StateoftheClimate2017 lowres

Transcript

1 STATE OF THE CLIMATE I N 2017 Special Supplement to the Bulletin of the American Meteorological Society Vol. 99, No. 8, August 2018

2

3 STATE OF THE CLIMATE I N 2017 Editors Jessica Blunden Derek S. Arndt Gail Hartfield Chapter Editors ánchez-Lugo Ahira S Peter Bissolli Gregory C. Johnson Ted A. Scambos Howard J. Diamond Tim Li Carl J. Schreck III Robert J. H. Dunn Ademe Mekonnen Sharon Stammerjohn Catherine Ganter Emily Osborne Diane M. Stanitski Nadine Gobron Jacqueline A. Richter-Menge Kate M. Willett Martin O. Jeffries Technical Editor Mara Sprain Al meric ociety A An m eteorologic S

4 : Credits over C r © : ront F on _ t homas/Spring desert wildflowers in Anza Borrego Desert State Park, CA/Getty Images. moke and Fire in Southern California: Thick smoke was streaming from several fires in Southern California when : s aCk B the Moderate Resolution Imaging Spectroradiometer (MODIS) on NASA’s Terra satellite acquired a natural-color image in the afternoon on December 5, 2017. On the same day, the Multi Spectral Imager (MSI) on the European Space Agency’s Sentinel-2 satellite captured the data for a false-color image of the burn scar. Active fires appear orange; the burn scar is brown. Unburned vegetation is green; developed areas are gray. The Sentinel-2 image is based on observations of visible, shortwave infrared, and near infrared light . NASA Earth Observatory images by Joshua Stevens, using MODIS data from LANCE/EOSDIS Rapid Response and modi - fied Copernicus Sentinel data (2017) processed by the European Space Agency. Story by Adam Voiland. Instrument(s): Terra - MODIS Sentinel-2 How to cite this document: Citing the complete report: A rndt, and G. Hartfield , Eds., 2018: State of the Climate in 2017. Bull. Amer. Blunden, J., D. S. Meteor. Soc. –S332, doi:10.1175/2018BAMSStateoftheClimate.1. , 99 (8), S i Citing a chapter (example): Richter-Menge, J., M. O. Jeffries, and E. Osborne, Eds., 2018: The Arctic [in “State of the Climate in (8), S143–173, doi:10.1175/2018BAMSStateoftheClimate.1. 2017”]. Bull. Amer. Meteor. Soc. , 99 Citing a section (example): Osborne, E., T. Cronin, and J. Farmer, 2018: Paleoclimate records: Providing context and ., Bull. Amer. Meteor. Soc understanding of current Arctic change [in “State of the Climate in 2017”]. (8), S150–S152, doi:10.1175/2018BAMSStateoftheClimate.1. 99

5 by ) EDITOR AND AUTHOR AFFILIATIONS ( alphabetical name Becker, Andreas, Met Office Hadley Centre, Exeter, United Abernethy, R., Global Precipitation Climatology Centre, Deutscher Wetterdienst, Offenbach, Germany Kingdom Bedka, Kristopher M., NASA Langley Research Center, Ackerman, Steven A., Cooperative Institute for Hampton, Virginia Meteorological Satellite Studies, University of Wisconsin– Behe, Madison, Madison, Wisconsin Inuit Circumpolar Council Alaska, Anchorage, Carolina, University of Maryland, College Park, Maryland Adler, R., Alaska Albanil Encarnación, Adelina, National Meteorological Bell, Gerald D., NOAA/NWS Climate Prediction Center, College Park, Maryland Service of Mexico, Mexico Servicio Meteorológico Nacional, Buenos Bellouin, Nicolas, Aldeco, Laura S., University of Reading, Reading, United Kingdom Aires, Argentina Eric J., Seychelles National Meteorological Services, Belmont, M., Alfaro, Center for Geophysical Research and School of Pointe Larue, Mahé, Seychelles Physics, University of Costa Rica, San José, Costa Rica European Centre for Medium-Range Benedetti, Angela, Aliaga-Nestares, Vannia, Servicio Nacional de Meteorología e Hidrología del Perú, Lima, Perú Weather Forecasts, Reading, United Kingdom Allan, Richard P., University of Reading, Reading, United Biospherical Instruments, San Diego, G. H., Bernhard , Kingdom California Allan, Rob, European Centre for Medium-Range Berrisford, Paul, Met Office Hadley Centre, Exeter, United Kingdom Weather Forecasts, Reading, United Kingdom Alves, Lincoln M., Centro de Ciencias do Sistema Terrestre, National Oceanography Centre, Instituto Nacional de Pesquisas Espaciais, Cachoeira Paulista, Berry, David I., Southampton, United Kingdom Sao Paulo, Brazil U. S., Center for Geophysical Research and Geophysical Institute, University of Alaska Bhatt, Amador, Jorge A., Fairbanks, Fairbanks, Alaska School of Physics, University of Costa Rica, San José, Costa Bissolli, Peter, Deutscher Wetterdienst, WMO RA VI Rica Department of Atmospheric and Planetary Regional Climate Centre Network, Offenbach, Germany Anderson, John, Bjerke, J., Norwegian Institute for Nature Research, Tromsø, Science, Hampton University, Hampton, Virginia Andreassen, L. M., Section for Glaciers, Ice and Snow, Norway Blake, Eric S., NOAA/NWS National Hurricane Center, Norwegian Water Resources and Energy Directorate, Oslo, Miami, Florida Norway Blenkinsop, Stephen, Argüez, Anthony, NOAA/NESDIS National Centers for School of Engineering, Newcastle Environmental Information, Asheville, North Carolina University, Newcastle-upon-Tyne, United Kingdom NOAA/NESDIS National Centers for Blunden, Jessica, Woodland Trust, Grantham, United Kingdom Armitage, C., NOAA/NESDIS National Centers for Arndt, Derek S., Environmental Information, Asheville, North Carolina Environmental Information, Asheville, North Carolina Swedish University of Agricultural Sciences, Bolmgren, K., Avalos, Grinia, Uppsala, Sweden Servicio Nacional de Meteorología e Hidrología del Perú, Lima, Perú Global Modelling and Assimilation Bosilovich, Michael G., Azorin-Molina, Office, NASA Goddard Space Flight Center, Greenbelt, César Regional Climate Group, Department , of Earth Sciences, University of Gothenburg, Gothenburg, Maryland Sweden Boucher, Olivier, Institut Pierre-Simon Laplace, CNRS/UPMC, Dirección de Meteorología e Hidrología de la Báez, Julián, Paris, France DINAC and Universidad Católica Ntra. Sra. de la Asunción, Instituto del Mar del Perú, Callao, Perú Bouchon, Marilú, Asunción, Paraguay Box, J. E., Geological Survey of Denmark and Greenland, Copenhagen, Denmark Bardin, Institute of Global Climate and Ecology of M. Yu., NOAA/NESDIS National Centers for Roshydromet and Russian Academy of Sciences, and Institute Boyer, Tim, of Geography of Russian Academy of Sciences, Russia Environmental Information, Silver Spring, Maryland Barichivich, Jonathan, Universidad Austral de Chile, WMO Atmospheric Environment Braathen, Geir O., Valdivia, Chile; Center for Climate and Resilience Research Research Division, Geneva, Switzerland (CR)2, Chile Bromwich, David H., Byrd Polar and Climate Research Center, The Ohio State University, Columbus, Ohio Baringer, Molly O., NOAA/OAR Atlantic Oceanographic and Meteorological Laboratory, Miami, Florida Brown, R., Climate Research Division, Environment and Climate Change Canada, Downsview, Ontario, Canada Barreira, Sandra, Argentine Naval Hydrographic Service, Buenos Aires, Argentina Buehler, S., Universitaet Hamburg, Hamburg, Germany Baxter, Stephen, Russian Institute for Hydrometeorological NOAA/NWS Climate Prediction Center, Olga N., Bulygina, Information, Obninsk, Russia College Park, Maryland Department of Civil and Environmental Beck, H.E., D., Geological Survey of Canada, Ottawa, Ontario, Burgess, Engineering, Princeton University, Princeton, New Jersey Canada Si | AUGUST 2018 STATE OF THE CLIMATE IN 2017

6 Daniel, The Pew Charitable Trusts, Washington, Calderón, Blanca, Center for Geophysical Research, Raychelle, D.C. University of Costa Rica, San José, Costa Rica Cooperative Institute for Research in Lamont-Doherty Earth Observatory, Camargo, Suzana J., Davis, Sean M., Columbia University, Palisades, New York Environmental Sciences, University of Colorado Boulder, and Campbell, Ethan C., School of Oceanography, University of NOAA/OAR Earth System Research Laboratory, Boulder, Washington, Seattle, Washington Colorado Russian Institute for Hydrometeorological S. G. , Department of Physics, The University Davletshin, Campbell, Jayaka D., Information, Obninsk, Russia of the West Indies, Jamaica Marine Institute, Newport, Ireland Danish Meteorological Institute, Copenhagen, J., Cappelen, de Eyto, Elvira, de Jeu, Richard A. M., Denmark EODC GmbH, Vienna, Austria NOAA/NESDIS Coral Reef De La Cour, Jacqueline L., Department of Meteorology, University of Carrea, Laura, Reading, Reading, United Kingdom Watch, College Park, Maryland, and Global Science and Carter, Brendan R., Technology, Inc., Greenbelt, Maryland Joint Institute for the Study of the Atmosphere and Ocean, University of Washington, and Royal Netherlands Meteorological Institute de Laat, Jos, (KNMI), DeBilt, Netherlands NOAA/OAR Pacific Marine Environmental Laboratory, Seattle, Washington DeGasperi, Curtis L., King County Water and Land Resources Division, Seattle, Washington Castro, Anabel, Servicio Nacional de Meteorología e University of Saskatchewan, Saskatoon, Hidrología del Perú, Lima, Perú Degenstein, Doug, Chambers, Don P., College of Marine Science, University of Saskatchewan, Canada EDYTEM Lab, University Savoie Mont Blanc, Deline, P., South Florida, St. Petersburg, Florida Chambéry, France Cheng, Lijing, International Center for Climate and Turkish State Meteorological Service, Demircan, Mesut, Environment Sciences, Institute of Atmospheric Physics, A nk ar a , Turkey Chinese Academy of Sciences, Beijing, China Climate Research Division, Environment and Derksen, Christiansen, Hanne H., C., Geology Department, University Climate Change Canada, Downsview, Ontario, Canada Centre in Svalbard, Longyearbyen, Norway Christy, John R., University of Alabama in Huntsville, Dewitte, Boris, Centro de Estudios Avanzado en Zonas Huntsville, Alabama Áridas, and Universidad Católica del Norte, Coquimbo, Chile, Chung, E.-S., and Laboratoire d’Etudes en Géophysique et Océanographie Rosenstiel School of Marine and Atmospheric Science, University of Miami, Key Biscane, Miami, Florida Spatiales, Toulouse, France Clem, Kyle R., Institute of Earth, Ocean, and Atmospheric Dhurmea, Mauritius Meteorological Service, Vacoas, R., Sciences, Rutgers, the State University of New Jersey, New Mauritius Brunswick, New Jersey Di Girolamo, Larry, University of Illinois at Urbana– CPTEC/INPE Center for Weather Coelho, Caio A.S., Champaign, Urbana, Illinois Forecasts and Climate Studies, Cachoeira Paulista, Brazil Diamond, Howard J., NOAA/OAR Air Resources German Aerospace Center Coldewey-Egbers, Melanie, Laboratory, Silver Spring, Maryland (DLR) Oberpfaffenhofen, Wessling, Germany C., Dickerson, Department of Environmental Sciences, British Antarctic Survey, Cambridge, United Colwell, Steve, University of Virginia, Charlottesville, Virginia Kingdom Dlugokencky, Ed J., NOAA/OAR Earth System Research Cooper, Owen. R., Cooperative Institute for Research in Laboratory, Boulder, Colorado Environmental Sciences, University of Colorado Boulder, and Earth and Space Research, Seattle, Dohan, Kathleen, NOAA/OAR Earth System Research Laboratory, Boulder, Washington Colorado Dokulil, Martin T., Research Institute for Limnology, Copland, Department of Geography, University of Ottawa, L., University of Innsbruck, Mondsee, Austria Ottawa, Ontario, Canada Dolman, A. Johannes, Department of Earth Sciences, Earth Antarctic Meteorological Research Center Costanza, Carol, and Climate Cluster, VU University Amsterdam, Amsterdam, and Space Science and Engineering Center, University of Netherlands Wisconsin-Madison, Madison, Wisconsin Domingues, Catia M., Institute for Marine and Antarctic Lawrence Livermore National Laboratory, Covey, Curt , Studies, University of Tasmania, Antarctic Climate and Livermore, California Ecosystems Cooperative Research Centre, and Australian Science Systems and Applications, Inc., NASA Coy, Lawrence, Research Council’s Centre of Excellence for Climate System Goddard Space Flight Center, Greenbelt, Maryland Science, Hobart, Tasmania, Australia U.S. Geological Survey, Reston, Virginia T. , Cronin, Cooperative Institute for Marine and Domingues, Ricardo, NOAA/NESDIS National Centers for Crouch, Jake, Atmospheric Studies, University of Miami, Miami, Florida Environmental Information, Asheville, North Carolina Climate Change Research Centre, Donat, Markus G., Servicio Nacional de Meteorología e Cruzado, Luis , University of New South Wales, Sydney, New South Wales, Hidrología del Perú, Lima, Perú Australia Sii | AUGUST 2018

7 Forbes, NOAA/OAR Atlantic Oceanographic and Dong, Shenfu, B. C., Arctic Centre, University of Lapland, Rovaniemi, Meteorological Laboratory, and Cooperative Institute for Finland Marine and Atmospheric Science, Miami, Florida Foster, Michael J., Cooperative Institute for Meteorological Dorigo, Wouter A., Department of Geodesy and Satellite Studies, University of Wisconsin–Madison, Madison, Geoinformation, Vienna University of Technology, Vienna, Wisconsin S. D., Francis, Austria National Weather Forecasting and Climate Drozdov, D. S., Earth Cryosphere Institute, and Tyumen State Research Centre, Nigerian Meteorological Agency, Abuja, Nigeria University, Tyumen, Russia Met Office Hadley Centre, Exeter, NASA Goddard Space Flight Center, Dunn, Robert J. H., Franz, Bryan A., United Kingdom Greenbelt, Maryland NOAA/NESDIS National Centers for Cooperative Institute for Meteorological Frey, Richard A., Durre, Imke, Environmental Information, Asheville, North Carolina Satellite Studies, University of Wisconsin–Madison, Madison, Cooperative Institute for Research in Dutton, Geoff S., Wisconsin Frith, Stacey M., Environmental Sciences, University of Colorado Boulder, and Science Systems and Applications, Inc. and NOAA/OAR Earth System Research Laboratory, Boulder, NASA Goddard Space Flight Center, Greenbelt, Maryland Colorado Froidevaux, Lucien, Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California NOAA/NESDIS Coral Reef Watch, College Eakin, C. Mark, Bureau of Meteorology, Melbourne, Park, Maryland Catherine, Ganter, Victoria, Australia Direction de la Météorologie Nationale Maroc, M., ElKharrim, NOAA/NESDIS Coral Reef Watch, College Rabat, Morocco Geiger, Erick F., NOAA/OAR Earth System Research Elkins, James W., Park, Maryland, and Global Science and Technology, Inc., Laboratory, Boulder, Colorado Greenbelt, Maryland Norwegian Polar Institute, Fram Centre, Tromsø, Gerland, S., Epstein, Department of Environmental Sciences, H. E., Norway University of Virginia, Charlottesville, Virginia Scripps Institution of Oceanography, University Gilson, John, Espinoza, Jhan C., Instituto Geofisico del Perú, Lima, Perú of California San Diego, La Jolla, California Jet Propulsion Laboratory, California Famiglietti, James S., Gobron, Nadine, European Commission, Joint Research Institute of Technology, Pasadena, California Centre, Ispra, Italy J., Department of Geosciences, Princeton University, Farmer, Goldenberg, Stanley B., University of Miami, Miami, Florida Princeton, New Jersey Gomez, Andrea M., NOAA/NESDIS Coral Reef Watch, S., Earth System Science Interdisciplinary Center, Farrell, College Park, Maryland, and Ecosystem Science Lab and University of Maryland, College Park, Maryland NOAA-CREST, New York, New York P. , Fauchald, Norwegian Institute for Nature Research, NOAA/OAR Atlantic Oceanographic and Goni, Gustavo, Tromsø, N or w ay Meteorological Laboratory, Miami, Florida Fausto, R. S., Geological Survey of Denmark and Greenland, Forschungszentrum Jülich, Jülich, Germany Grooß, Jens-Uwe, Copenhagen, Denmark Gruber, Alexander, Department of Geodesy and NOAA/OAR Pacific Marine Environmental Feely, Richard A., Geoinformation, Vienna University of Technology, Vienna, Laboratory, Seattle, Washington Austria; Department of Earth and Environmental Sciences, Atmospheric Sciences and Global Change Division, Feng, Z., KU Leuven, Heverlee, Belgium Pacific Northwest National Laboratory, Richland, Washington NOAA/NWS Weather Forecast Charles “Chip” P., Guard, NOAA/NESDIS National Centers for Fenimore, Chris, Office, Guam Environmental Information, Asheville, North Carolina Gugliemin, Mario, Department of Theoretical and Applied University of Liège, Liège, Belgium X., Fettweis, Sciences, Insubria University, Varese, Italy and Climate Change Environment Fioletov, Vitali E., Canada, Gupta, S. K., Science Systems and Applications, Inc., Hampton, Toronto, Ontario, Canada Virginia Flemming, Johannes, European Centre for Medium-Range Gutiérrez, Dimitri, Instituto del Mar del Perú, Callao, Perú Weather Forecasts, Reading, United Kingdom C., Haas, Alfred Wegener Institute, Helmholtz Centre for Polar Department of Geography, Ohio University, Fogt, Ryan L., and Martine Research, Bremerhaven, Germany Athens, Ohio Hagos, S., Atmospheric Sciences and Global Change Division, Met Office Hadley Centre, Exeter, and Folland, Chris, Pacific Northwest National Laboratory, Richland, Washington School of Environmental Sciences, University of East Hahn, Sebastian, Department of Geodesy and Anglia, Norwich, United Kingdom, and Department of Geoinformation, Vienna University of Technology, Vienna, Earth Sciences, University of Gothenburg, Gothenburg, Austria Sweden, and International Centre for Applied Climate Department of Meteorology and Haimberger, Leo, Sciences, University of Southern Queensland, Toowoomba, Geophysics, University of Vienna, Vienna, Austria Queensland, Australia Siii | AUGUST 2018 STATE OF THE CLIMATE IN 2017

8 Finnish Meteorological Institute, Helsinki, Iolanda, Ialongo, Hall, Brad D., NOAA/OAR Earth System Research Finland Laboratory, Boulder, Colorado M. M., Department of Meteorology, Al-Azhar Ibrahim, NOAA/NWS Climate Prediction Halpert, Michael S., University, Egypt Center, College Park, Maryland Ijampy, J. A., Nigerian Meteorological Agency, Abuja, Nigeria Hamlington, Benjamin D., Center for Coastal Physical Inness, Antje, European Centre for Medium-Range Weather Oceanography, Old Dominion University, Norfolk, Virginia Forecasts, Reading, United Kingdom Department of Geography, University of Lincoln, E., Hanna, Isaac, Victor, Environment and Climate Change Canada, Lincoln, United Kingdom Toronto, Ontario, Canada Geological Survey of Denmark and Greenland, K., Hansen, Isaksen, K., Norwegian Meteorological Institute, Blindern, Copenhagen, Denmark Oslo, Norway Norwegian Meteorological Institute, L., Hanssen-Bauer, Ishii, Masayoshi, Climate Research Department, Blindern 0313, Oslo, Norway Meteorological Research Institute, Japan Meteorological National Centre for Atmospheric Science, Harris, Ian, Agency, Tsukuba, Japan University of East Anglia, Norwich, and Climatic Research Bureau of Meteorology, Melbourne, Stephanie J., Jacobs, Unit, School of Environmental Sciences, University of East Victoria, Australia Anglia, Norwich, United Kingdom Jeffries, Martin O., Office of Naval Research, Arlington, Hartfield, Gail, NOAA/NWS Weather Forecast Office, Virginia Raleigh, North Carolina National Oceanography Centre, Jevrejeva, Svetlana, NOAA/NESDIS/STAR University of Heidinger, Andrew K., Liverpool, United Kingdom Wisconsin–Madison, Madison, Wisconsin Jiménez, C., Estellus, Paris, France NOAA/NESDIS National Centers for Heim, Jr., Richard R., Woods Hole Oceanographic Institution, Woods Jin, Xiangze, Environmental Information, Asheville, North Carolina Hole, Massachusetts NOAA/NESDIS Center for Satellite Applications S., Helfrich, John, Viju, EUMETSAT, Darmstadt, Germany, and Met Office and Research, College Park, Maryland Hadley Centre, Exeter, United Kingdom Met Office Hadley Centre, Exeter, Hemming, D. L., Johns, William E., Rosenstiel School of Marine and United Kingdom; Birmingham Institute of Forest Research, Atmospheric Science, Miami, Florida Birmingham University, Birmingham, United Kingdom Johnsen, Bjørn, Norwegian Radiation Protection Authority, Hendricks, S., Alfred Wegener Institute, Helmholtz Centre for Østerås, Norway Polar and Martine Research, Bremerhaven, Germany NOAA/OAR Earth System Research Johnson, Bryan, Instituto Nacional de Meteorología e Hernández, Rafael, Laboratory, Global Monitoring Division, and University of Hidrología de Venezuela (INAMEH), Caracas, Venezuela Colorado Boulder, Boulder, Colorado Hernández, Sosa Marieta, Climate Center, Institute of Johnson, Gregory C., NOAA/OAR Pacific Marine Meteorology of Cuba, Havana, Cuba Environmental Laboratory, Seattle, Washington NOAA/NESDIS Coral Reef Watch, Heron, Scott F., Monterey Bay Aquarium Research Johnson, Kenneth S., College Park, Maryland, and ReefSense Pty Ltd, Townsville, Institute, Moss Landing, California Queensland, Australia Climatic Research Unit, School of Jones, Philip D., Department of Marine Sciences, University of Heuzé, C., Environmental Sciences, University of East Anglia, Norwich, Gothenburg, Sweden United Kingdom Center for Geophysical Research and Hidalgo, Hugo G., Météo France, Direction Interrégionale Guillaume, Jumaux, School of Physics, University of Costa Rica, San José, Costa pour l’Océan Indien, Réunion Rica Kabidi, Khadija, Direction de la Météorologie Nationale COSMIC, UCAR, Boulder, Colorado Ho, Shu-peng (Ben), Maroc, Rabat, Morocco Hobbs, William R., Antarctic Climate and Ecosystems Max Planck Institute for Chemistry, Mainz, Kaiser, J. W., Cooperative Research Centre, University of Tasmania, Germany Tasmania, Australia NASA Goddard Space Flight Center, Erdem M., Karaköylü, Department of Ecology and Environmental Horstkotte, T. , Greenbelt, Maryland, and Science Application International Sciences, Umeå University, Umeå, Sweden Corporation, Beltsville, Maryland NOAA/NESDIS National Centers for Huang, Boyin, NASA Langley Research Center, Hampton, Virginia Kato, Seiji, Environmental Information, Asheville, North Carolina A., Kazemi, Islamic Republic of Iranian Meteorological Hubert, Daan, Royal Belgian Institute for Space Aeronomy Organization, Iran (BIRA), Brussels, Belgium Keller, Linda M., Department of Atmospheric and Oceanic Department of Marine Sciences, University of Céline Hueuzé, , Sciences, University of Wisconsin–Madison, Madison, Gothenburg, Gothenburg, Gothenburg, Sweden Wisconsin Hurst, Dale F., Cooperative Institute for Research in Environmental Met Office Hadley Centre, Exeter, United Kennedy, John, Sciences, University of Colorado Boulder, and NOAA/OAR Earth Kingdom System Research Laboratory, Boulder, Colorado Siv | AUGUST 2018

9 Trinidad and Tobago Meteorological Service, Kerr, Kenneth, NOAA/NWS NCWCP Laboratory for Leuliette, Eric, Piarco, Trinidad Satellite Altimetry, College Park, Maryland Khan, Geological Survey of Denmark and Greenland, M. S., L’Heureux, Michelle, NOAA/NWS Climate Prediction Copenhagen, Denmark Center, College Park, Maryland Kholodov, A. L., Geophysical Institute, University of Alaska Li, Bailing, Hydrological Sciences Laboratory, NASA Goddard Fairbanks, Fairbanks, Alaska Space Flight Center, Greenbelt, Maryland; Earth System Mahbobeh, Khoshkam, Islamic Republic of Iranian Science Interdisciplinary Center, University of Maryland, Meteorological Organization, Iran College Park, Maryland Killick, Rachel, Met Office Hadley Centre, Exeter, United Li, Department of Atmospheric Sciences, Universtiy of Tim, Kingdom Hawaii, Honolulu, Hawaii Institute of Industrial Science, University of Kim, Hyungjun, Lieser, Jan L., Antarctic Climate and Ecosystems Cooperative Tokyo, Japan Research Centre and Institute for Marine and Antarctic S.-J., Kim, Korea Polar Research Institute, Incheon, Republic of Studies, University of Tasmania, Hobart, Tasmania, Australia Korea National Taiwan University, Taipei, Taiwan Lin, I.-I., Klotzbach, Philip J., Department of Atmospheric Science, NOAA/NESDIS Coral Reef Watch, College Liu, Gang, Colorado State University, Fort Collins, Colorado Park, Maryland, and Global Science and Technology, Inc., NOAA/NESDIS Center for Satellite Knaff, John A., Greenbelt, Maryland Applications and Research, Fort Collins, Colorado Liu, Hongxing, Department of Geography, University of Kohler, Norwegian Polar Institute, Tromsø, Norway J., Cincinnati, Cincinnati, Ohio Freshwater Centre, Finnish Environment Korhonen, Johanna, Locarnini, Ricardo, NOAA/NESDIS National Centers for Institute (SYKE), Helsinki, Finland Environmental Information, Silver Spring, Maryland Korshunova, Natalia N., All-Russian Research Institute Loeb, Norman G., NASA Langley Research Center, of Hydrometeorological Information - World Data Center, Hampton, Virginia Obninsk, Russia Long, Craig S., NOAA/NWS National Centers for Kramarova, Natalya, NASA Goddard Space Flight Center, Environmental Prediction, College Park, Maryland Greenbelt, Maryland López, Luis A., Instituto de Hidrología, Meteorología y Kratz, D. P., NASA Langley Research Center, Hampton, Estudios Ambientales de Colombia, Bogotá, Colombia Virginia National Institute of Water and Lorrey, Andrew M., Kruger, South African Weather Service, Pretoria, Andries, Atmospheric Research, Ltd., Auckland, New Zealand South Africa German Aerospace Center (DLR) Loyola, Diego, Kruk, Michael C., ERT, Inc., NOAA/NESDIS National Centers Oberpfaffenhofen, Wessling, Germany for Environmental Information, Asheville, North Carolina Lumpkin, Rick, NOAA/OAR Atlantic Oceanographic and T. , Alfred Wegener Institute, Helmholtz Centre for Krumpen, Meteorological Laboratory, Miami, Florida Polar and Martine Research, Bremerhaven, Germany Luo, Jing-Jia, Australian Bureau of Meteorology, Melbourne, NOAA/OAR Pacific Marine Environmental C., Ladd, Victoria, Australia Laboratory, Seattle, Washington K., Luojus, Finnish Meteorological Institute, Helsinki, Finland , Lakatos, Climatology Division, Hungarian Mónika Luthcke, S., NASA Goddard Space Flight Center, Greenbelt, Meteorological Service, Budapest, Hungary Maryland Finnish Meteorological Institute, Arctic Lakkala, Kaisa, NOAA/OAR Pacific Marine Environmental Lyman, John M., Research Centre, Sodankylä, Finland Laboratory, Seattle, Washington, and Joint Institute for Mark A., University of Guam, Mangilao, Guam Lander, Marine and Atmospheric Research, University of Hawaii, Landschützer, Peter, Max Planck Institute for Meteorology, Honolulu, Hawaii Hamburg, Germany School of Geography and the Macias-Fauria, M., Landsea, Chris W., NOAA/NWS National Hurricane Center, Environment, University of Oxford, Oxford, United Kingdom Miami, Florida G. V., Malkova, Earth Cryosphere Institute, Tyumen Science Lankhorst, Matthias, Scripps Institution of Oceanography, Center, Tyumen, Russia University of California, San Diego, La Jolla, California NorthWest Research Associates, and Manney, Gloria L., Servicio Nacional de Lavado-Casimiro, Waldo, New Mexico Institute of Mining and Technology, Socorro, Meteorología e Hidrología del Perú, Lima, Perú New Mexico Department of Physical Sciences, Lazzara, Matthew A., Marcellin, Vernie, Dominica Meteorological Service, School of Arts and Sciences, Madison Area Technical College, Dominica and Space Science and Engineering Center, University of Geophysical Institute, University of Alaska S. S., Marchenko, Wisconsin–Madison, Madison, Wisconsin Fairbanks, Fairbanks, Alaska Korea Meteorological Administration, South Korea S.-E., Lee, Marengo, José A., Centro Nacional de Monitoramento Lee, T. C . , Hong Kong Observatory, Hong Kong, China e Alertas aos Desastres Naturais, Cachoeira Paulista, Sao Paulo, Brazil Sv | AUGUST 2018 STATE OF THE CLIMATE IN 2017

10 Space Science and Engineering Center, Menzel, W. Paul, Servicio Nacional de Meteorología e Hidrología Marín, Dora, University of Wisconsin–Madison, Madison, Wisconsin de Perú, Lima, Perú Merchant, Christopher J., Department of Meteorology, NOAA/NESDIS National Centers for Marra, John J., University of Reading, Reading, and National Centre for Earth Environmental Information, Honolulu, Hawaii Observation, University of Reading, Reading, United Kingdom Marsh, Benjamin L., NOAA/NESDIS Coral Reef Watch, British Antarctic Survey, Cambridge, Meredith, Michael P., College Park, Maryland, and ReefSense Pty Ltd, Townsville, United Kingdom Queensland, Australia Merrifield, Mark A., Joint Institute for Marine and Department of Hydrology and Marszelewski, Wlodzimierz, Atmospheric Research, University of Hawaii, Honolulu, Water Management, Nicolaus Copernicus University, Toruń, Hawaii Poland Cooperative Institute for Research in Miller, Ben, Martens, B., Laboratory of Hydrology and Water Management, Environmental Sciences, University of Colorado Boulder, and Ghent University, Ghent, Belgium NOAA/OAR Earth System Research Laboratory, Boulder, Department of Zoology, University of Oxford, Martin, A., Colorado Oxford, United Kingdom Laboratory of Hydrology and Water Miralles, Diego G., Instituto Geofísico del Perú, Lima, Martínez, Alejandra G., Management, Ghent University, Ghent, Belgium Perú College of Marine Science, University of Mitchum, Gary T., Martínez-Güingla, Rodney, Centro Internacional para la South Florida, St. Petersburg, Florida Investigación del Fenómeno de El Niño, Guayaquil, Ecuador Meteorological Service Suriname, Paramaribo, Mitro, Sukarni, Martínez-Sánchez, Odalys, NOAA/NWS San Juan, Suriname Puerto Rico Moat, Ben, National Oceanography Centre, Southampton, Massom, Robert A., Australian Antarctic Division, and United Kingdom Antarctic Climate and Ecosystems Cooperative Research Y., Tokyo Climate Center, Japan Meteorological Mochizuki, Centre, University of Tasmania, Hobart, Tasmania, Australia Agency, Japan May, Linda, Centre for Ecology and Hydrology, Edinburgh, CSIRO Oceans and Atmosphere, Monselesan, Didier, United Kingdom Hobart, Tasmania, Australia Mayer, Michael, Department of Meteorology and Geophysics, NOAA/OAR Earth System Research Montzka, Stephen A., University of Vienna, Vienna, Austria; European Centre for Laboratory, Boulder, Colorado Medium-Range Weather Forecasts, Reading, United Kingdom Mora, Natalie, Center for Geophysical Research and School Mazloff, Matthew, Scripps Institution of Oceanography, of Physics, University of Costa Rica, San José, Costa Rica University of California, San Diego, La Jolla, California Met Office Hadley Centre, Exeter, United Morice, Colin, South African Weather Service, Charlotte, McBride, Kingdom Pretoria, South Africa Instituto Geofísico del Perú, Kobi, Mosquera-Vásquez, McCabe, M. F., Water Desalination and Reuse Center, Division Lima, Perú of Biological and Environmental Sciences and Engineering, Mostafa, Awatif E., Department of Seasonal Forecast and King Abdullah University of Science and Technology, Thuwal, Climate Research, Cairo Numerical Weather Prediction, Saudi Arabia Egyptian Meteorological Authority, Cairo, Egypt McCarthy, Mark, Met Office Hadley Centre, Exeter, Mote, Department of Geography, University of Georgia, T. , United Kingdom Athens, Georgia McVicar, Tim R., CSIRO Land and Water Flagship, Canberra, Climate Research Division, Environment and L., Mudryk, Australian Capital Territory, and Australian Research Council Climate Change Canada, Downsview, Ontario, Canada Centre of Excellence for Climate System Science, Sydney, Scripps Institution of Oceanography, University of Mühle, Jens, New South Wales, Australia California, San Diego, La Jolla, California Mears, Carl A., Remote Sensing Systems, Santa Rosa, National Institute of Water and Mullan, A. Brett, California Atmospheric Research, Ltd., Wellington, New Zealand Meier, W., National Snow and Ice Data Center, University of Müller, Rolf, Forschungszentrum Jülich, Jülich, Germany Colorado, Boulder, Boulder, Colorado R., Department of Earth and Environment, Boston Myneni, Meijers, Andrew J. S., British Antarctic Survey, Cambridge, University, Boston, Massachusetts United Kingdom Science Systems and Applications, Inc., Lanham, Nash, Eric R., Department of Energy and Ademe, Mekonnen, Maryland Environmental Systems, North Carolina A & T State Nerem, R. Steven, Colorado Center for Astrodynamics University, Greensboro, North Carolina Research, Cooperative Institute for Research in Department of Earth and Environmental Mengistu Tsidu, G., Environmental Sciences, University of Colorado Boulder, Sciences, Botswana International University of Science and Boulder, Colorado Technology, Palapye, Botswana, and Department of Physics, Addis Ababa University, Addis Ababa, Ethiopia Svi | AUGUST 2018

11 Newman, L., Environment and Climate Change Canada, Phillips, David, SOOS International Project Office, Institute for Toronto, Ontario, Canada Marine and Antarctic Studies, University of Tasmania, Hobart, Department of Animal and Plant Sciences, G., Phoenix, Australia. NASA Goddard Space Flight Center, University of Sheffield, Sheffield, United Kingdom Newman, Paul A., European Commission, Joint Research Centre, Pinty, Bernard, Greenbelt, Maryland Ispra, Italy Nielsen-Gammon, John W., Texas A&M University, College Pinzon, St at ion , Tex a s J., NASA Goddard Space Flight Center, Greenbelt, Centro Internacional para la Investigación Nieto, Juan José, Maryland Po-Chedley, S., del Fenómeno de El Niño, Guayaquil, Ecuador Lawrence Livermore National Laboratory, Livermore, California Noetzli, Jeannette, WSL Institute for Snow and Avalanche C., Research, Davos, Switzerland Polashenski, USACE, ERDC, Cold Regions Research and Noll, Engineering Laboratory, and Thayer School of Engineering, Ben E., National Institute of Water and Atmospheric Research, Ltd., (NIWA), Auckland, New Zealand Dartmouth College, Hanover, New Hampshire USGS, Alaska Science Center, Anchorage, Alaska O’Neel, S., Purkey, Sarah G., Scripps Institution of Oceanography, Climatic Research Unit, School of Osborn, Tim J., University of California, San Diego, La Jolla, California Quispe, Environmental Sciences, University of East Anglia, Norwich, Nelson, Servicio Nacional de Meteorología e Hidrología del Perú, Lima, Perú United Kingdom Earth System Science Organization, Rajeevan, Osborne, Madhavan, NOAA/OAR Arctic Research Program, Emily, Ministry of Earth Sciences, New Delhi, India Silver Spring, Maryland Madagascar Meteorological Service, NOAA/OAR Pacific Marine Environmental Overland, Rakotoarimalala, C., J., Antananarivo, Madagascar Laboratory, Seattle, Washington Lamjav, Rayner, Darren, Hydrology and Environmental National Oceanography Centre, Oyunjargal, Southampton, United Kingdom Monitoring, Institute of Meteorology and Hydrology, National M. K., Agency for Meteorology, Ulaanbaatar, Mongolia Institute of Arctic Biology, University of Raynolds, Department of Earth and Environment, Boston Alaska Fairbanks, Fairbanks, Alaska T. , Park, University, Boston, Massachusetts Earth System Science Interdisciplinary Center/ Reagan, James, Cooperative Institute for Climate and Satellites–Maryland, Pasch, Richard J., NOAA/NWS National Hurricane Center, Miami, Florida University of Maryland, College Park, Maryland, and NOAA/ NESDIS National Centers for Environmental Information, Pascual-Ramírez, Reynaldo, National Meteorological Silver Spring, Maryland Service of Mexico, Mexico Reid, Phillip, Agencia Estatal de Pastor Saavedra, Maria Asuncion, Australian Bureau of Meteorology, and Antarctic Meteorología, Madrid, Spain Climate and Ecosystems Cooperative Research Centre, Paterson, Andrew M., Dorset Environmental Science Hobart, Tasmania, Australia Centre, Ontario Ministry of the Environment and Climate EODC, Vienna, Austria Reimer, Christoph, Change, Dorset, Ontario, Canada Rémy, Samuel, Institut Pierre-Simon Laplace, CNRS / UPMC, Paulik, Christoph, Paris, France VanderSat B.V., Haarlem, the Netherlands Indian Institute of Tropical National Institute of Water and Pearce, Petra R., Jayashree V., Revadekar, Atmospheric Research, Ltd., Auckland, New Zealand Meteorology, Pune, India Richardson, A. D., Peltier, Alexandre, Météo-France en Nouvelle-Cáledonie, School of Informatics, Computing and Noumea, Caledonia Cyber Systems and Center for Ecosystem Science and Society, Nichols College, Dudley, Massachusetts Mauri S., Pelto, Northern Arizona University, Flagstaff, Arizona University of Alaska Fairbanks, Peng, Liang, State University of New York, Albany, New York Jacqueline, Richter-Menge, Climate Change Research Fairbanks, Alaska Perkins-Kirkpatrick, Sarah E., Ricker, R., Centre, University of New South Wales, Sydney, New South Alfred Wegener Institute, Helmholtz Centre for Wales, Australia Polar and Marine Research, Bremerhaven, Germany Thayer School of Engineering, Dartmouth Yigal Allon Kinneret Limnological Laboratory, Rimmer, Alon, Perovich, Don, College, Hanover, New Hampshire Israel Oceanographic and Limnological Research, Migdal, Israel Cooperative Institute for Research Petropavlovskikh, Irina, Department of Geography, Rutgers in Environmental Sciences, University of Colorado Boulder, Robinson, David A., University, Piscataway, New Jersey and NOAA/OAR Earth System Research Laboratory, Rodell, Matthew, Hydrological Sciences Laboratory, NASA Boulder, Colorado Pezza, Alexandre B., Goddard Space Flight Center, Greenbelt, Maryland Greater Wellington Regional Council, Wellington, New Zealand Rodriguez Camino, Ernesto, Agencia Estatal de Department of Atmospheric and Oceanic Sciences, Meteorología, Madrid, Spain Phillips, C., Geophysical Institute, University Romanovsky, Vladimir E., University of Wisconsin-Madison, Madison, Wisconsin of Alaska Fairbanks, Fairbanks, Alaska Svii | AUGUST 2018 STATE OF THE CLIMATE IN 2017

12 Ronchail, Josyane, Université Paris Diderot/Laboratoire Department of Geography, George Nikolai I., Shiklomanov, L’OCEAN-IPSL, Paris, France Washington University, Washington, D.C. Rosenlof, Karen H., NOAA/OAR Earth System Research Institute of Biology, Irkutsk State Shimaraeva, Svetlana V., Laboratory, Boulder, Colorado University, Russia Rösner, Benjamin, Laboratory for Climatology and Remote University of California–Santa Barbara, Santa Siegel, David A., Sensing, Faculty of Geography, University of Marburg, Barbara, California Marburg, Germany Silow, Eugene, Institute of Biology, Irkutsk State University, University of Saskatchewan, Saskatoon, Roth, Chris, Russia Saskatchewan, Canada Sima, Fatou, Division of Meteorology, Department of Water NOAA/NWS Weather Prediction Center, Roth, David Mark, Resources, Banjul, The Gambia College Park, Maryland Simmons, Adrian J., European Centre for Medium-Range Dorset Environmental Science Centre, Rusak, James A., Weather Forecasts, Reading, United Kingdom Ontario Ministry of the Environment and Climate Change, Skirving, William J., NOAA/NESDIS Coral Reef Watch, Dorset, Ontario, Canada College Park, Maryland, and ReefSense Pty Ltd, Townsville, Swiss Academies of Arts and Science, Berne, Rutishäuser, T. , Queensland, Australia Switzerland National Oceanography Centre, Smeed, David A., Sallée, Jean-Bapiste, Sorbonne Universités, L’OCEAN-IPSL, Southampton, United Kingdom Paris, France, and British Antarctic Survey, Cambridge, United Smeets, Institute for Marine and Atmospheric C. J. P. P., Kingdom Research Utrecht, Utrecht University, Utrecht, Netherlands NOAA/NESDIS National Centers for Sánchez-Lugo, Ahira, Smith, Adam, NOAA/NESDIS National Centers for Environmental Information, Asheville, North Carolina Environmental Information, Asheville, North Carolina Jet Propulsion Laboratory, California Santee, Michelle L., Smith, Sharon L., Geological Survey of Canada, Natural Institute of Technology, Pasadena, California Resources Canada, Ottawa, Ontario, Canada Sasgen, L., Climate Sciences Department, Alfred Wegener Rosenstiel School of Marine and Atmospheric Soden, B., Institute, Bremerhaven, Germany Science, University of Miami, Key Biscane, Miami, Florida Sawaengphokhai, P., Science Systems and Applications, Inc., Sofieva, Viktoria, Finnish Meteorological Institute (FMI), Hampton, Virginia Helsinki, Finland Department of Meteorology, Al-Azhar University, Sayad, T. A . , Sparks, T. H., Coventry University, Coventry, United Kingdom Egypt Spence, Jacqueline M., Meteorological Service, Jamaica, Direction de la Météorologie Nationale Maroc, Amal, Sayouri, Kingston, Jamaica Rabat, Morocco Met Éireann, Irish Meteorological Service, Spillane, Sandra, National Snow and Ice Data Center, Scambos, Ted A., Dublin, Ireland University of Colorado Boulder, Boulder, Colorado Srivastava, India Meteorological Department, Jaipur, A. K., Scanlon, T., Department of Geodesy and Geoinformation, India Vienna University of Technology, Vienna, Austria NASA Langley Research Center, Stackhouse, Jr., Paul W., Schenzinger, Verena, Department of Meteorology and Hampton, Virginia Geophysics, University of Vienna, Austria Stammerjohn, Sharon, Institute of Arctic and Alpine Tahoe Environmental Research Schladow, S. Geoffrey, Research, University of Colorado Boulder, Boulder, Colorado Center, University of California at Davis, Davis, California Stanitski, Diane M., NOAA/OAR Earth System Research Schmid, Claudia, NOAA/OAR Atlantic Oceanographic and Laboratory, Boulder, Colorado Meteorological Laboratory, Miami, Florida Steinbrecht, Wolfgang, German Weather Service (DWD), Eawag, Swiss Federal Institute of Aquatic Schmid, Martin, Hohenpeissenberg, Germany Science and Technology, Kastanienbaum, Switzerland Servicio Meteorológico Nacional, Buenos Aires, Stella, José L., North Carolina State University, Schreck III, Carl J., Argentina Cooperative Institute for Climate and Satellites-North Deutscher Wetterdienst, Offenbach, Germany Stengel, M., Carolina, Asheville, North Carolina Stephenson, Kimberly, Department of Physics, Universities Space Research Association, NASA Selkirk, H. B., The University of the West Indies, Jamaica Goddard Space Flight Center, Greenbelt, Maryland Department of Physics, Stephenson, Tannecia S., Scripps Institution of Oceanography, University of Send, Uwe, The University of West Indies, Jamaica California, San Diego, La Jolla, California Strahan, Susan, Universities Space Research Association, Turkish State Meteorological Service, Ankara, Sensoy, Serhat, NASA Goddard Space Flight Center, Greenbelt, Maryland Turkey Streletskiy, Department of Geography, George Dimitri A., Department of Earth and Atmospheric Sciences, M., Sharp, Washington University, Washington, D.C. University of Alberta, Edmonton, Alberta, Canada NOAA/NESDIS Coral Reef Watch, College Strong, Alan E., NOAA/NESDIS National Centers for Environmental Shi, Lei, Park, Maryland, and Global Science and Technology, Inc., Information, Asheville, North Carolina Greenbelt, Maryland Sviii | AUGUST 2018

13 Triñanes, Laboratory of Systems, Technological Joaquin A., Science Systems and Applications, Inc., Sun-Mack, Sunny, Research Institute, Universidad de Santiago de Compostela, Hampton, Virginia Santiago de Compostela, Spain, and NOAA Atlantic Sutton, Adrienne J., NOAA/OAR Pacific Marine Oceanographic and Meteorological Laboratory, and Environmental Laboratory, Seattle, Washington Cooperative Institute for Marine and Atmospheric Studies, Swart, Sebastiaan, Department of Marine Sciences, Rosenstiel School of Marine and Atmospheric Science, University of Gothenburg, Gothenburg, Sweden, and University of Miami, Miami, Florida Department Caribbean Institute for Meteorology Trotman, Adrian R., of Oceanography, University of Cape Town, Rondebosch, and Hydrology, Bridgetown, Barbados South Africa Aerospace Engineering Sciences, University of M., Tschudi, NOAA/NOS Center for Operational Sweet, William, Colorado, Boulder, Boulder, Colorado Oceanographic Products and Services, Silver Spring, Maryland NASA Goddard Space Flight Center, Greenbelt, Tu c ke r, C. J., Takahashi, Kenneth S., Servicio Nacional de Meteorología e Maryland Hidrología del Perú, Lima, Perú Capacity Center for Climate and Weather Tye, Mari R., Tamar, Gerard, Grenada Airports Authority, St. George’s, Extremes (C3WE), National Center for Atmospheric Grenada Research, Boulder, Colorado Department of Physics, The University of Taylor, Michael A., van As, D., Geological Survey of Denmark and Greenland, the West Indies, Jamaica Copenhagen, Denmark Tedesco, Lamont–Doherty Earth Observatory, Columbia M., Institute for Marine and Atmospheric R. S. W., van de Wal, University, Palisades, New York, and NASA Goddard Institute Research Utrecht, Utrecht University, Utrecht, Netherlands of Space Studies, New York, New York Royal Netherlands Meteorological van der A, Ronald J., Thackeray, S. J., Centre for Ecology and Hydrology, Lancaster, Institute (KNMI), De Bilt, Netherlands United Kingdom van der Schalie, Robin, VanderSat B.V., Haarlem, NOAA/National Weather Service, Alaska R. L., Thoman, Netherlands Region, Fairbanks, Alaska Royal Netherlands Meteorological van der Schrier, Gerard, Thompson, Philip, Joint Institute for Marine and Atmospheric Institute (KNMI), De Bilt, Netherlands Research, University of Hawaii, Honolulu, Hawaii van der Werf, Guido R., Faculty of Earth and Life Sciences, L., Department of Earth Sciences, Simon Fraser Thomson, VU University Amsterdam, Netherlands University, Burnaby, British Columbia, Canada Caribbean Institute for Van Meerbeeck, Cedric J., Icelandic Meteorological Office, Reykjavik, T. , Thorsteinsson, Meteorology and Hydrology, Bridgetown, Barbados Iceland Cooperative Institute for Velden, Christopher S., Singapore Meteorological Service, Bertrand, Timbal, Meteorological Satellite Studies, University of Wisconsin– Singapore Madison, Madison, Wisconsin Yale University, New Haven, M.-L., Timmermans, University of California, Irvine, California Velicogna, I., Connecticut Verburg, Piet, National Institute of Water and Atmospheric TImofeyev, Maxim A., Institute of Biology, Irkutsk State Research, Hamilton, New Zealand University, Russia , Vickers, H., Norut Northern Research Institute, Tromsø NOAA/NESDIS Coral Reef Watch, College Tirak, Kyle V., Norway Park, Maryland, and Global Science and Technology, Inc., Vincent, Lucie A., Environment and Climate Change Canada, Greenbelt, Maryland Toronto, Ontario, Canada To b i n , Skie, Bureau of Meteorology, Melbourne, Victoria, Earth Observing Laboratory, National Center Vömel, Holger, Australia for Atmospheric Research, Boulder, Colorado Togawa, H., Tokyo Climate Center, Japan Meteorological NOAA/NESDIS National Centers for Vose, Russell S., Agency, Japan Environmental Information, Asheville, North Carolina Norwegian Institute for Nature Research, H., Tømmervik, Wagner, Wolfgang, Department of Geodesy and Tromsø, N or w ay Geoinformation, Vienna University of Technology, Vienna, Aristotle University, Thessaloniki, Greece Tourpali, Kleareti, Austria Trachte, Katja, Laboratory for Climatology and Remote Institute of Arctic Biology, University of Alaska Walker, D. A., Sensing, Faculty of Geography, University of Marburg, Fairbanks, Fairbanks, Alaska Marburg, Germany J., Walsh, International Arctic Research Center, University of Trewin, Blair C., Australian Bureau of Meteorology, Alaska Fairbanks, Fairbanks, Alaska Melbourne, Victoria, Australia Wang, Bin, Department of Atmospheric Science and IPRC, University of Hawaii, Honolulu, Hawaii State University of New York, Albany, New Wang, Junhong, York Six | AUGUST 2018 STATE OF THE CLIMATE IN 2017

14 Department of Geography and Anthropology, Wang, Lei, NASA Langley Research Center, Hampton, Wong, Takmeng, Louisiana State University, Baton Rouge, Louisiana Virginia M., Wang, Joint Institute for the Study of the Atmosphere and Department of Civil and Environmental Wood, E. F., Ocean, University of Washington, Seattle, Washington Engineering, Princeton University, Princeton, New Jersey Georgia Institute of Technology, Atlanta, Georgia Wang, Ray, Joint Institute for the Study of the Atmosphere and K., Wood, Wang, Sheng-Hung, Byrd Polar and Climate Research Ocean, University of Washington, Seattle, Washington Center, The Ohio State University, Columbus, Ohio Woolway, R. Iestyn, Department of Meteorology, University Wanninkhof, Rik, NOAA/OAR Atlantic Oceanographic and of Reading, Reading, United Kingdom Meteorological Laboratory, Miami, Florida B., Wouters, Institute for Marine and Atmospheric Research Tahoe Environmental Research Center, Watanabe, Shohei, Utrecht, Utrecht University, Utrecht, Netherlands University of California at Davis, Davis, California Xue, Yan, NOAA/NWS National Centers for Environmental Weber, Mark, University of Bremen, Bremen, Germany Prediction, Climate Prediction Center, College Park, NASA Goddard Space Flight Center, Greenbelt, M., Webster, Maryland Maryland Yin, Xungang, ERT Inc., NOAA/NESDIS National Centers for Weller, Robert A., Woods Hole Oceanographic Institution, Environmental Information, Asheville, North Carolina Woods Hole, Massachusetts Yoon, Huang, Department of Oceanography, University of Westberry, Toby K., Department of Botany and Plant Hawaii, Honolulu, Hawaii Pathology, Oregon State University, Corvallis, Oregon A., Alaska Fire Science Consortium, International Arctic York, Department of Ecology and Weyhenmeyer, Gesa A., Research Center, University of Alaska Fairbanks, Fairbanks, Genetics/Limnology, Uppsala University, Uppsala, Sweden Alaska Environment and Climate Change Whitewood, Robert, Woods Hole Oceanographic Institution, Woods Yu, Lisan, Canada, Toronto, Ontario, Canada Hole, Massachusetts Widlansky, Matthew J., Joint Institute for Marine and Centro Internacional para la Zambrano, Eduardo, Atmospheric Research, University of Hawaii, Honolulu, Investigación del Fenómeno El Niño, Guayaquil, Ecuador Hawaii Zhang, Huai-Min, NOAA/NESDIS National Centers for Jet Propulsion Laboratory, California Wiese, David N., Environmental Information, Asheville, North Carolina Institute of Technology, Pasadena, California Beijing Climate Center, Beijing, China Zhang, Peiqun, Wijffels, Susan E., Woods Hole Oceanographic Institution, Zhao, Guanguo, University of Illinois at Urbana–Champaign, Woods Hole, Massachusetts Urbana, Illinois Science Systems and Applications, Inc., Wilber, Anne C., Cold and Arid Regions Environmental and Zhao, Lin, Hampton, Virginia Engineering Research Institute, Lanzhou, China INNOVIM, NOAA Climate Prediction Wild, Jeanette D., Nanjing University of Information Science and Zhiwei, Zhu, Center, College Park, Maryland Technology, China Met Office Hadley Centre, Exeter, United Willett, Kate M., Alaska Fire Science Consortium, International Arctic R., Ziel, Kingdom Research Center, University of Alaska Fairbanks, Fairbanks, Jet Propulsion Laboratory, California Institute Willis, Josh K., Alaska of Technology, Pasadena, California Goddard Earth Sciences Technology and Ziemke, Jerry R., Alaska Division of Geological and Geophysical G., Wolken, Research, Morgan State University, Baltimore, Maryland, and Surveys, and International Arctic Research Center, University NASA Goddard Space Flight Center, Greenbelt, Maryland of Alaska Fairbanks, Fairbanks, Alaska Ziese, Markus G., Global Precipitation Climatology Center, Deutscher Wetterdienst, Offenbach am Main, Germany EDITORIAL AND PRODUCTION TEAM Griffin, Jessicca, Graphics Support, Cooperative Institute for Riddle, Deborah B., Graphics Support, NOAA/NESDIS National Centers for Environmental Information, Asheville, Climate and Satellites–NC, North Carolina State University, Asheville, North Carolina North Carolina Slagle, Mary, Graphics Support, TeleSolv Consulting LLC, Hammer, Gregory, Content Team Lead, Communications NOAA/NESDIS National Centers for Environmental and Outreach, NOAA/NESDIS National Centers for Information, Asheville, North Carolina Environmental Information, Asheville, North Carolina Sprain, Mara, Lead Graphics Production, Love-Brotak, S. Elizabeth, Technical Editor, LAC Group, NOAA/NESDIS NOAA/NESDIS National Centers for Environmental National Centers for Environmental Information, Asheville, Information, Asheville, North Carolina North Carolina Misch, Deborah J., Visualization Team Lead, Communications Graphics Support, TeleSolv Consulting Veasey, Sara W., LLC, NOAA/NESDIS National Centers for Environmental and Outreach, NOAA/NESDIS National Centers for Environmental Information, Asheville, North Carolina Information, Asheville, North Carolina Sx | AUGUST 2018

15 TABLE OF CONTENTS TABLE OF CONTENTS ... List of authors and affiliations i xvi ... Abstract 1 ... INTRODUCTION 1. 1.1: e ssential varia Bles ... 2 s ide Bar Climate ... 5 2. GLOBAL CLIMATE a ... 5 . Overview T ... 11 b. emperature ... lobal surface temperature . 11 1 G . L ake surface temperature ... 13 2 3 L and surface temperature extremes ... 15 . ropospheric temperature ... 16 4. T . S tratospheric temperature ... 18 5 . c Cryosphere ... 20 . P ermafrost thermal state 1 ... 20 2 . N orthern Hemisphere continental snow cover extent ... 22 3 . Alpine glaciers ... 23 d . H ydrological cycle ... 25 ... . urface humidity S 25 1 . T otal column water vapor ... 26 2 3 U pper tropospheric humidity . ... 27 4 . P recipitation ... 28 s 2.1: l Bar and sur FaCe pre Cipitation extremes ... 29 ide 5 C loudiness ... 31 . 6 . R iver discharge and runoff ... 33 7 . G roundwater and terrestrial water storage ... 34 S oil moisture . ... ... 35 8 9 . Dr ought ... ... 36 ... 1 0. Land evaporation ... 37 e . Atmospheric circulation ... ... 39 1 . M ean sea level pressure and related modes of variability ... 39 ... . urface winds S 41 2 . U pper air winds ... 43 3 f E arth radiation budget . ... 45 1 . Earth radiation budget at top-of-atmosphere ... 45 g. A tmospheric composition ... 46 1. ... 46 Long-lived greenhouse gases . O zone-depleting gases 2 ... 49 3. Aerosols ... 49 . S tratospheric ozone ... 51 4 tratospheric water vapor 5 S . ... 54 6. Tropospheric ozone ... 56 s Bar 2.2: t he t ropospheri C o zone a ssessment r eport ... 58 ide 7. Carbon monoxide ... 59 h . L and surface properties ... 61 1 L and surface albedo dynamics ... 61 . errestrial vegetation activity T . ... 62 2 s ide Bar 2.3: p henology oF terrestrial and Freshwater primary produ Cers ... 63 3 Biomass burning ... ... 67 . GLOBAL OCEANS ... ... 69 3. a . Overview ... 69 . S ea surface temperatures ... 69 b c . O cean heat content ... 72 74 ... s ide Bar 3.1: u npre Cedented three years oF glo Bal Coral Blea Ching 2014–2017 Sxi | AUGUST 2018 STATE OF THE CLIMATE IN 2017

16 d S alinity ... ... 77 . 1 . I ntroduction ... 77 ... . ea surface salinity S 78 2 . S ubsurface salinity ... 79 3 e G lobal ocean heat, freshwater, and momentum fluxes . ... 81 1 . S urface heat fluxes ... 82 2 S urface freshwater fluxes ... 83 . . Wind stress ... 83 3 4 . L ong-term perspective ... 84 f S ea level variability and change ... 84 . g. ... S urface currents 87 acific Ocean 1 P ... 87 . s ide Bar 3.2: n u ‘ a k ai : F looding in h awaii Caused By a “ sta Ck ” oF oCeanographi C pro Cesses ... 88 2 I ndian Ocean ... 91 . 3 . A tlantic Ocean ... 91 . M eridional overturning and oceanic heat transport circulation observations in the h N orth Atlantic Ocean ... 91 i G lobal ocean phytoplankton ... 94 . j lobal ocean carbon cycle G ... 96 . ir–sea carbon dioxide fluxes A . ... 96 1 2 . O cean acidification ... 98 3 C arbon inventories ... 98 . TROPICS 4. ... 101 THE . Overview ... 101 a b . E NSO and the tropical Pacific ... 102 ceanic conditions O . ... 102 1 2 . A tmospheric circulation: Tropics and subtropics ... 102 ropical intraseasonal activity . T c ... 104 d . ... I ntertropical convergence zones 107 1 P acific ... 107 . 2 . Atlantic ... 109 . lobal monsoon summary ... 110 e G T ropical cyclones ... 112 f . . O verview ... 112 1 ... 2 A tlantic basin ... ... 114 . . E astern North Pacific and Central North Pacific basins ... 118 3 4 . W estern North Pacific basin ... 120 . N orth Indian Ocean basin ... 124 5 6 S outh Indian Ocean basin ... 125 . 7 . A ustralian basin ... 126 8 S outhwest Pacific basin ... 128 . g. T ropical cyclone heat potential ... 129 . I ndian Ocean dipole ... 132 h s ide Bar 4.1: h urri Cane i rma : r ewriting the re Cord Books ... 136 Bar 4.2: t he new goes -r series : m uCh improved “ glasses ” to view the tropi Cs ... 138 ide s ide Bar 4.3: h urri Cane h arvey : t he hallmark s oF a Busy and wet 2017 storm hurri Cane season For the u nited s tates ... 140 5. ... 143 THE ARCTIC a ... I ntroduction 143 . . S urface air temperature ... 144 b c . S ea surface temperature ... 146 . S ea ice cover ... 147 d 1 . S ea ice extent ... 147 . A 2 ge of the ice ... 148 ... 149 3 . S ea ice thickness and snow depth Sxii | AUGUST 2018

17 s Bar 5.1: p aleo Climate re Cords : p roviding Context and understanding ide oF Current a rCti C Change ... 150 ... G reenland ice sheet 152 e . . S urface melting ... 152 1 2 S urface mass balance . ... ... 153 3 . A lbedo ... ... 153 4 T otal mass balance ... ... 154 . 5 . M arine-terminating glaciers ... 154 6 S urface air temperatures ... 154 . f . G laciers and ice caps outside Greenland ... 156 ide Bar 5.2: i ndigenous knowledge and the s Coprodu Ction oF knowledge pro Cess : C reating a holisti C understanding oF a rCti C Change ... 160 g. T errestrial permafrost ... 161 1 . P ermafrost temperatures ... 162 ... 2 A ctive layer thickness .. 164 . h . T undra greenness 165 ... ide Bar 5.3: w ildland Fire in Boreal s a rCti C n orth a meri Ca ... 167 and i . T errestrial snowcover in the Arctic ... 169 j O zone and UV radiation ... 171 . ... 6. 175 ANTARCTICA a Overview ... 175 . b . A tmospheric circulation and surface observations ... 176 (P – E) N et precipitation . ... ... 179 c d . S easonal melt extent and duration ... 181 ea ice extent, concentration, and seasonality . S e ... 183 S f . ... outhern Ocean ... 185 1 U pper ocean ... 185 . 2 . I ntermediate ocean ... 187 ... . iogeochemical status: Continued ocean acidification B 187 3 ide Bar 6.1: r eturn oF the m aude r ise polynya : C litmus or sea iCe anomaly ? ... 188 s limate g. 2 017 Antarctic ozone hole ... 190 REGIONAL CLIMATES ... 193 7. a Overview ... 193 . b . N orth America ... 193 1 C anada ... 193 . 2 . U nited States ... ... 195 3 M exico ... ... 197 . c entral America and the Caribbean C ... 199 . entral America C . ... 199 1 2 . C aribbean ... 200 s Bar 7.1: i mpa Cts From h urri Canes i rma and m aria in the C ari BB ean ... 202 ide d . South America ... 204 1 N orthern South America ... 204 . entral South America C . ... 205 2 3 . S outhern South America ... 207 Bar 7.2: t he 2017 Coastal e l n iño ... 210 ide s . Af e ... ... 212 rica 1 . N orth Africa ... ... 212 est Africa W . ... 214 2 3 . E astern Africa ... 216 . S 4 outhern Africa ... 217 ... 220 5 . W estern Indian Ocean island countries Sxiii | AUGUST 2018 STATE OF THE CLIMATE IN 2017

18 f E urope and the Middle East ... 222 . 1 . O verview ... ... 222 . C entral and western Europe ... 224 2 3 . T ... 226 he Nordic and the Baltic countries . I berian Peninsula ... 227 4 editerranean, Italy, and Balkan states 5 M . ... 229 6 . E astern Europe ... ... 230 7 M iddle East ... 232 . g. ... Asia ... 232 1 O verview ... ... 233 . 2 . Russia ... 233 . E ast and Southeast Asia ... 237 3 4 . S outh Asia ... 239 . S 5 outhwest Asia ... 242 s ide Bar 7.3: a Bnormal w est C hina autumn rain Fall in 2017 and persisten Ce oF p aCiFiC –J apan pattern in a ugust 2017 ... 243 the h Oc eania ... 245 . 1 . O verview ... ... 245 2 N orthwest Pacific and Micronesia ... 245 . outhwest Pacific S . ... 247 3 4 . A ustralia ... 248 N ew Zealand 5. ... ... 251 s ide Bar 7.4: s ummer arrives early in a ustralia as the austral spring Breaks re Cords ... 253 ... 255 APPENDIX 1: Relevant Datasets and Sources 265 ... ACKNOWLEDGMENTS ... 269 ACRONYMS AND ABBREVIATIONS 270 ... ... REFERENCES Sxiv | AUGUST 2018

19 Sxv | AUGUST 2018 STATE OF THE CLIMATE IN 2017

20 —J. BLUNDEN, G. HARTFIELD, AND D. S. ARNDT ABSTRACT melt index were the second highest since 2005, mostly due to In 2017, the dominant greenhouse gases released into Earth’s strong positive anomalies of air temperature over most of the atmosphere—carbon dioxide, methane, and nitrous oxide— West Antarctic coast. In contrast, the East Antarctic Plateau reached new record highs. The annual global average carbon saw record low mean temperatures in March. The year was dioxide concentration at Earth’s surface for 2017 was 405.0 also distinguished by the second smallest Antarctic ozone hole ± 0.1 ppm, 2.2 ppm greater than for 2016 and the highest in observed since 1988. the modern atmospheric measurement record and in ice core Across the global oceans, the overall long-term SST warming records dating back as far as 800 000 years. The global growth trend remained strong. Although SST cooled slightly from 2016 rate of CO has nearly quadrupled since the early 1960s. 2 to 2017, the last three years produced the three highest annual With ENSO-neutral conditions present in the central and values observed; these high anomalies have been associated eastern equatorial Pacific Ocean during most of the year and with widespread coral bleaching. The most recent global coral weak La Niña conditions notable at the start and end, the global bleaching lasted three full years, June 2014 to May 2017, and - temperature across land and ocean surfaces ranked as the sec was the longest, most widespread, and almost certainly most ond or third highest, depending on the dataset, since records destructive such event on record. Global integrals of 0–700- began in the mid-to-late 1800s. Notably, it was the warmest m and 0–2000-m ocean heat content reached record highs in non-El Niño year in the instrumental record. Above Earth’s 2017, and global mean sea level during the year became the surface, the annual lower tropospheric temperature was also highest annual average in the 25-year satellite altimetry record, either second or third highest according to all datasets ana - rising to 77 mm above the 1993 average. lyzed. The lower stratospheric temperature was about 0.2°C In the tropics, 2017 saw 85 named tropical storms, slightly higher than the record cold temperature of 2016 according to above the 1981–2010 average of 82. The North Atlantic basin most of the in situ and satellite datasets. was the only basin that featured an above-normal season, its Several countries, including Argentina, Uruguay, Spain, and seventh most active in the 164-year record. Three hurricanes Bulgaria, reported record high annual temperatures. Mexico in the basin were especially notable. Harvey produced record broke its annual record for the fourth consecutive year. On 27 rainfall totals in areas of Texas and Louisiana, including a storm January, the temperature reached 43.4°C at Puerto Madryn, total of 1538.7 mm near Beaumont, Texas, which far exceeds Argentina—the highest temperature recorded so far south the previous known U.S. tropical cyclone record of 1320.8 mm. (43°S) anywhere in the world. On 28 May in Turbat, western Irma was the strongest tropical cyclone globally in 2017 and the Pakistan, the high of 53.5°C tied Pakistan’s all-time highest strongest Atlantic hurricane outside of the Gulf of Mexico and temperature and became the world-record highest tempera - −1 Caribbean on record with maximum winds of 295 km h ture for May. . Maria In the Arctic, the 2017 land surface temperature was 1.6°C caused catastrophic destruction across the Caribbean Islands, above the 1981–2010 average, the second highest since the including devastating wind damage and flooding across Puerto record began in 1900, behind only 2016. The five highest annual Rico. Elsewhere, the western North Pacific, South Indian, and Arctic temperatures have all occurred since 2007. Exceptionally Australian basins were all particularly quiet. high temperatures were observed in the permafrost across Precipitation over global land areas in 2017 was clearly above the Arctic, with record values reported in much of Alaska and - the long-term average. Among noteworthy regional precipita northwestern Canada. In August, high sea surface temperature tion records in 2017, Russia reported its second wettest year (SST) records were broken for the Chukchi Sea, with some on record (after 2013) and Norway experienced its sixth wet - regions as warm as +11°C, or 3° to 4°C warmer than the long- test year since records began in 1900. Across India, heavy rain term mean (1982–present). According to paleoclimate studies, and flood-related incidents during the monsoon season claimed today’s abnormally warm Arctic air and SSTs have not been around 800 lives. In August and September, above-normal observed in the last 2000 years. The increasing temperatures precipitation triggered the most devastating floods in more have led to decreasing Arctic sea ice extent and thickness. On than a decade in the Venezuelan states of Bolívar and Delta 7 March, sea ice extent at the end of the growth season saw Amacuro. In Nigeria, heavy rain during August and September its lowest maximum in the 37-year satellite record, covering caused the Niger and Benue Rivers to overflow, bringing floods 8% less area than the 1981–2010 average. The Arctic sea ice that displaced more than 100 000 people. minimum on 13 September was the eighth lowest on record Global fire activity was the lowest since at least 2003; how - and covered 25% less area than the long-term mean. ever, high activity occurred in parts of North America, South Preliminary data indicate that glaciers across the world lost America, and Europe, with an unusually long season in Spain mass for the 38th consecutive year on record; the declines and Portugal, which had their second and third driest years are remarkably consistent from region to region. Cumulatively on record, respectively. Devastating fires impacted British since 1980, this loss is equivalent to slicing 22 meters off the Columbia, destroying 1.2 million hectares of timber, bush, and top of the average glacier. grassland, due in part to the region’s driest summer on record. Antarctic sea ice extent remained below average for all of In the United States, an extreme western wildfire season 2017, with record lows during the first four months. Over burned over 4 million hectares; the total costs of $18 billion the continent, the austral summer seasonal melt extent and tripled the previous U.S. annual wildfire cost record set in 1991. Sxvi | AUGUST 2018

21 1. INTRODUCTION — D. S. Arndt, J. Blunden, and of impacts and extremes. Several sidebars deal with G. Hartfield extreme precipitation, how it is assessed, or weather This is the 28th issuance of the annual assessment systems that delivered extreme precipitation during now known as State of the Climate , published in the 2017. A multiyear look at this decade’s repeated coral Bulletin since 1996. As a supplement to the Bulletin, bleaching episodes provides a thorough, if sobering, its foremost function is to document the status and assessment. Other sidebars address the profound trajectory of many components of the climate system. physical and human toll of the 2017 North Atlantic However, as a series, the report also documents the hurricane season. Some feature new and advanced status and trajectory of our capacity and commitment technologies for observing the climate system, while to observe the climate system. others highlight the value of less familiar observation The year was nominally characterized as “ENSO approaches: those dealing with phenology, paleocli - neutral,” although most metrics indicate La Niña mate records, and, for the second consecutive State of or nearly La Niña status early and late in the year. the Climate report, indigenous knowledge. As is typical for this series, the characterization of Our cover this year ref lects the interplay between ENSO status varies slightly by discipline, region, and - the climate and living systems in the state of Cali available pertinent data. Readers may notice some fornia. The early-2017 “superbloom” depicted on the variation in the characterization and timing of ENSO front cover was an immediate response to the first status from section to section. productive wet season in several years for the region. If the report’s authors and their datasets are the Unfortunately, much of that new additional biomass lifeblood of this series, the chapter editors are surely became fuel for raging wildfires later in the year, as - the heart. They drive the development of their chap depicted on the back cover. The two covers illustrate ters, keeping pace with the evolution of available data, these dichotomous outcomes—sublime and serene available authors, and the state of the science. The on the front, destructive on the back—with the same majority of sections of this report are updates. Al - color palette. though new technologies, new analysis methods, and - We are saddened at the news of Dr. Olga Buly new datasets contribute dynamism to this volume, it gina’s passing in June, as this document was being is inevitable that some passages, particularly those - assembled. Dr. Bulygina was a constant in the build that describe observational or analytical methods, ing, sharing, and analysis of climatological datasets. borrow heavily from the text of previous reports. She was a reliable and skillful author of the Russian Changes in this year’s report, relative to recent years, State of the Climate section for many editions of the , include: explicit treatments of ocean acidification including this one. We will remember her fondly, and observations in both the Global Oceans chapter’s we wish her family and her colleagues well. global ocean carbon cycle section and the Antarctica An overview of findings is presented in the chapter’s Southern Ocean section; a subdividing of Abstract, Fig. 1.1, and Plate 1.1. Chapter 2 features the Arctic chapter’s section on sea ice cover to include global-scale climate variables; Chapter 3 highlights explicit and separate analyses of ice age, extent, and the global oceans; and Chapter 4 discusses tropical thickness, including snow depth; a diversification of climate phenomena including tropical cyclones. The - authors and nations addressed in the African sec Arctic and Antarctica respond differently through tion; and streamlining and combining of subsections time and are reported in separate chapters (5 and 6, within several chapters. respectively). Chapter 7 provides a regional perspec - This edition’s 16 sidebar articles remind us that the tive authored largely by local government climate climate is not experienced in annual averages, and specialists. A list of relevant datasets and their sources - that living systems, including humans, experience cli for all chapters is provided as an Appendix. mate change and variability most deeply in the form | S1 AUGUST 2018 STATE OF THE CLIMATE IN 2017

22 | S2 AUGUST 2018

23 J. BLUNDEN, R. J. — ESSENTIAL CLIMATE VARIABLES . DUNN, D. S. ARNDT, H AND G. HARTFIELD ECVs in this edition that are considered “partially Time series of major climate indicators are monitored,” meeting some but not all of the above again presented in this introductory chapter. Many requirements, include: of these indicators are essential climate variables Atmospheric Upper Air: cloud properties (ECVs), originally defined in GCOS (2003) and • • Atmospheric Composition: aerosols and updated again by GCOS (2010). The following ECVs, included in this edition, are considered “fully their precursors monitored,” in that they are observed and analyzed • Ocean Surface: carbon dioxide, ocean acidity across much of the world, with a sufficiently long- • Ocean Subsurface: current, carbon term dataset that has peer-reviewed documentation: Terrestrial: soil moisture, permafrost, glaciers Atmospheric Surface: air temperature, pre - • and ice caps, river discharge, groundwater, ice sheets, fraction of absorbed photosynthetically ac - cipitation, air pressure, water vapor, wind tive radiation, lakes, biomass, fire disturbance speed and direction Atmospheric Upper Air: Earth radiation bud Remaining ECVs that are desired for the future - • get, temperature, water vapor, wind speed include: Atmospheric Surface: surface radiation and direction • budget • Atmospheric Composition: carbon dioxide, methane, other long-lived gases, ozone • Ocean Surface: sea state • • Ocean Subsurface: nutrients, ocean tracers, Ocean Surface: temperature, salinity, sea level, sea ice, current, ocean color, phyto ocean acidity, oxygen - Terrestrial: water use, land cover, leaf area plankton • Ocean Subsurface: temperature, salinity • index, soil carbon • Terrestrial: snow cover, albedo 1. Global (or representative) average time series for essential climate variables through 2017. Anomalies lAte 1. P are shown relative to the base period in parentheses although base periods used in other sections of the report may differ. The numbers in the square brackets that follow in this caption indicate how many reanalysis (blue), satellite (red), and in situ (black) datasets are used to create each time series in that order. (a) N. Hemisphere polar stratospheric ozone (March) [0,0,1]; (b) S. Hemisphere polar stratospheric ozone (October) [0,0,1]; (c) - Arctic air temperature (60°–90°N) [0,0,1]; (d) Surface temperature [0,0,4]; (e) Lower tropospheric tempera ture [3,2,4]; (f) Lower stratospheric temperature [3,3,4]; (g) Extremes [warm days (solid) and cool nights (dotted)] [0,0,1]; (h) Arctic sea ice extent [max (solid) and min (dashed)] [0,0,1]; (i) Antarctic sea ice extent [max (solid) and min (dashed)] [0,0,1]; (j) Glacier cumulative mean specific balance [0,0,1]; (k) N. Hemisphere snow cover extent [0,1,0]; (l) Lower stratospheric water vapor [0,0,1]; (m) Cloudiness [0,8,0]; (n) Total column water vapor - land [3,1,1]; (o) Total column water vapor - ocean [3,2,0]; (p) Upper tropospheric humidity [0,2,0]; (q) Specific humidity - land [3,0,4]; (r) Specific humidity - ocean [3,1,3]; (s) Relative humidity - land [3,0,4]; (t) Relative humidity - ocean [3,0,2]; (u) Precipitation - land [0,0,4]; (v) Southern Oscillation index [0,0,1]; (w) Ocean heat content (0–700m) [0,0,5]; (x) Sea level rise [0,0,1]; (y) Tropospheric ozone [0,1,0]; (z) Tropospheric wind speed at 850 hPa for 20°–40°N [4,0,1]; (aa) Land wind speed [0,0,1]; (ab) Ocean wind speed [3,1,0]; (ac) Biomass burning [0,3,0]; (ad) Soil moisture [0,1,0]; (ae) Terrestrial groundwater storage [0,1,0]; (af) Fraction of absorbed photosynthetically active radiation (FAPAR) [0,1,0]; (ag) Land surface albedo - visible (solid) and infrared (dashed) [0,1,0]. | S3 AUGUST 2018 STATE OF THE CLIMATE IN 2017

24 1.1. Geographical distribution of selected notable climate anomalies and events in 2017. . ig F | S4 AUGUST 2018

25 Anomalously high upper-level divergence along 2. A l c lim A t e —R. J. H. Dunn, D. M. Stanitski, lo g B with strong tropical easterly wave disturbances may N. Gobron, and K. M. Willett, Eds. have contributed to the high levels of storm activity a. Overview— R. J. H. Dunn, D. M. Stanitski, N. Gobron, and during the Atlantic hurricane season. More gener - ally, upper-air winds from radiosonde measurements K. M. Willett The global land and ocean surface temperature continued to show no strong trend, with reanalyses was remarkably high in 2017. Depending on the data indicating a slight increase in average wind speed - (see Dee et al. 2011b about the use of reanalyses for set considered, the past year ranked as the second or climate monitoring). Surface winds over land contin - third highest since records began in the mid-to-late 1800s at 0.38°–0.48°C above the 1981–2010 average. ued a slow increase from the multidecadal decrease in globally averaged wind speeds observed since Notably, as ENSO conditions were neutral throughout much of 2017, it was the warmest year not inf luenced the ~1960s, most clearly seen in central and eastern by El Niño in the instrumental record, as well as being Asia. Over the oceans, there is disagreement between warmer than any year before 2015. satellite and reanalysis estimates as to whether wind - Unsurprisingly, lake surface temperatures, fre speeds were above or below average. quencies of land surface temperature extremes, and The emissions and atmospheric abundance of most - tropospheric temperatures also had high, but not re ozone-depleting substances continued to decline due to the positive effects of the Montreal Protocol and its cord-breaking, global anomalies in 2017. Many other Amendments; however, the atmospheric abundance essential climate variables (ECVs; Bojinski et al. 2014) of CFC-11 declined more slowly than expected from and other measures of the climate system responded to the predominantly above-average temperatures mid-2015 to mid-2017, potentially leading to a delay in the recovery of stratospheric ozone. (see also Plate 1.1). Exceptionally high temperatures were observed in the permafrost across the American Annual mean total stratospheric ozone levels and European Arctic, with record values observed in in 2017 were above average over almost the entire large parts of Alaska and northwestern Canada. Pre - Southern Hemisphere, with Antarctic values more than 10 Dobson units above the 1998–2008 average. liminary data indicate that glaciers across the world This is due to a weakened polar vortex when the continued to lose mass for the 38th consecutive year quasi-biennial oscillation (QBO) was in the east phase on record; the declines are remarkably consistent in late 2017, an enhanced Brewer–Dobson circulation from region to region. Cumulatively since 1980, this loss is the equivalent of slicing 22 meters off the top transporting ozone into the middle to high latitudes, of the average glacier. and the small size and depth of the ozone hole. The long-term upward trend of hemispheric and global The continued warmth resulted in a humid year over both land and oceans in terms of specific humid - average tropospheric ozone continued into 2017. There were lower concentrations of aerosols in ity, but more arid in terms of relative humidity over 2017 over highly populated areas in Europe, North land. Total column water vapor corroborated the surface specific humidity record, dropping slightly America, and China. Trends of total aerosol optical compared to the previous year over both land and depth (AOD) since 2003 have been negative over - ocean, but still remaining above average in most lo - Amazonia, the eastern U.S., southern Europe, north cations. A similar drop from 2016 was observed over ern Africa, China, and Japan, possibly from declining deforestation and anthropogenic aerosol emissions as the land surface area affected by drought. Global land evaporation was much lower than 2016 and below the well as reduced dust episodes in desert regions; but trends were positive over the Indian subcontinent. long-term average for the year. However, precipitation Near-record high stratospheric water vapor over global land areas was above the long-term aver - age (by 15–80 mm depending on the dataset used). - anomalies occurred by the middle of 2017 after a re cord low in December 2016, as confirmed by both the This year we include a sidebar (2.1) on precipita - tion extremes. Extreme precipitation is multifaceted, Microwave Limb Sounder satellite measurement Aura and balloon-borne frost point hygrometer soundings. depending on the timescales over which it is assessed and the average conditions experienced by a given This was possibly caused by tropical upwelling linked region. A particular focus is on Hurricane Harvey, to the QBO. where 5-day total rainfall amounts broke previous A sidebar (2.2) describes the first Tropospheric Ozone Assessment Report (TOAR), completed in Oc station records in some locations in Texas by over a - factor of three. tober 2017, highlighting a wide range of tropospheric ozone metrics produced using data from thousands | S5 AUGUST 2018 STATE OF THE CLIMATE IN 2017

26 P 2.1. (a) NOAA/NCEI surface temperature lAte (NOAAGlobalTemp); (b) Satellite-derived lake surface water temperature); (c) GHCNDEX warm day threshold exceedance (TX90p); (d) GHCNDEX warm night threshold exceedance (TN90p); (e) ERA-Interim lower tropospheric temperature grid anomalies; (f) ERA-Interim gridpoint lower stratosphere temperature anoma - lies; (g) ESA CCI average surface soil moisture anomalies; | S6 AUGUST 2018

27 P 2.1. ( cont .) (h) GRACE difference in annual lAte mean terrestrial water storage between 2016 and 2017; (i) GPCP v2.3 map of annual mean precipitation anomalies; (j) Percentile of annual precipitation total from 2017 GPCC First Guess Daily; (k) GHCNDEX 2017 anomalies for maxi - mum 1 day precipitation total (Rx1day); (l) JRA-55 global distribution of runoff anomaly; (m) JRA-55 global distribution of river discharge anomaly; (n) HadISDH annual average anomaly surface specific humidity over land; | S7 AUGUST 2018 STATE OF THE CLIMATE IN 2017

28 P 2.1. ( cont .) (o) ERA-Interim annual average lAte anomaly surface relative humidity; (p) PATMOS-x/ AVHRR global cloudiness anomaly; (q) Microwave UTH anomalies; (r) Total column water vapor anomaly from satellite radiometers (oceans) and COSMIC (land); (s) Mean scPDSI for 2017. Droughts are indicated by negative values (brown), wet episodes by positive values (green); (t) GLEAM land evaporation anomalies; (u) HadSLP2r sea level pressure anomalies; | S8 AUGUST 2018

29 P 2.1. ( cont .) (v) Land surface wind speed anomalies (circles: observational HadISD2 and Austra - lAte lian datasets, and worldwide shaded grids: MERRA-2); (w) ERA-Interim upper air winds; (x) Global distribution of OMI/MLS tropospheric column ozone annual mean anomalies (in Dobson Units) for year 2017 relative to the 2005-2016 average field. White areas poleward of 60°N and 60°S were flagged as missing due to lack of sufficient OMI ozone measurements during winter polar night to calculate annual averages; (y) GOME-2 2017 total column ozone anomalies [using GOME, SCIAMACHY, and GOME-2 (GSG) for 1998–2008 climatology]; (z) Anomalies of total AOD at 550 nm; (aa) Anomalies of dust AOD at 550 nm; | S9 AUGUST 2018 STATE OF THE CLIMATE IN 2017

30 (ab) Anomalies of biomass burning AOD at 550 nm; (ac) Visible broadband albedo anoma P 2.1. ( cont .) lAte - lies; (ad) Near-infrared broadband albedo anomalies; (ae) FAPAR anomalies; (af) GFASv1.4 carbonaceous emission from biomass burning; (ag) CAMS total column CO anomalies. | S10 AUGUST 2018

31 of global surface sites. The metrics are focused on availability of high-quality, high-resolution, and timely datasets is impinging on the ability to monitor the impacts of tropospheric ozone on human health, the climate in these cases. Improved (open) access to vegetation, and climate, and are based on the TOAR’s large database of surface hourly ozone observations. data, continued stable monitoring, and near-real time In 2017, there were no regional biomass burning data releases all help in allowing accurate assessments events that had a global impact on the annual carbon of current changes. monoxide (CO) regional burden, evidenced by the Time series and anomaly maps for many variables described in this chapter are shown in Plates 1.1 and fact that 2017 had the lowest CO burden since 2003. 2.1 respectively. Many sections refer to online fig - In Indonesia and central Africa, the CO burden was considerably lower than in previous years due to ures that can be found here (http://doi.org/10.1175 reduced fire activity. Globally during 2017, the levels /2018BA MSStateof t heClimate.2). of fire activity (as opposed to impacts or losses) were the lowest since at least 2003, 15% below the 2003–16 b. Temperature average. However, stronger activity occurred in North lobal 1) G A. Sánchez-Lugo, — temperatures surface America, Europe, and Siberia, with an unusually long C. Morice, P. Berrisford, and A. Argüez The 2017 global surface temperature was the season in Portugal and northwestern Spain, and the second or third highest annual global temperature worst fires experienced in recent history in British since records began in the mid-to-late 1800s at Columbia in terms of burned area. 0.38°–0.48°C above the 1981–2010 average (Table 2.1; A sidebar (2.3) focusing on land surface phenology observations in the Northern Hemisphere is also in - Fig. 2.1), according to four independent in situ analy - - cluded. In contrast to meteorological and hydrologi ses (NASA-GISS, Hansen et al. 2010; HadCRUT4, cal observations, which give a physical description Morice et al. 2012; NOAAGlobalTemp, Smith et al. of the current climate, phenology information shows 2008, Huang et al. 2015; JMA, Ishihara 2006). The how the natural environment is responding as the 2017 value was lower than the record set in 2016 state of the climate changes over time. and, depending on the dataset, 2015, both of which A common theme across a number of sections were years inf luenced by a strong El Niño episode. is the lack of available data to adequately monitor In contrast, ENSO-neutral conditions were present the climate and make assessments of change. While across the tropical Pacific Ocean during much of 2017, naturally an issue for this publication, we believe it is transitioning to La Niña in October. Despite this, worth highlighting more widely. Some examples from global temperature anomalies were high throughout this chapter include surface humidity, where no ob - the year, resulting in the warmest non-El Niño year servational marine product is currently available; ter - - on record. Separately, the global land annual tem restrial water storage, where no satellite observations perature ranked as either the second or third highest currently exist past June 2017; and subdaily (extreme) on record, again, depending on the dataset, and the globally averaged sea surface temperature (SST) was precipitation. Furthermore, there are several ECVs where various estimates are not in good agreement third highest. with each other or with reanalysis products. Limited t AB le 2.1. Temperature anomalies (°C) and uncertainties (where available) for 2017 wrt the 1981–2010 base period. Temperature anomalies provided in the table are the central values of a range of possible estimates. Uncertainty ranges are represented in terms of a 95% confidence interval. Note that the land values com - puted for HadCRUT4 used the CRUTEM.4.6.0.0 dataset (Jones et al. 2012), the ocean values were computed using the HadSST.3.1.1.0 dataset (Kennedy et al. 2011a, 2011b), and the global land and ocean values used the HadCRUT4.6.0.0 dataset. NOAA- HadCRUT4 NASA–GISS Global JMA ERA-Int JRA-55 MERRA-2 Global Temp Land +0.73 +0.66 ± 0.13 +0.70 ± 0.15 +0.69 +0.73 +0.70 +0.47 Ocean +0.30 ± 0.07 +0.31 ± 0.16 +0.28 +0.45 +0.38 +0.36 +0.35 Land and +0.39 +0.48 ± 0.05 +0.38 ± 0.08 +0.41 ± 0.15 +0.38 +0.53 +0.48 Ocean | S11 AUGUST 2018 STATE OF THE CLIMATE IN 2017

32 The global surface temperature analyses assessed here are derived from air temperatures observed at weather stations over land and SSTs observed from ships and buoys. Differences between analyses are mainly due to how each methodology treats areas with little to no data, such as the polar regions, and - how each analysis accounts for changes in measure ment methods [for more details see Kennedy et al. (2010); Hansen et al. (2010); Huang et al. (2015); and Sánchez-Lugo et al. (2017)]. The ranges of tempera - ture anomalies provided in this summary are ranges of best estimates for the assessed in situ analyses. - These ranges do not include uncertainty informa tion from each in situ analysis, which can be found in Table 2.1. The ten warmest years on record have all occurred since 1998, with the four warmest years occurring since 2014. Incrementally adding years to the analysis starting from 1988, each year initially ranks among the ten warmest years on record (with the exception of 2011, which ranked among the top twelve years at the time). The median value for the initial ranking since 1988 for a newly ended year is second or third highest, suggesting that the current ranking of 2017 is consistent with recent tendencies. In addition to the ranking, it is illustrative to dis - tinguish between warmer and colder years relative to the sustained trend (e.g., looking at the residuals from an ordinary least squares regression, Fig. 2.2). The average rate of change of global average surface −1 . How temperature since 1901 is 0.7°–0.9°C century - ever, this rate of change has nearly doubled in the −1 ). Relative to period since 1975 (1.5°–1.8°C century - the trend, the years 2008 and 2011 (both years inf lu F ig - . 2.1. Global average surface temperature anoma lies (°C; 1981–2010 base period). In situ estimate are shown from NOAA/NCEI (Smith et al. 2008), NASA- °C; Annual global temperature anomalies ( . 2.2. F ig GISS (Hansen et al. 2010), HadCRUT4 (Morice et al. displayed as dots) from 2007–17. Lines represent the - 2012), CRUTEM4 (Jones et al. 2012), HadSST3 (Ken linear trends over the 1975–2017 period, while the size nedy et al. 2011a,b), JMA (Ishihara 2006). Reanalyses of the dot represents the trend residuals. The black, estimates are shown from ERA-Interim (Dee et al. gray, red, and blue colors represent the NOAAGlo - 2011a), MERRA-2 (Bosilovich et al. 2015; Gelaro et balTemp, NASA GISS, JMA, and HadCRUT datasets, al. 2017) and JRA-55 (Ebita et al. 2011; Kobayashi et respectively. al. 2015). | S12 AUGUST 2018

33 enced by a strong La Niña) were considerably cooler cooler than average, including Antarctica. The 2017 than surrounding years and below the overall trend global ocean temperature is the second highest on record in all three reanalyses, whereas over global line, whereas 1998 and 2016 were not only considered land the temperature is the second highest in JRA- the warmest years on record when reported, but their values are considerably above the trend line. The 55 and ERA-Interim but only the fourth highest in MERRA-2, where temperatures were lower than in year 2014, on the other hand, was considered to be the warmest year on record at the time, even though 2016, 2005, and 2002. its value is near the 1975–2017 trend line. The 2017 ake 2) l anomaly is near the trend line for the HadCRUT4 R. I. Woolway, L. Carrea, — temperature surface series (~50th percentile) and above the trend in the C. J. Merchant, M. T. Dokulil, E. de Eyto, C. L. DeGasperi, other in situ datasets (~60th to 80th percentile). While J. Korhonen, W. Marszelewski, L. May, A. M. Paterson, the value of residuals may shift with the addition of A. Rimmer, J. A. Rusak, S. G. Schladow, M. Schmid, S. V. Shimaraeva, E. A. Silow, M. A. Timofeyev, P. Verburg, each new year of data, the current data suggest that the 2017 annual global temperature and ranking are S. Watanabe, and G. A. Weyhenmeyer - consistent with the progression of the upward trend Observed lake surface water temperature anoma lies in 2017 are placed in the context of the recent since the mid-1970s. During 2017, much-warmer-than-average condi - warming observed in global surface air temperature (Section 2b1) by collating long-term in situ lake tions were present across most of the world’s land and surface temperature observations from some of the ocean surfaces, with limited areas (parts of the north, world’s best-studied lakes and a satellite-derived central, and eastern Pacific Ocean, the southern global lake surface water temperature dataset. The Atlantic Ocean, eastern Indian Ocean, and a small area in western North America) experiencing near- to period 1996–2015, 20 years for which satellite-derived lake temperatures are available, is used as the base cooler-than-average conditions (Plate 2.1a). period for all lake temperature anomaly calculations. Global average surface air temperatures are also Warm-season averages (i.e., time periods without ice estimated using reanalyses. Reanalysis produces cover: July–September in the Northern Hemisphere datasets with uniform temporal and spatial coverage above 23.5°N and January–March in the Southern of the whole globe, but can suffer from regional model biases and the effects of changes in the observation Hemisphere below 23.5°S) are analyzed in line with previous lake surface temperature analyses (Schnei - network during the analysis period. However, surface der and Hook 2010; O’Reilly et al. 2015; Woolway and temperatures from reanalyses should be consistent Merchant 2017). Temperatures of lakes located within with observations in regions of good observational coverage. Here we consider three reanalyses: ERA- 23.5° of the equator are averaged over the whole year. Interim (Dee et al. 2011a), JRA-55 (Ebita et al. 2011; Satellite-derived lake surface water temperatures Kobayashi et al. 2015), and MERRA-2 (Bosilovich et for 688 lakes are used in this analysis to investigate al. 2015; Gelaro et al. 2017). The ERA-Interim 2-m global variations in lake surface water temperature. Satellite-derived surface water temperatures were temperature was adjusted by merging analyses over - retrieved during the day using the methods of Mac land with short forecasts over ocean and subtracting 0.1°C from the latter before 2002, in order to account Callum and Merchant (2012) on image pixels filled with water according to both the inland water dataset for a change in SST provider, following Simmons et of Carrea et al. (2015) and a ref lectance-based water al. (2017) and Simmons and Poli (2014). ERA-Interim detection scheme (Xu 2006). The satellite tempera - provides data from 1979, JRA-55 from 1958, and MERRA-2 from 1980. tures represent midmorning observations throughout the record (except at the highest latitudes, where According to the reanalyses, the annual global 2-m observations may be available at other times of day). temperature for 2017 was the second highest since their records began and was between 0.39°C and The observations were generated using data from the ATSR (Along Track Scanning Radiometer) series 0.53°C above average, depending on the reanalysis (Table 2.1). The temperatures for the warmest year, including ATSR-2 (1995–2003) and the Advanced 2016, ranged between 0.47°C and 0.62°C above aver - ATSR (AATSR) (2002–12), extended with MetOp-A AVHRR (2007–17). In this study, lake-wide average age. surface temperatures are used to remove the intralake ERA-Interim and MERRA-2 for 2017 also show heterogeneity of surface water temperature responses warmer-than-average conditions over many regions of the world (Online Figs. S2.1–S2.3), particularly to climate change (Woolway and Merchant 2018). over higher northern latitudes. A few regions were | S13 AUGUST 2018 STATE OF THE CLIMATE IN 2017

34 In 2017, satellite-derived lake surface tempera - temperature anomalies agree in this respect. For ex - ample, in situ measurements of temperature anomaly tures were lower than observed in 2016 by 0.3°C in the 688-lakes average (Fig. 2.3a), though the mean in Vättern (Sweden) were −0.03°C (i.e., below the anomaly for 2017 was still +0.4°C above the baseline, 20-year base period mean) in summer 2017. There is continuing the long-term lake surface warming trend a clear contrast between Scandinavian lake surface temperature anomalies and those in central Europe, identified in previous analyses (e.g., Woolway et al. with lake temperature anomalies in the latter region 2017) and ref lecting the observed increase in global surface air temperature (section 2b1). Lake surface up to 1°C higher than average (Plate 2.1b; Fig. 2.4). water temperatures in 2017 were the second highest This is also confirmed by in situ lake temperature since 1995 (the earliest satellite data used), behind anomalies, for example, +0.7°C in 2017 for Lake Zu - only 2016. Eight of the ten warmest years for lake - rich (Switzerland). Above-average lake surface tem surface waters in the record have occurred since 2007 - perature anomalies are also observed from the satel (1998 and 2001 rank fifth and ninth, respectively). lite data in northwest Canada and the western United Lake surface water temperatures in 2017 were not States, confirmed by in situ data (e.g., +0.8°C in Lake above average in all regions (Figs 2.3b,c; Plate 2.1b). Washington). Lakes in the central and eastern U.S. Below-average lake surface temperatures prevailed experienced near-normal lake surface temperatures in 2017, with some regions showing below-average throughout north and northwestern Europe (Plate 2.1b; Fig. 2.4) in summer, where lake surface tem - peratures were up to 1°C cooler than the 20-year base period mean. The satellite data and in situ lake F ig . 2.3. Annual lake surface water temperature anomalies 1995–2017 (°C; relative to 1996–2015). (a) Global average (with 95% confidence intervals) satel - lite-derived lake surface temperature anomalies; (b) satellite-derived lake surface temperature anomalies . 2.4. Comparisons of satellite-derived lake surface ig F for 688 lakes; and (c) in situ lake surface temperature water temperature anomalies (colored dots) to air anomalies for 34 globally distributed lakes. Annual lake surface temperature anomalies (calculated from the surface water temperatures anomalies are calculated NASA GISS Surface Temperature Analysis) in (a) for the warm season (Jul–Sep in NH; Jan–Mar in SH), North America and (b) Europe in 2017. Temperatures except within 23.5° of the equator, where the averages anomalies (°C; relative to 1996–2015) are calculated are taken over the whole year. for the NH warm season (Jul–Sep). | S14 AUGUST 2018

35 lake surface temperatures. These regional differences such values are still typically well above and below in lake surface temperature anomalies in 2017 ref lect the climatologically defined threshold of 36.5 days per year, respectively (Fig. 2.5). the July–September average surface air temperature Over areas where observations exist, the an - anomalies (relative to 1996–2015), calculated from the NASA GISS surface temperature analysis (Fig. nual occurrence of warm days (TX90p) and nights 2.4; Hansen et al. 2010; GISTEMP Team 2016). In (TN90p; Plates 2.1c,d) was typically well above the summary, surface air and lake water temperatures climatological average. In particular, eastern Asia experienced 20 more warm days than the threshold, in 2017 were generally coherent. whereas southern Europe and eastern Australia ex - s u r f a c e e x t r e m e s l a n d t e m p e r at u r e perienced more than 40 additional warm days. The — 3) frequency of warm nights was less than warm days S. E. Perkins-Kirkpatrick, M. G. Donat, and R. J. H. Dunn Changes in temperature extremes are important over Australia and southern Europe but was still 10–30 days and 30–40 days more than the threshold, for climate monitoring due to their sensitivity to respectively. Conversely, the U.S. and Canada expe - relatively small changes in average conditions. Small rienced slightly more warm nights than warm days. changes in average temperature can induce much Cool days and nights (TX10p, TN10p; Fig. 2.6) - larger changes in the intensity and frequency of cor responding heat extremes. Land surface temperature were less frequent than the threshold over some extremes during 2017 were characterized by overall regions, with around 20 fewer cool nights over the increased occurrences of warm temperatures and re - U.S. and Canada and 30 fewer nights for Europe. For duced occurrences of cooler temperatures compared northern regions with available data, annual minima during both daytime and nighttime (TXn, TNn, to long-term averages. A number of anomalously high temperature events occurred in 2017, in both Online Figs. S2.4c,d) were very high. The respective maximum and minimum daily temperatures. As in annual maxima, however, did not always display - similar anomalies (TXx, TNx; Online Figs. S2.4a,b). previous reports, the GHCNDEX quasi-global grid The frequency of warm daytime temperatures ded dataset (Donat et al. 2013b) is used to monitor (TX90p; Online Fig. S2.5) varied across the seasons. global temperature extremes over land. This is quasi- - global, as an absence of data over some locations During boreal winter (DJF 2016/17), warm day oc currences much higher than the threshold occurred hinders the robust calculation of extremes indices and their trends. A suite of temperature and precipitation over northern Europe and eastern China, Russia, extremes indices (Zhang et al. 2011) is first calculated and Australia. However, western Australia and the from observed daily station time series in the GHCN- Daily archive (Menne et al. 2012), before interpolat - ing the indices on global grids. Some of the fields of extremes indices have limited spatial coverage, especially across central and eastern Asia, for those derived from minimum temperatures compared to those from maximum temperatures. Therefore, complete coverage derived from the ERA-Interim reanalysis (Dee et al. 2011a) is shown separately in Online Figs. S2.7–S2.9. Results are presented for a selection of the tem - perature indices in GHCNDEX: TX90p (frequency of warm days when daily temperatures exceed the 90th percentile of daily maximum temperatures calculated over the 1961–90 base period), TX10p (cool day fre - quency, daily temperatures below the 10th percentile), Global average time series of the number of . 2.5. ig F TN90p and TN10p (warm and cool night frequency, (a) warm days (TX90p) and (b) cool nights (TN10p) respectively), and TXx, TXn, TNx, and TNn (extrema from GHCNDEX relative to 1961–90. (This reference - of annual maximum and minimum temperatures, re period is used for consistency with other ETCCDI in - spectively; see online supplement for full definitions). dex products.) By construction, these indices have an Averaged over areas where there are observations, average of 36.5 days over the reference period. The there were fewer warm days (TX90p) and more cool dotted black line shows the percent of land area with data. Units: days. nights (TN10p) in 2017 compared to 2016 . However, | S15 AUGUST 2018 STATE OF THE CLIMATE IN 2017

36 (https://public.wmo.int/en/media/news /records-fall-amid-heatwaves, accessed 16 February 2018). A heatwave engulfed southern and eastern Europe during late July and early August, causing human casualties. Boreal autumn (SON) saw higher-than-threshold occurrences of warm days (TX90p) over most regions except for northern Russia and Europe, where around five fewer warm days than the threshold occurred. For most regions, the anomalous frequency of seasonal warm days was larger than that of seasonal cool days (TX10p; Online Figs. S2.5e–h); however, there were still fewer cool days than the climatological average, giving warm anomalies. During each season, the vast majority of the globe experienced two to five fewer cool nights than the threshold (TN10p; Online Figs. S2.5m–p) but higher numbers of warm nights (TN90p; Online Figs. S2.5i–l), with scattered areas experiencing warm nights close to the climatological average. With the exception of JJA, much of the globe experienced minimum daytime temperatures (TXn; Online Figs. S2.6e–h) that were at least 2°C above the 1961–90 average. During JJA, such anomalies F . 2.6. (a) Cool days (TX10p) and (b) cool nights ig (TN10p) anomaly maps for 2017 (GNCHDEX). were up to 2°C below the climatological average over northeast Europe and East Asia. The former were caused by cyclonic activity, especially in June. The western U.S. and Canada experienced occurrences of signature of these events is also evident in the reduced warm days lower than the threshold. Near-average or frequency of warm days (TX90p) during JJA (Online higher than the threshold occurrences of warm days occurred during boreal spring (MAM), particularly Figs. S2.5a,b). Anomalies 2°C below average were also experienced over East Asia during SON. Minimum over northwest Russia and China, where more than nighttime temperatures (TNn, Online Figs. S2.6m–p) ten extra warm days were observed. were consistently warm throughout most seasons and With the exception of the eastern U.S. and northern Europe, all areas with available data saw were quite large (3°C or higher) over the U.S., Canada, around five or more warm days than average in the and Europe during MAM and DJF. boreal summer (JJA). During this period, numerous ropospheric 4) t J. R. Christy, S. Po-Chedley, — warm temperature events occurred worldwide: temperature Australia experienced its warmest winter on record and C. Mears based on daily maximum temperature observations Following the record high global lower tropo - spheric temperature (LTT) in 2016, LTT decreased by (w w w.bom.gov.au/climate/current/season /aus/archive/201708.summary.shtml, accessed 16 more than 0.1°C in 2017. The annual, globally aver - aged LTT (the bulk atmosphere below 10-km altitude) February 2018). However, over Southern Australia was, depending on the dataset, +0.38° to +0.58°C cool day occurrences during the austral winter above the 1981–2010 mean. 2017 was generally the (TN10p; Online Fig. S2.5) were higher than average, second or third warmest year since measurements indicative of very dry conditions early in the season. began in 1958 (Fig. 2.7) and the warmest non-El A severe heatwave also impacted the southwest Niño year. U.S. during June, resulting in temperatures so - high that some aircraft in Arizona and California Direct measurements of LTT by radiosonde da were grounded (www.climate.gov/news-features tasets have reasonable spatial coverage since 1958. Radiosonde data are complemented by satellites /event-tracker/heat-roasts-western-united-states, accessed 16 February 2018). Also during June, and reanalysis products since late 1978, except JRA- extreme temperatures of at least 50°C were 55 reanalyses which begin in 1958. These datasets reported for multiple locations in the Middle East are described in Christy (2016). These bulk-layer | S16 AUGUST 2018

37 F . 2.8. (a) Multivariate ENSO index (MEI; Wolter ig and Timlin 2011). (b) Latitude–time depiction of ERA-I LTT anomalies (°C; base period of 1981–2010, cosine ig (°C; 1981–2010 F . 2.7. Anomalies of global mean LTT ). g latitude weightin (a) radiosondes, (b) satellites, and (c) base period): reanalyses. Annually averaged LTT was above average over atmospheric temperatures are closely related to the most of the globe in 2017 (Plate 2.1e). Regionally, warm anomalies for the year occurred throughout heat content of the atmospheric climate system and - the Arctic poleward of 65°N. The midlatitude belts in thus are valuable indicators for quantifying heat en ergy changes expected from rising concentrations of both hemispheres featured areas with mostly above- greenhouse gases and other forcings. normal temperatures with centers in southwestern The latitude–time depiction of the LTT anomalies North America, southwestern Europe, central China, (Fig. 2.8) beginning in 1979 illustrates major tropo - the northern Pacific Ocean, southern midlatitude spheric responses to El Niño events, most clearly oceans, and eastern Australia. The Antarctic was evident in the tropics (1983, 1987, 1998, 2010, and generally cooler than average as were other scattered 2016). The major El Niños in 1998 and 2016 reveal locations. (Plate 2.1e). - comparable magnitudes of peak anomalies, but 2016 The long-term global LTT trend based on radio −1 is set against higher background temperatures. Since sondes (starting in 1958) is +0.17° ± 0.02°C decade . 2013, few zonal average anomalies have been negative. Starting in 1979 and using the average of radiosondes, Annual global LTT anomalies are closely tied to El satellites, and reanalyses (weighted one-third each), −1 . the trend is fairly similar, at +0.16° ±0.04°C decade Niños and La Niñas, which can be characterized by - the Multivariate El Niño Southern Oscillation Index The range represents the variation among the indi - vidual datasets which serves as a proxy for the struc (MEI; Wolter and Timlin 2011) shown in Fig. 2.8a. tural uncertainty seen in Fig. 2.7 and Table 2.2. Efforts As noted, 2017 followed a major El Niño (MEI > 2 to document and understand the differences among in early 2016) yet its global LTT experienced a small datasets continue. Accounting for the magnitude decline of less than 0.2°C while previous year-to-year of the year-to-year variations results in a statistical declines were greater (e.g., 1999 was over 0.4°C cooler −1 , meaning that than the El Niño year of 1998). Part of the reason was confidence range of ±0.06°C decade the trends are significantly positive. the rise of El Niño-like characteristics (MEI > 1.4) The positive trends noted in this assessment by May 2017 before La Niña conditions ensued. The LTT anomaly, which generally lags the MEI by 3–5 represent the net effect of both anthropogenic (e.g., months, apparently responded with record high val - increasing concentrations of greenhouse gases) and - ues in September and October, thereby mitigating the natural forcings. For example, major volcanic erup tions injected solar-ref lecting aerosols into the strato - late-year La Niña cooling effect in the annual average. | S17 AUgUSt 2018 StAte oF tHe climAte in 2017

38 − 1 2.2. Estimates le of lower tropospheric temperature decadal trends (°C decade t AB ) beginning in 1958 and 1979 from the available datasets. Tr o p i c a l LT T Tropical TTT Global LTT 1958 1979 1958 Start Year: 1958 1979 1979 Radiosondes +0.13 RAOBCORE +0.15 +0.15 +0.14 +0.13 +0.14 +0.19 RICH + 0.17 + 0.17 + 0.17 +0.18 +0.20 +0.20 +0.15 +0.15 +0.15 +0.15 +0.18 R ATPAC + 0.17 + 0 .11 +0.16 +0.15 UNSW (to 2015) +0.13 +0.10 Satellites a +0.13 — +0.12 — +0.12 — UAHv6.0 b +0.19 — +0.15 +0.19 — — RSSv4.0 — — — — +0.21 NOAAv4.0 — c — — — — — + 0.17 UWv1.0 Reanalyses and Climate Models after Reanalyses — +0.13 — +0.13 — ERA-I +0.10 +0.14 — +0.16 — +0.13 — JRA-55 +0.15 MERRA-2 — +0.14 — + 0.17 — CMIP5 Mean +0.22 +0.31 +0.25 +0.29 +0.27 +0.21 a The UAH LTT vertical profile is slightly different than the others with much less emphasis on surface emissions and slightly −1 warmer if using the traditional more in the midtroposphere. Calculations indicate UAH LTT would be +0.01°C decade LTT profile represented by other datasets here. b RSS value of TTT utilizes RSSv4.0 of MTT and RSSv3.3 of LST. c UW value of TTT utilizes MTT from UWv1.0 and LST from NOAAv4.0. the method of Fu et al. (2004), this year’s results are sphere in 1963, 1982, and 1991, depressing global temperatures for a few years each time. The latter two provided from a weighted average of the MTT and events, being early in the current 1979–2017 period, LST channels that largely removes the stratospheric tilted the global trend since 1979 to be more positive portion, producing a better estimate of the full tro - −1 by about +0.06°C decade (Christy and McNider posphere itself, - 2017). There is evidence that other time-varying fac Examining the various datasets of the tropical - tors such as internal climate variability related to oce TTT trend for 1979−2017 (Table 2.2), it is noted that the magnitude of the trend is similar to LTT in most anic processes, a recent reduction in solar irradiance, and/or the presence of aerosols from natural (e.g., cases and always greater than MTT (not shown). Using the average of 102 climate model simulations minor volcanic activity) and anthropogenic sources from the IPCC CMIP-5 (Flato et al. 2013), we see that also affected the temperature and likely had a role in because of the incorporation of more inf luence of the reducing post-2000 values and thus contributed to upper level tropospheric layers, for which trends are the so-called “warming hiatus” from 2000 to 2014 (Wuebbles et al. 2017). more positive than the lower troposphere (Christy Christy (2017) examined tropical trends for the 2017), the TTT trends slightly exceed those of LTT in most cases. layer centered in the midtroposphere (MTT), where trends are expected to respond rapidly to increases — J. R. Christy and C. Covey in greenhouse gases. However, the tropical MTT temperature tratospheric 5) s profile includes a small portion of the stratosphere The stratosphere is the atmospheric layer above where long-term cooling has occurred (not shown). the tropopause (~17 km altitude near the equator, ~9 km at the poles). Its upper boundary is ~50 km. This inf luence leads to an MTT trend that is cooler than would be measured in the troposphere alone Radiosondes have observed the stratosphere, typi - −1 cally up to ~20 km, with coverage sufficient for global . Following by approximately 0.03°–0.04°C decade | S18 AUGUST 2018

39 averaging since 1958. Since 1978 satellites have car - (1963 Mt. Agung, 1982 El Chichón, and 1991 Mt. ried microwave sounding units (MSUs) to monitor Pinatubo), whose stratospheric aerosols each led to warming spikes. After Pinatubo (and perhaps El the intensity of radiances which is directly related to lower stratospheric temperature (LST). The MSU Chichón), LST declined to levels lower than prior to the eruption, giving a stair-step appearance. Ozone LST channel detects emissions from ~14 to ~27 km - depletion and increasing CO in the atmosphere con with maximum signal at ~18 km; thus in the deep 2 tribute an overall decline, so trends in global LST are tropics there is some upper tropospheric inf luence. clearly negative until approximately 1996. Stratospheric sounding units (SSUs) monitor layers In Figs. 2.9a–c, the global trends through 2017, completely above the tropopause. In 2017, the annual globally averaged LST rose based on the average of all displayed datasets, are −1 for periods about 0.2°C from its value in 2016, which was the −0.29°, −0.27°, and +0.01°C decade record low in six of the nine datasets (Figs. 2.9a–c). beginning in 1958, 1979, and 1996, respectively. The Episodes in which the tropopause rises into typically satellite time series are in exceptional agreement with each other (r > 0.99) and with reanalyses (highest r stratospheric levels lead to cooler MSU LST values because upper tropospheric air is cooler than the with JRA-55 > 0.96). The radiosonde datasets are stratospheric air it displaces. This occurred during the limited by geographical coverage; even so, satellites major El Niño event of 2016. The 2017 anomaly was and radiosondes achieved r > 0.95. Absence of lower stratospheric cooling in the approximately −0.4°C, but varied among the datasets analyzed here by ± 0.2°C. global mean since 1996 is due to recovery of the ozone Observed long-term globally averaged LST time layer, especially at high latitudes, as the Montreal series in Figs. 2.9a–c include three volcanic events Protocol and its Amendments on ozone-depleting substances has taken effect (Solomon et al. 2017; Randel et al. 2017). The pattern of LST anomalies in 2017 is depicted in Plate 2.1f. Warmer-than-average conditions occurred poleward of 50°S and over the north polar Western Hemisphere (180°E to 360°E). In general below-average temperatures prevailed elsewhere, consistent with the generally negative trends in Figs. 2.9a–c. Two prominent features of LST are sudden strato - spheric warmings (SSWs) and the quasi-biennial oscillation (QBO). SSWs usually appear during the northern polar night. Figures 2.10a,b shows pentad (5-day average) LST anomalies for the north and south polar caps (65°–85° latitude average, values smoothed 1–2–1 in time). Excursions over the North Pole often exceed 10°C, with 5°C departures in almost every year. 2017 did not experience an event > 5°C in the north, but in pentad 66, near the end of 2016, both polar caps exceeded 5°C (unsmoothed). Because these events are related to the breakdown of the polar night vortex, they occur less frequently and with less intensity over the South Pole due to its more zonally symmetric circulation. Sudden cooling episodes also occur and are related to the impact of ozone depletion in spring over the south polar cap. The QBO is typically defined by the time–height pattern of zonal wind anomalies in the tropics, but it can also be detected in the LST temperature anoma - lies. The QBO alternates between westerly (warm) (°C; . 2.9. Time series of annual LST anomalies ig F and easterly (cold) wind shear regimes in the tropical 1981–2010 base period) : (a) radiosondes, (b) satellites, stratosphere in which the feature propagates down - (c) reanalyses, and (d) coupled climate models. (e) Up - per stratospheric temperature anomalies. ward from the upper stratosphere and dissipates near | S19 AUGUST 2018 STATE OF THE CLIMATE IN 2017

40 and most of the remaining decline from ozone loss (Aquila et al. 2016). c. Cryosphere t h e r m a l 1) J. Noetzli, — p e r m a f r o s t s tat e H. H. Christiansen , P. Deline, M. Gugliemin, K. Isaksen, V. E. Romanovsky, S. L. Smith, L. Zhao, and D. A. Streletskiy Permafrost is an invisible component of the cryosphere in polar and high mountain areas and is defined as earth materials (eg., soil, rock) that exist at or below 0°C continuously for at least two consecu - tive years. Long-term monitoring of its conditions primarily relies on ground temperatures measured in boreholes. Overlying the permafrost is the active layer, which thaws in summer and refreezes in winter. Globally, permafrost observation data (thermal state and active layer dynamics) are collected in the data - base of the Global Terrestrial Network for Permafrost (GTN-P; Biskaborn et al. 2015), which is part of the Global Climate Observing System (GCOS) of the World Meteorological Organization (WMO). - The long-term trend of rising permafrost tempera tures worldwide continued in 2017. There is, however, . 2.10. Time series of pentad (5-day averages, ig F considerable regional variability, mainly depending smoothed by 1-2-1 in time) LST anomalies (°C) for (a) northern and (b) southern polar caps bounded by 65° on the temperature range, surface characteristics, and and 85° latitude. (c) Monthly time series of QBO index ground ice content at the site. The general picture is as averaged from UAH, RSS, and NOAA LST. that a more substantial increase is observed in regions the tropopause. Figure 2.10c extends the temperature- with cold continuous permafrost compared to areas with warm permafrost at temperatures within 2°–3°C based QBO index of Christy and Drouilhet (1994) of the freezing point. This is mainly a result of latent t h r o u g h 2 017. The 16 QBO periods in Fig. 2.10c indicate a mean heat effects associated with melting of ground ice. The length of 27.4 months. The longest (35 months) ended lowest permafrost temperatures—and thus highest in April 2002, and the shortest by a substantial mar - warming rates—were observed in the high Arctic of northern Alaska, Canada, Svalbard, and Russia, gin (17 months) concluded in March 2017. The cycle as well as in shaded f lanks of high mountain peaks. that finished in 2017 included the weakest (warmest) easterly regime in this 39-year history. Record high temperatures were observed in 2017 for There is relatively high confidence in explaining nearly all sites in Alaska and in northwestern Canada. In other areas (northeastern Canada, Nordic coun - the variations of global mean stratospheric tempera - ture. When climate models used in the IPCC AR5 tries), permafrost temperatures measured in 2016/17 were among the highest ever recorded (updates from (Flato et al. 2013) are provided with forcing estimates related to changes in ozone, carbon dioxide, volcanic Christiansen et al. 2010; Romanovsky et al. 2017; aerosols, solar variability, etc., the multi-model mean Smith et al. 2015, 2017; Ednie and Smith 2015; Boike agrees with the satellite observations to a high level et al. 2018). A detailed description of permafrost (r > 0.96). Figure 2.9d shows the mean of 102 CMIP-5 conditions in the Arctic and sub-Arctic is provided simulations of the LST time series. Aquila et al. (2016) in Section 5g. Here, the focus is on updated results examined forcing agents and concluded that about from mountain permafrost (European Alps, Nordic ⅓ of the decline was due to increasing concentra countries, and central Asia) and permafrost in con - - tinental Antarctica (Streletskiy et al. 2017). tions of greenhouse gases and ⅓ to ozone-depleting substances. At higher levels of the stratosphere, in In mountain permafrost in the European Alps the layer monitored by the SSU channel 3 (~ 40–50 most boreholes are located between 2600 and 3000 m a.s.l. (above sea level), with permafrost temperatures km altitude; Fig. 2.9e), the observed trend is approxi - −1 of which 75% is estimated to typically above –3°C. Permafrost temperatures have mately −0.7°C decade result from enhanced greenhouse gas concentrations generally increased in the upper 20 m (Fig. 2.11), espe - | S20 AUGUST 2018

41 Aiguille du Midi Mont Blanc (France), permafrost temperature at 10-m depth continued to increase in the past two years and is expected to be at a very high level compared to the past decades (Fig. 2.11; updated from Magnin et al. 2015). In Nordic countries, mountain permafrost tem - - peratures continued to increase (updated from Isak sen et al. 2007; Christiansen et al. 2010). In southern Norway (Juvvasshøe) ground temperatures in 2017 were near-record high, a warming that followed a period of cooling between 2010 and 2013. Moni - toring (since 2008) in northern Norway (Iškoras) shows evidence of thawing permafrost with ground temperatures well above 0°C at 10-m depth since 2013/14 (Fig. 2.11). In the warm permafrost of the higher elevations of central Asia, ground tempera - tures are estimated to be in the range of –2° to –0.5°C (Zhao et al. 2017) and have increased by up to 0.5°C −1 since the early 1990s (update from Zhao et decade al. 2010). On the Qinghai–Tibetan Plateau (Fig. 2.12), the increase in ground temperature at 10-m depth F . 2.11. Temperature (°C) measured in permafrost ig boreholes in the European Alps and Nordic countries at depths of (a) ~10 m (monthly means) and (b) ~20 m - Monitor (Sources: Swiss Permafrost (annual means). - ing Network PERMOS; Norwegian Meteorological In stitute and the Norwegian Permafrost D atabase N OR - PERM; French Permafrost Monitoring Network PermaFRANCE.) cially since 2009 and accentuated in 2015 (PERMOS 2016). The past two winters (2015/16 and 2016/17) interrupted this warming trend: a late and thin snow cover resulted in lower permafrost temperatures in debris slopes and on rock glaciers, which were visible down to about 20-m depth (updated from PERMOS 2016; Noetzli et al. 2018, paper to be presented at ), for ex 5th European Conf. Permafrost, EUCOP - ample, in the borehole on Corvatsch-Murtèl. This short-term cooling has also led to a decrease of rock glacier creep velocities relative to the previous years at multiple sites in Switzerland (updated from PER - MOS 2016; Noetzli et al. 2018, paper to be presented ). Rock at 5th European Conf. Permafrost, EUCOP temperatures in shaded f lanks of the highest peaks can be as low as temperatures measured in the Arctic . 2.12. Temperature (°C) measured in permafrost F ig (Fig. 2.11; Noetzli et al. 2016). They closely follow air boreholes along the Qinghai–Xizang Highway on the temperatures and the inf luence of snow is negligible Tibetan Plateau at (a) 10 and (b) 20 m depth. (Source: (Gruber et al. 2004; PERMOS 2007), but only few Cryosphere Research Station on Qinghai–Xizang Plateau, CAS.) and relatively short time series are available. In the | S21 AUGUST 2018 STATE OF THE CLIMATE IN 2017

42 - in most permafrost regions except the Antarctic Pen insula, where ALT has been stable or even decreased since 2009 (Hrbáček et al. 2018). Extremely warm summer conditions in the Arctic in 2016 resulted in extremely high ALT values. They were reinforced in North America in the summer of 2017 resulting in the ALT close to the recorded maximum. In contrast, in Europe and along the Russian Arctic coast, the cold summer of 2017 led to a decrease in ALT relative to the previous year to values around the long-term mean (see Chapter 5 for more details). orthern snow 2) n cover h continental emisphere —D. A. Robinson extent . 2.13. Observed borehole temperatures (°C) in Ant F ig - Annual snow cover extent (SCE) over Northern arctica at 20-m depth (monthly means): WV Wright = 2 Hemisphere (NH) lands averaged 25.8 million km Valley; MP = Marble Point; Oasi in continental Antarc - 2 tica; and Rothera in maritime Antarctica. (Source: more than the 48-year in 2017. This is 0.7 million km Insubria Permafrost Database.) average (mapping extends back to late 1967; however, several early years in the record are incomplete) and ranks as the eighth most extensive cover on reached 0.5°C (site QTB15) between 2005 and 2016, record (Table 2.3 and Fig. 2.14). This is 1.2 million and up to 0.3°C at 20-m depth. Along the latitudinal 2 transect in Victoria Land, continental Antarctica km greater than the 2016 mean extent. Snow extent over both Eurasia and North America, including the [between 77°31'S, Wright Valley (WV in Fig. 2.13) and 74°41'S, Oasi] per - t AB and annual climatological information on Northern Hemisphere 2.3. Monthly le mafrost temperature at and continental snow extent between Nov 1966 and Dec 2017. Included are the 20-m depth is among numbers of years with data used in the calculations, means, standard deviations, the lowest recorded 2 2017 values and rankings. Areas are in millions of km . 1968, 1969, and 1971 have during the period of 1, 5, and 3 missing months respectively, thus are not included in the annual observation. It contin- calculations. North America (N. Am.) includes Greenland. Ranks are from most ued to increase despite extensive (1) to least (ranges from 48 to 52 depending on the month). stable air tempera - 2017 Std. Mean Eurasia N. Am. tures since 1960, and 2017 NH Ye a r s Rank Rank SCE Dev. the temperature rise Rank is more pronounced 6 51 1.6 7 13 49.2 47. 2 Jan at the southern coast (Marble Point; MP in 37 19 22 46.0 Feb 51 46.0 1.8 Fig. 2.13). In contrast, 40.5 27 32 Mar 51 28 1.9 40.1 temperatures mea - sured in the Rothera 21 51 30.5 21 16 31.2 1.7 Apr borehole (67°S) in the 12 51 May 18 12 20.7 2.0 19.3 northern Antarctic Peninsula decreased Jun 50 9.6 2.4 9.3 27 20 40 in the past two years. 36 28 48 Jul 19 4.0 1.2 3.5 This is mainly due to regional cooling of 3.0 Aug 49 14 0.7 2.9 21 28 the atmosphere and 6.2 0.9 5.4 49 15 12 10 Sep the inf luence of snow cover (Guglielmin et 18.4 7 Oct 50 11 2.7 21.2 9 al. 2014). 52 7 36.0 Nov 17 34.1 2.1 9 An increasing trend 43.7 24 Dec 52 31 1.9 43.6 33 in active layer thick - ness (ALT) since the Ann 48 25.1 0.8 25.8 8 11 15 mid-1990s is observed | S22 AUGUST 2018

43 There was an early onset of the 2017/18 snow season across the NH, with NH September cover the tenth most extensive for the month on record. This behavior continued through the remainder of autumn over both EU and NA, with NH SCE ninth - most extensive on record for both October and No vember, each month close to one SD above average. Autumn (September–November) NH SCE was the third highest among the 49 years with complete data, behind 2014 and 1996. As winter began, the pace of the southward snow advance into the middle latitudes slowed, resulting in the 20th least extensive NH snow ig F . 2.14. Twelve-month running anomalies of monthly 2 6 ) over NH lands as a whole km snow cover extent (× 10 cover of the past 52 Decembers. (black), Eurasia (red) and North America (blue) plot - SCE over the contiguous United States was well ted on the 7th month using values from Nov 1966 to - above average in January 2017, the 13th most exten Dec 2017. Anomalies are calculated from NOAA snow sive on record; however, cover decreased considerably maps. Mean hemispheric snow extent is 25.1 million in February and was 12th lowest on record for the 2 km for the full period of record. Monthly means for month, remaining below average throughout spring. the period of record are used for 9 missing months Autumn cover started out on the high side but rank - between 1968 and 1971 in order to create a continuous series of running means. Missing months fall between ings declined through the end of 2017, with December Jun and Oct, no winter months are missing. SCE ranking 20th lowest. Maps depicting daily, weekly, and monthly conditions, daily and monthly anomalies, and Greenland ice sheet, is considered in this analysis. 2 monthly climatologies for the entire period of record Monthly SCE in 2017 ranged from 49.2 million km 2 - may be viewed at the Rutgers Global Snow Lab website in August. SCE is calcu in January to 2.9 million km (http://snowcover.org). Monthly SCE for the NH, EU, lated at the Rutgers Global Snow Lab from daily SCE maps produced by meteorologists at the National Ice NA, contiguous US, Alaska, and Canada are also posted, along with information on how to acquire Center (a U.S. joint NOAA, Navy, and Coast Guard weekly areas and the weekly and monthly gridded - facility), who rely primarily on visible satellite imag products. Section 5: describes SCE as well as snow ery to construct the maps. 2 cover duration and snow water equivalent. January 2017 NH SCE was over 2 million km above average, which exceeds the average by greater 3) a Glaciers M. Pelto and the WGMS network — than one standard deviation (SD) and ranks sixth lpine The World Glacier Monitoring Service (WGMS) highest of the past 51 Januaries. Eurasia (EU) main - record of mass balance and terminus behavior tained above-average SCE in February, ranking 19th highest, while SCE over North America (NA) - (WGMS 2017) provides a global index for alpine gla decreased considerably to 15th lowest. Given the cier behavior. Glacier mass balance is the difference greater land area of EU than NA, this resulted in a between accumulation and ablation, reported here in mm of water equivalence (w.e.) and is a GCOS NH continental ranking of 22nd highest. Conditions across the two continents evened out in March, with headline indicator. Mean annual glacier mass bal - ance in 2016 was −847 mm for the 37 long-term each having the 21st highest SCE in April. Melt over reference glaciers and −761 mm for all 140 monitored both continents was delayed compared to many recent glaciers (Fig. 2.15). Of the reporting reference glaciers, springs, with May SCE the 12th most extensive. Sea - sonally, spring (March–May) NH SCE was the largest only one had a positive mass balance. Preliminary since 2003 and the third most extensive since 1987. data reported to the WGMS in 2017 from Austria, June 2017 SCE over Northern Hemisphere land was Canada, China, France, Italy, Kazakhstan, Norway, close to the long-term average, yet the most extensive Russia, Switzerland, and United States indicate that 2017 will be the 38th consecutive year of negative since 2004, and by far the most extensive since 2007. June cover disappeared rather quickly over NA early annual balances with a mean loss of −1036 mm for in the month, resulting in the 11th smallest SCE on 29 reporting reference glaciers, with three glaciers record, while Eurasia SCE was slightly above average reporting a positive mass balance (http://wgms.ch but the largest since 2003 and third largest since 1997. /latest-glacier-mass-balance-data/, accessed 2 Feb 2018). | S23 AUGUST 2018 STATE OF THE CLIMATE IN 2017

44 pattern of substantial negative balances in the Alps that continue to lead to terminus retreat. In 2016, in Switzerland 94 glaciers were observed: 82 retreated, 7 were stable, and 5 advanced (Huss et al. 2017b). In 2016, Austria observed 84 glaciers: 82 retreated, 1 was stable, and 1 advanced; the average retreat rate was 25 m (Lieb and Kellerer-Pirklbauer 2018). In Norway and Svalbard, terminus f luctuation data from 36 glaciers with ongoing assessment in - dicate that in 2016, 32 retreated, 3 advanced, and 1 was stable. The average terminus change was -12.5 F . 2.15. Global alpine glacier annual mass balance ig m (Kjøllmoen et al. 2017). Mass balance surveys record (mm w.e.) of reference glaciers submitted to with completed results for 2017 are available for nine the WGMS 1980–2017 (see also: http://wgms.ch/latest glaciers; six of the nine had negative mass balances -glacier-mass-balance-data/). with an average loss of -80 mm w.e. In western North The ongoing global glacier retreat affects human America, data for 2017 have been submitted from society by raising sea levels, changing seasonal stream eight reference glaciers in Alaska and Washington in the United States, and British Columbia in Canada. runoff, and increasing geohazards (Huss et al. 2017a). Seven of these eight glaciers reported negative mass Huss and Hock (2018) indicate that approximately balances with an overall mean of -1020 mm. Winter half of 56 glaciated watersheds globally have already - and spring 2017 had above-average snowfall, but abla passed peak glacier runoff. Rounce et al. (2017) iden - tion conditions were above average. In Alaska mass tify the widespread expansion of glacier lakes due to retreat in Nepal from 2000 to 2015, which pose a losses from 2002 to 2014 have been -52 ± 4 gigatons −1 glacier lake outburst f lood hazard. yr , as large as any alpine region in the world (Wahr et al. 2016). Glacier retreat is a ref lection of strongly negative In the high mountains of central Asia, four glaciers mass balances over the last 30 years (Zemp et al. 2015). reported data from China, Kazakhstan, and Nepal. Marzeion et al. (2014) indicate that most of the recent mass loss, during 1991–2010, is due to anthropogenic forcing. The cumulative mass balance loss from 1980 to 2016 is −19.9 m, the equivalent of cutting a 22-m thick slice off the top of the average glacier (http: //wgms.ch/latest-glacier-mass-balance-data, see Figure 2). The trend is remarkably consistent from region to region (WGMS 2017). WGMS mass balance based on 41 reference glaciers with a minimum of 30 years of record is not appreciably different from that of all glaciers at -19.1 m. The decadal mean annual mass balance was -228 mm in the 1980s, -443 mm in the 1990s, -676 mm for the 2000s, and -896 mm for 2010–17 (WGMS 2017). The declining mass balance trend during a period of retreat indicates alpine glaciers are not approaching equilibrium, and retreat will continue to be the dominant terminus response. Exceptional glacier melt was noted across the European Alps in 2017, along with high snowlines (Fig. 2.16), and contributed to large negative mass balances of glaciers on this continent (Swiss Acad - emy of Sciences 2017). In the European Alps, annual mass balance has been reported for nine reference F ig . 2.16. Landsat image from 19 Aug 2017 illustrating glaciers from Austria, France, Italy, and Switzerland. the snowline on Mont Blanc glaciers with one month All had negative annual balances exceeding -1000 left in the melt season (M = Mer de Glace; A = Argen - tière; S Tr i e nt) . = Saleina; L Le Tour; T = = m with a mean of -1664 mm. This continues the | S24 AUGUST 2018

45 All four were negative, with a mean of −674 mm. weak La Niña conditions both at the beginning and This is a continuation of regional mass losses, such as end of the year, whereas 1999 and 2011 each had a strong La Niña present. Over ocean, the moisture reported by King et al. (2017) who found for 2000–15 levels at the surface over the last ~3 years have been the mean annual mass balance of 32 glaciers in the higher than at any other time during the record ac Mount Everest region was −520 ± 220 mm. - cording to the reanalyses. There are currently no in The mass balance of the Arctic glaciers reported situ-only datasets for comparison beyond 2015, but in the WGMS is described in Section 5f. this feature is consistent with high global sea surface d. Hydrological cycle temperatures (Section 2b1) and total column water vapor (Section 2d2). —K. Willett, D. Berry, M. Bosilovich, humidity 1) s urface Despite high surface moisture levels (specific and A. Simmons humidity), in terms of relative humidity (RH), 2017 was a humid year over land and ocean in q terms of surface specific humidity ( the atmosphere remained drier than average over ; Figs. 2.17a–d). land and near average over oceans (Figs. 2.17e–h). Over land it was comparable with the El Niño–driven peak in 2010 but lower than those of 1998 and 2016. ERA-Interim and JRA-55 reanalyses show low RH anomaly values, comparable with the lowest years on Interestingly, compared with other post–El Niño years of 1999 and 2011, the decline from the El record. The HadISDH in situ RH product is similar Niño–driven peak was much smaller. However, interannually to ERA-Interim and JRA-55, but more moderate. MERRA-2 is similar interannually but 2017 saw generally neutral ENSO conditions with with significant deviations that are thought to be linked to vari - ability in precipitation forcing (Reichle and Liu 2014; Willett et al. 2016). Month-to-month, ERA-In - terim and HadISDH also track similarly. Variability was low over ocean during 2017, but over q land both RH and declined throughout the year. December had the driest monthly mean anomaly with respect to both variables. Global average HadISDH is consistently higher than ERA- - Interim for both variables. Had ISDH has gaps over the par - ticularly dry regions of South America, Africa, and Australia and also over Antarctica where dry anomalies are widespread . 2.17. Global average surface humidity annual anomalies (1979–2003 base ig F in ERA-Interim (Plate 2.1 and period). For the in situ datasets 2-m surface humidity is used over land and Online Fig. S2.12). This is a large ~10-m over the oceans. For the reanalysis 2-m humidity is used over the source of uncertainty for in situ whole globe. For ERA-Interim ocean series-only points over open sea are selected and background forecast values are used as opposed to analysis products and may explain some values because of unreliable use of ship data in producing the analysis. All of the difference. Indeed, when data have been adjusted to have a mean of zero over the common period spatially matched, ERA-Interim 1980–2003 to allow direct comparison, with HOAPS given a zero mean and HadISDH are more similar over the 1988–2003 period. Additional dotted lines are plotted for ERA- (Figs. 2.17 a and e dotted). The Interim and MERRA-2 reanalyses where they have been spatially matched data sparse regions and regions to HadISDH for comparison. [Sources: HadISDH (Willett et al. 2013, 2014); of poor data quality are also HadCRUH (Willett et al. 2008); Dai (Dai 2006); HadCRUHext (Simmons where reanalyses differ most et al. 2010); NOCSv2.0 (Berry and Kent 2009, 2011); HOAPS (Fennig et al. 2012) and reanalyses as in Fig. 2.1]. (Plate 2.1b and Online Figs. | S25 AUGUST 2018 STATE OF THE CLIMATE IN 2017

46 S2.10 and S2.11) due to model differences and how/ a comprehensive suite of observations with sufficient which observational datasets are used. numbers that are long term and of climate quality Spatially, ERA-Interim and HadISDH are similar. (Thorne et al. 2018). The annual average q anomaly patterns for 2017 C. Mears, S. P. Ho, 2) t otal were predominantly moist but more muted than in water vapor — column 2016 (Plate 2.1a; Online Figs. S2.10 and S2.11). The J. Wang, and L. Peng strongly moist anomalies over India to China and As Earth’s surface and the lower troposphere south to Southeast Asia were still prominent but to a warm, the total column water vapor (TCWV) is expected to increase under the assumption of near- lesser degree. Dry anomalies were more widespread and quite zonal; there were bands of dry anomalies constant relative humidity, and in turn amplify around 30°–60°N and 0°–30°S. Dry anomalies over the initial warming through positive water vapor feedback. Thus, measurements of TCWV provide eastern Brazil, South Africa, and Iran/Afghanistan/ - an important check to estimates of temperature in Pakistan persisted from 2016. Additional dry anoma - crease in addition to the role of changing TCWV in lies developed over eastern Australia and Spain dur - ing 2017, stronger in ERA-Interim than HadISDH. the global hydrological cycle. In 2017, total column water vapor (TCWV) retreated from record levels in The dry anomaly over East Africa, a data sparse and 2016 but remained above the 1981–2010 climatologi - therefore uncertain region, was stronger than 2016 in cal average in most locations. Estimates are available ERA-Interim but not MERRA-2. Spatial patterns of RH were predominantly nega - from satellite-borne microwave radiometers over ocean (Mears et al., 2018), COSMIC GPS-RO (Global tive (Plate 2.1b; Online Fig. S2.12) and broadly similar to both 2016 and the long-term drying trend patterns. Positioning System–Radio Occultation) over land and Humid anomalies were apparent over southern Africa ocean (Ho et al. 2010a,b; Teng et al. 2013; Huang et around Botswana and Zimbabwe in 2017 that were al. 2013), and ground-based GNSS (Global Naviga - tion Satellite System) stations (Wang et al. 2007) dry in 2016. All regions showing dry q anomalies in over land. An anomaly map for 2017 (Plate 2.1r) was 2017 had spatially more extensive corresponding dry RH anomalies over land and ocean. This shows the importance of looking at both variables in tandem. While a region may be moister than average it could be relatively drier if the regional temperature anomaly is particularly high. The combination of moisture, closeness to saturation, and temperature can lead to - different societal impacts in terms of water availabil ity for people and plants, f looding, and heat stress. There is currently no in situ-only marine product (Willett et al. 2017). The decline in spatial coverage and data quality has made it difficult to continue or develop new in situ marine humidity monitoring products and resulted in the use of background fore - cast fields instead of analysis fields for ERA-Interim marine humidity in this section. Although satellite products can provide measures of water vapor, and total column water vapor (TCWV) has good interan - nual agreement with global surface , these are not q directly comparable with in situ observations and can be derived from q records are shorter. Surface - brightness temperature based on empirical relation ships with in situ data. Several datasets were com - F . 2.18. Global mean total column water vapor an - ig pared alongside the in situ-only NOCSv2.0 marine q nual anomalies (mm) for (a),(b) ocean only and (c),(d) product (Berry and Kent 2009, 2011; Prytherch et al. land only for observations and reanalysis (see Fig. 2.1 2015) and to reanalyses (Schröder et al. 2018). Consid - for reanalysis references) averaged over 60°N–60°S. erable differences were found. Over land, derivations Shorter time series have been adjusted so that there are complicated by the diverse surface properties. is zero mean difference relative to the mean of the The ability to monitor the climate fully depends on three reanalyses over the 2006–14 period. | S26 AUGUST 2018

47 made by combining data from satellite radiometers the tropics. Previous strong El Niño events (1983/84 over ocean and COSMIC GPS-RO over land. Much of and 1997/98) showed pronounced drying events in the the globe showed small wet anomalies, except for dry years following the El Niño events, which were not seen anomalies in the central tropical Pacific Ocean and after the 2015/16 event. the southeastern tropical Indian Ocean, and a large wet anomaly in the western Pacific warm pool and pper u 3) —V. O. John, L. Shi, humidity tropospheric - E.-S. Chung, R. P. Allan, S. A. Buehler, and B. J. Soden over the maritime continent, consistent with the pres ence of La Niña conditions. The patterns in TCWV In the atmosphere as a whole, water vapor is the over the ocean are confirmed by COSMIC ocean principal greenhouse gas (Held and Soden 2000). Despite water vapor in the upper troposphere being measurements and by output from the MERRA-2, ERA-Interim, and JRA-55 reanalyses (not shown). insignificant by total mass when compared to the total column, it nevertheless contributes a major part to the Over land, the patterns from COSMIC are in general feedbacks present in the climate system. Thus, up agreement with the reanalysis output. - Over the ocean, the TCWV anomaly time series per tropospheric water vapor is responsible for most - (Figs. 2.18a,b) from reanalysis and microwave radi of the tropospheric radiative cooling (Manabe and Möller 1961), and the radiative effect of water vapor ometers show maxima in 1982/83, 1987/88, 1997/98, is proportional to relative changes in the amount 2009/10, and 2015/16 associated with El Niño events. The 2015/16, anomaly is the largest recorded in all of water vapor in the upper troposphere (John and - Soden 2007). datasets. The radiometer data show a discernible in creasing trend, while the different reanalysis products Following John et al. (2017), upper tropospheric (relative) humidity (UTH) is monitored on a global show a wide range of long-term trends over the entire - scale by two independent global satellite UTH datas period but agree well with the radiometer data after the ets: (1) the High Resolution Infrared Sounder (HIRS) mid-1990s. The COSMIC data show the same general features as both the radiometer and reanalysis data dataset (Shi and Bates 2011) and (2) the microwave- after COSMIC began in 2007. After the 2015/16 peak, - based UTH dataset (Chung et al. 2013). In these da all datasets show a return to drier conditions due to tasets, UTH represents a Jacobian weighted average the onset of La Niña but remained wetter than the of relative humidity in a broad layer which is roughly between 500 and 200 hPa but varies depending upon 1981–2010 normal for almost all latitudes. atmospheric humidity profile. Both datasets have Over land, average anomalies from the ground- been constructed through careful bias corrections based GNSS stations are used in place of the satellite and intersatellite calibration. As the microwave-based radiometer measurements (Figs. 2.18c,d). The various reanalysis products, COSMIC, and GNSS are in good UTH dataset only begins in 1999, anomalies are com - agreement. A land-and-ocean Hovmöller plot derived puted relative to the 2001–10 base period. from JRA-55 (Fig. 2.19) indicates a long-term increase Figure 2.20 shows the area-weighted mean de - in TCWV at all latitudes, with less variability outside seasonalized anomaly time series of UTH for 60°N– 60°S using two observational datasets as described later in this section: one from HIRS and the other . 2.20. Time series of UTH anomalies (% RH; 2001– ig F 10 base period) using HIRS (black) and microwave . 2.19. Hovmöller plot of TCWV anomalies (mm; ig F (blue) datasets. Time series are smoothed to remove base period 1981–2010) including both land and ocean variability on time scales shorter than three months. derived from the JRA-55 reanalysis | S27 AUGUST 2018 STATE OF THE CLIMATE IN 2017

48 Climatology Centre (GPCC; Becker et al. 2013) and the blended gauge–satellite product from the Global Precipitation Climatology Project (GPCP; Adler et al. 2003), both depict 2017 as about 15 mm above average. The operational version of the gauge-based Global Historical Climatology Network (GHCN; Peterson and Vose 1996) dataset is somewhat wetter, with an anomaly of about 40 mm, while a new experimental version of GHCN (with five times as many stations) has an anomaly of about 80 mm (implying 2017 was the wettest year in the historical record). Notably, when the blended gauge–satellite GPCP product is F ig . 2.21. Annual average UTH anomalies (% RH, adjusted using a new gauge analysis for 2014–present, 2001–10 base period) for 2017 based on the “clear-sky” the anomaly for 2017 increases to about 30 mm, which HIRS UTH dataset. would be the wettest year in the satellite era. Large from microwave humidity sounder measurements. areas with above-normal precipitation in 2017 (Plate The anomalies are close to zero in 2017 and are fairly 2.1i) include northwestern North America, northern Eurasia, interior sub-Saharan Africa, southeastern similar to the previous year, even though 2016 began with an El Niño whereas there were neutral or weak - Asia, the Maritime Continent, and western Austra La Niña conditions in 2017. To maintain the same lia. Areas with below-normal precipitation include - southern Alaska, central Canada, southeastern Brazil, relative humidity, warmer air has to contain more wa western Europe, eastern Africa, northern India, the ter vapor (have a higher specific humidity). Therefore, the presence of a near-zero decadal trend in the UTH Korean peninsula, and eastern Australia (Plate 2.1i). Relative to 2016, aridity was much less pronounced requires an increase in absolute (specific) humidity in many areas, particularly Central America, South in step with the warming upper troposphere (Section America, and southern Africa. 2b4) and hence is consistent with a positive water According to GPCP, precipitation over the global vapor feedback (Chung et al. 2016). Good agreement between the two independent datasets despite their ocean surface in 2017 was near the long-term aver - differences in sampling—microwave data having an age for the satellite era. In the tropics, the annual almost all-sky sampling whereas HIRS data sample anomaly pattern (Plate 2.1i) ref lects much wetter- than-normal conditions stretching from the eastern mainly clear-sky scenes—provides confidence in the observed long-term behavior of UTH (John et al. Indian Ocean across the Maritime Continent to the 2011). The higher short-term variability in the HIRS western equatorial Pacific Ocean, with much drier- than-normal conditions extending eastward across time series arises from the sampling issues discussed by John et al. (2011). The spatial variation of annual the rest of the equatorial Pacific. Indeed, the seasonal - average UTH anomalies for 2017 (Plate 2.1q for mi anomaly patterns during 2017 (not shown) indicate crowave data and Fig. 2.21 for HIRS data) shows dry that similar features in the Pacific Ocean/Maritime anomalies over the central Pacific and moist anoma - Continent area existed in varying strengths during lies over the maritime continent, which ref lect neutral ENSO conditions during the year; however, the moist anomaly seen in the eastern Pacific is typical for El Niño conditions. The dry anomaly over the Indian subcontinent is an indication of the below-normal monsoon rainfall in 2017 (Sections 2d9, 7g4). —R. S. Vose, R. Adler, A. Becker, and X. Yin recipitation 4) p Precipitation over global land areas in 2017 was clearly above the long-term average (Fig. 2.22). All available datasets are consistent on this point, but there is a wide range of estimates across the analyses . 2.22. Globally averaged precipitation anomalies F ig - (ranging from 15 to 80 mm above average). The datas - 1 (mm yr ) over land areas relative to the 1961–90 base ets with the most complete global coverage, that is, the period (except GPCP v2.3, which is 1981–2000). gauge-based product from the Global Precipitation | S28 AUGUST 2018

49 SIDEBAR 2 .1: —M. R. TYE, LAND SURFACE PRECIPITATION EXTREMES S. BLENKINSOP, M. DONAT, I. DURRE, AND M. ZIESE - 2017 was a mixed year in terms of extreme precipita making only indices such as Rx1day or R95P (Table SB2.1) tion, with remarkable tropical and post-tropical cyclone meaningful for global comparisons. During 2017 some of precipitation at one end of the scale and extended - these regions also experienced anomalously high precipi - droughts at the other (see Sections 2d4, 2d9, respec tation events that resulted in significant impacts, such as tively). The range of events demonstrates that extreme Rx1day (Plate 2.1k, Online Fig. S2.17) for 31 March–1 April precipitation is not evenly distributed across the globe in Colombia (130 mm in Mocoa). (Herold et al. 2015), while anthropogenic climate change - While boreal spring 2017 saw high seasonal precipita has likely increased their probability of occurrence (e.g., tion totals across all continents (Section 2d4), individual Risser and Wehner 2017). Annual precipitation totals indices do not reflect the full picture of extremity. For were above the 90% percentile in coastal east Asia, instance, although atmospheric rivers (e.g., Dettinger et al. western and northern Australia, northeastern Europe, 2011) were very active along the U.S. West Coast early in parts of northern North America, Central America, and the year, Rx1day and Rx5day rank low for that region with southeastern South America (Plate 2.1j; see also Section respect to climatology (Plate 2.1k). The year as a whole 2d4). The year was also notable for the large proportion is notable for the moderate extreme indices. That is, the of moderate to heavy extreme precipitation (i.e., days anomalies in total precipitation (PRCPTOT) appear to be with accumulations � 90th and 95th percentiles; Zhang attributable more to anomalies in R10mm and/or R20mm et al. 2011) with respect to previous years. than, say, R95P (Plate 2.1j, Fig. SB2.1, Online Fig. S2.18). Robust and reliable global datasets for extreme Munich Re (2018) summarized 2017 global loss events precipitation that include data throughout 2017 are lim - within four hazard categories. From the meteorological ited, particularly for sub-monthly resolutions. Here we - and hydrological events hazard categories, the pre make use of the Global Historical Climatology Network cipitation induced insured losses from Severe Tropical Daily (GHCND; Menne et al. 2012), GHCND-based Cyclone Debbie (27 March–6 April) in eastern Australia GHCNDEX (Donat et al. 2013a) and GPCC-FirstGuess and New Zealand, and Hurricane Harvey (25 August–1 Daily (Schamm et al. 2013) gridded extremes to calculate September) in Texas and Louisiana are unprecedented. the precipitation indices summarized in Table SB2.1. How - Hurricane Harvey set numerous daily rainfall records ever, near-realtime updates including 2017 are primarily at NOAA weather stations throughout the area, with available from North America, Europe, Australia, and Rx1day exceeding 600mm and Rx5day at almost double parts of Asia for GHCNDEX. previous records (Fig. SB2.2; see online supplemental The highest daily precipitation totals are typically found information; see Sidebar 4.3 for more details on Harvey). as orographically enhanced rainfall in tropical regions, - Severe Tropical Cyclone Debbie strengthened to a Saf t AB SB2.1. Extreme precipitation indices from the Expert Team on Climate Change Detection and le Indices (ETCCDI) . Index Name Definition Unit PRCPTOT Total annual precipitation mm Annual wet day (R≥1 mm) precipitation total Max 1 day precipitation amount Annual maximum 1 day precipitation mm R x1day Annual maximum consecutive 5 day Max 5 day precipitation amount mm Rx5day precipitation R10mm Number of heavy precipitation days Annual count when precipitation ≥10 mm days R20mm Annual count when precipitation ≥20 mm days Number of very heavy precipitation days Annual total precipitation from days >95th Very wet days R95P mm percentile of wet days | S29 AUGUST 2018 STATE OF THE CLIMATE IN 2017

50 CONT. SIDEBAR 2 .1: LAND SURFACE PRECIPITATION EXTREMES —M. R. TYE, S. BLENKINSOP, M. DONAT, I. DURRE, AND M. ZIESE at least three times at several locations since 1918 (Online fir–Simpson category 3 storm before making landfall over Fig. S2.19). Other events such as those in South Asia the North Queensland coast. While Debbie is considered (monsoon rains June–October), Peru (wet season floods to be one of the costliest and most deadly cyclones to January–March), and Sierra Leone (landslides 14 August) - affect Australia (Insurance Council Australia 2017), as were notable for their longevity and/or high human cost. sociated Rx1day and Rx5day totals have been exceeded Further information is available at http://floodlist. com/?s=2017&submit=. - Many of the events witnessed in 2017 origi nated from stationary mesoscale convective systems (MCS). MCSs are organized systems of thunderstorms, larger than individual storms but smaller than extratropical cyclones. They can last for over 12 hours and can rival tropi - cal cyclones for impacts due to their intensity and repetition leading to soil saturation, and they contribute up to $20 billion (U.S. dollars) economic losses each year (Munich Re 2016). Across the North American Midwest, where the conditions are often suitable for generating these events, it is notable that the pattern of extreme deviations from climatology (negative and positive) for R10mm and R20mm are similar to PRCPTOT. That is, the greatest contributions to the annual total came from more moderate extremes. Many of the flood inducing extremes in 2017 (Brakenridge 2018) appear to be derived from these moderate extremes, suggesting that more attention should be paid to “nuisance events” (i.e., unusual but not so rare that they are unknown within the record; Schroeer and Tye 2018, manuscript submitted to J. Flood Risk ). It should be noted that the available Manage. data, such as GHCNDEX, are often too coarse to resolve MCSs and may miss these extremes. Limited availability of in situ high-resolution rainfall observations has confounded long-term assessments of changes in sub-daily extreme precipitation (significant in generating flash floods; Westra et al. 2014). Historical analyses of change have typically been conducted on regional scales using a variety of methods (Online Fig. S2.20) making it difficult to assess the state of the climate and to place notable events in con - text. However, studies do indicate a tendency towards more intense extremes. A global data collection exercise is underway (Lewis et al. J. Climate ) under 2018, manuscript submitted to . SB2.1. 2017 anomalies (1961–90 base period) from F i g the INTENSE (Intelligent Use of Climate Models GHCNDEX for (a) PRCPTOT (mm), (b) R10mm (days), and for Adaptation to Nonstationary Hydrological (c) R20mm (days). | S30 AUGUST 2018

51 Extremes) project (Blenkinsop et al. 2018, manuscript intensify with temperature according to or exceeding the −1 ) Adv. Sci. Res. submitted to ) and will result in a quality- Clausius–Clapeyron (CC) relation (a rate of ∼ 6%–7% °C (Trenberth et al. 2003; Pall et al. 2007), although changes - controlled sub-daily dataset. This will include the produc in dynamics may lead to regionally higher or lower rates tion of comparable sub-daily extreme indices to those in of intensification. Improved quality and global coverage of Table SB2.1 to enable the monitoring of these events. New sub-daily observations will enable a much needed advance analyses using these data have indicated continental-scale in understanding of how local thermodynamics and large- increases in the intensity of hourly rainfall (Barbero et al. scale circulations interact to generate short-duration 2017; Guerreiro et al. 2018, manuscript submitted to Nat. intense rainfall (Pfahl et al. 2017). ). Observational evidence, coupled with Climate Change that from climate models, suggests that heavy rainfall will ig . SB2.2. (a) Rx5day from GHCNDEX for Hurricane Harvey and (b) ratio of the 2017 value to the F previous maximum in the record. records including PATMOS-x/AVHRR (Pathfinder the entirety of 2017. Large parts of the North Pacific, North Atlantic, South Atlantic, and south Indian Atmospheres Extended/Advanced Very High Resolu - Aqua tion Radiometer; Heidinger et al. 2014), Oceans had notable negative anomalies, whereas MODIS C6 (Moderate Resolution Imaging Spectroradiom parts of the South Pacific convergence zone and the - equatorial Atlantic Ocean were wetter than normal. eter Collection 6; Ackerman et al. 2008), CALIPSO (Cloud-Aerosol Lidar and Infrared Pathfinder Satel A negative anomaly feature across the very southern - lite Observation; Winker et al. 2007), CERES (Clouds tip of the African continent was also apparent, where and the Earth’s Radiant Energy System; Minnis et al. Cape Town, South Africa, experienced its driest year MODIS, MISR (Multi- since 1933. Aqua 2008; Trepte et al. 2010) For an assessment of precipitation extremes in angle Imaging SpectroRadiometer; Di Girolamo et al. 2017, see Sidebar 2.1, and for more detailed discussion 2010), and SatCORPS (satellite cloud and radiative on regional precipitation quantities, see Chapter 7. property retrieval system; Minnis et al. 2016). All of these records show a decrease in cloudiness from 2016 —M. J. Foster, S. A. Ackerman, K. Bedka, 5) c loudiness to 2017 ranging from 0.1% to 0.34%, depending on the dataset. Figure 2.23 shows global cloudiness from L. Di Girolamo, R. A. Frey, A. K. Heidinger, S. Sun-Mack, 1981 to present with additional records: HIRS High C. Phillips, W. P. Menzel, M. Stengel, and G. Zhao Cloud observations are important for monitoring Cloud (High Resolution Infrared Sounder; Wylie et al. 2005; Menzel et al. 2016), CLOUD_CCI (Cloud climate because they modulate energy f low through Climate Change Initiative AVHRR-PM v3.0; Stengel - ref lection of incoming solar radiation and absorp et al. 2017), CLARA-A2 (cloud, albedo and radiation tion of outgoing terrestrial radiation, and they affect Aqua dataset; Karlsson et al. 2017), and PATMOS-x/ global water distribution through storage and pre - cipitation of atmospheric water. Global cloudiness MODIS that do not currently extend through 2017. - While there is interannual and inter-record variabil in 2017 decreased incrementally (~0.2%) from that of 2016. This analysis is based on several satellite cloud ity in the early part of the record, there is an overall | S31 AUGUST 2018 STATE OF THE CLIMATE IN 2017

52 ating an aliasing effect. Therefore, when generating a cloud data product, the selection of which satellite records to include is significant. SatCORPS, CLARA- A2, and CLOUD_CCI are derived from afternoon satellites, while PATMOS-x/AVHRR uses afternoon and morning satellites. PATMOS-x, SatCORPS, and CLARA-A2 have a diurnal correction applied (Foster and Heidinger 2013). This correction usually takes the form of a cloudiness adjustment to a single local overpass time based on a linear regression. Several international collaborative efforts exist with the goal of better characterizing these differences and ad - dressing some of these issues, including the Global Energy Water Cycle Experiment (GEWEX) Cloud Climatology Assessment (Stubenrauch et al. 2013), the International Clouds Working Group (ICWG; . 2.23. (a) Annual global cloudiness anomalies (%) ig F for 1981–2017, defined as the annual value minus the formerly the EUMETSAT Cloud Retrieval Evaluation - mean, derived between 2003 and 2015, a period com Workshops; Wu et al. 2017), the WMO Sustained and mon to the satellite records excluding CALIPSO , where Coordinated Processing of Environmental Satellite the entire record was used instead. (b) Annual actual data for Climate Monitoring (SCOPE-CM; Kearns global cloudiness (%). The datasets include PATMOS- and Doutriaux-Boucher 2015) AVHRR Climate x/AVHRR, HIRS High Cloud, MISR, Aqua MODIS C6, Initiative, and the WMO Global Space-based Inter- MODIS, SatCORPS, CLARA- CALIPSO , CERES Aqua Calibration System (GSICS). Aqua A 2 , PATM O S -x / MODIS, and CLOUD_CCI. - There were a few noteworthy cloudiness anoma lies (those found to be significant at the 5% level) tendency for convergence after 2000. Much of the convergence can be explained by the use of a com - in 2017 relative to the PATMOS-x/AVHRR base mon baseline of 2003–15, though it does not explain period of 1981–2010. Almost all of these anomalies interannual variability. Online Figure S2.13 plots the were less cloudy than average with two exceptions of cloudier-than-average areas over the Arctic Ocean. records that extend back before 2000 and removes the common baseline, which results in the spread among Global cloudiness patterns frequently correspond the records to be similar throughout. Figure 2.23b with large-scale circulation patterns. SST and low- - level wind anomalies between the central equatorial shows absolute cloudiness and the overall interan Pacific and Indonesia characteristic of ENSO drive nual stability of these records. It also shows there is convection, which, in turn, drives global cloudiness no consensus on global cloudiness trends. We should note the HIRS record is noticeably lower because it focuses on detecting high cloud. It is included here because comparison with anomalies is still valuable, and it is the only non-AVHRR record we have that extends back into the 80s and 90s. Although global-scale events such as ENSO and volcanic eruptions may be responsible for some early-record interannual variability, it is likely that much of the interannual and most of the inter-record variability relates to the combinations of satellites and sensors used in the records. Four of the records that extend back into the 1980s—PATMOS-x/AVHRR, SatCORPS, CLARA-A2, and CLOUD_CCI—are derived from the AVHRR sensor f lown on NOAA POES. The morning satellites f lown in the 1980s and 1990s lack a second infrared channel and have a great - F ig . 2.24. Annual global cloudiness anomalies (%; er tendency to drift from their original orbit thereby relative to 1981–2010) from the PATMOS-x/AVHRR record calculated using the same method as Fig. 2.23 shifting the local overpass time and potentially cre - but zonally for each degree latitude. | S32 AUGUST 2018

53 distribution. El Niño events often correspond with for water resources required by humans and their cloudier conditions over the central equatorial and activities and industries. Sixty years (1958–2017) of global runoff and river southeastern Pacific, while La Niña events correspond discharge were estimated by off-line land surface with less cloudy conditions. This can be seen in Fig. 2.24 where cloudiness anomalies are consistent with simulations on the ensemble land surface estimator phases of ENSO in the PATMOS-x/AVHRR record. In (ELSE; Kim et al. 2009). The simulation configuration 2017 the ENSO index was largely neutral, beginning remains the same as in the previous report (e.g., Kim and ending the year with weak La Niña conditions. 2017), and atmospheric boundary conditions were - Seasonal cloudiness ref lects this evolution (Online extended by combining the Japanese global atmo spheric reanalysis (JRA-55; Kobayashi et al. 2015) and Fig. S2.14). Due to this pattern, Plate 2.1p shows cloudiness anomaly patterns between the western the Global Precipitation Climatology Centre (GPCC) Pacific and Indonesia generally consistent with weak Monitoring Product version 5 (Schneider et al. 2015). In 2017, the global distributions of runoff (Plate La Niña conditions but lacking significance at the 2.1l) and discharge (Plate 2.1m) anomalies show that 5% level, with the exception of small areas off the large areas of South America, Southeast Asia, eastern west coasts of Mexico and Chile. The Indian Ocean Europe, and western and eastern Siberia were under dipole (IOD) is an interannual weather pattern that significantly wet conditions. In contrast, Africa, affects the tropical Indian Ocean. 2017 saw sporadic central Siberia, India, the eastern United States, and negative phases of the IOD at the beginning and end eastern Europe including the Mediterranean were of the year, which typically correspond with cooler under drier conditions compared to their normal sea surface temperatures in the western Indian Ocean climate. Among these, the African, European, and that likely contributed to below-average cloudiness seen in the northern and southwestern parts of the Siberian regions tended to experience a similar state Indian Ocean. Continental below-average cloudiness as the previous year. Long-term variability of global frequently corresponds with warm and dry condi - runoff is shown with the El Niño–Southern Oscilla - tion (ENSO) and the Pacific decadal oscillation (PDO; tions as experienced in Alaska, western Europe, and large portions of Russia and China. Mantua et al. 1997) in Fig. 2.25. It has been found that the La Niña phase of ENSO and a neutral phase and runoff — H. Kim of PDO result in global runoff that is weakly wetter 6) r iver dischar Ge than the long-term average. The ENSO and PDO in After evapotranspiration, induced by the available - energy at the land surface and from moisture update dices explain approximately 50% of the variability of - from the root zone by the photosynthesis of vegeta the global runoff (Kim 2017). After a strong positive tion, the remaining precipitated water is transported - phase of the 2015/16 ENSO, the weak La Niña condi by gravity. The water eventually forms narrow and tions in 2017 and the relatively weak positive phase of meandering rivers, transporting it to the oceans. the PDO led to slightly wetter conditions. Figure 2.26 indicates a monthly time series Freshwater in the channel network is the first source deviation (i.e., excess or deficiency) runoff from the long-term mean of season - al variations globally and on each continent. South America shows the typi - cal seasonal variation of a wet year that has excessive runoff in the earlier season (i.e., February–April) com - pared to the wet season in dry years (i.e., May–July). The Amazon River is wetter than normal except in a few . 2.25. Interannual variability of ONI (Oceanic Niño Index, lower), PDO ig F sub-basins such as the Rio (upper), and global runoff (middle; mm; thick line is 12-month moving aver - Madeira, the Rio Tocantins, age). ONI and PDO are shaded red (positive phase) or blue (negative phase). and the Rio Araguaia. The Shading above and below the zero-line of global runoff is proportional to PDO and ONI, respectively. Rio Parana and the Rio Sao | S33 AUGUST 2018 STATE OF THE CLIMATE IN 2017

54 and roundwater 7) G terrestrial water stora Ge — M. Rodell, D. N. Wiese, B. Li, and J. S. Famiglietti Precipitation that falls on the land and does not immediately evaporate or run off becomes terrestrial water storage (TWS; the sum of groundwater, soil - moisture, surface water, snow, and ice). Ground water and total TWS exhibit larger variations on multiannual timescales than the near-surface TWS components (Li et al. 2015). Both are difficult to monitor using in situ observations, but from 2002 to 2017 the Gravity Recovery and Climate Experiment (GRACE; Tapley et al. 2004) satellite mission mapped TWS variations on a monthly basis at regional scales, worldwide. During the last few years of the mission, on-board battery issues caused frequent, multimonth measurement gaps, and no TWS data are available past June 2017. To create the 2016–2017 difference map (Plate 2.1h) output from a GRACE data assimi - lating land surface model (Li et al. 2018, manuscript ) was used. submitted to Water Resour. Res. Changes in TWS between 2016 and 2017, plotted as equivalent heights of water in Plate 2.1h, integrate the effects of other hydroclimatic conditions (see Plates 2.1i, n, o, p, t, and v). All continents experienced a F . 2.26. Interannual variability of global and con ig - somewhat even mix of TWS increases and decreases, 1 − tinental runoff (mm yr - and ) for 1958–2017. The x with many reversals. The Amazon basin recovered -axes correspond to annual and seasonal variations, y from huge, widespread water losses in 2016, with large respectively. Europe and South America refer to the upper scale of the color bar, and the others refer to - gains in the eastern part of the basin. However, south the lower scale. ern Brazil endured significant TWS reduction. Across the Atlantic the reverse scenario occurred—with - Francisco maintained the same wet and dry condi large-scale, deep drying in central and eastern Africa tions as in the previous year, respectively. In Asia, the and wetting to the south. Southern Europe suffered dry condition of high latitude regions in 2016 was serious drought-related water losses, most notably in alleviated slightly. The state of the Ob and Kolyma Portugal, while northern Europe was normal to wet. Rivers (Siberia) shifted from dry to wet. However, the Northwestern Australia regained water lost in the Yenisey River still remained dry and the Lena River took on dry conditions. Over mid- and low latitudes in Asia, many regions were wetter than normal (e.g., - the Yangtze, Huang He, Chao Phraya, Mekong Riv ers), while Amur and Ganges–Brahmaputra were drier than normal. Europe showed a considerable deviation from the climatological seasonality, with considerably drier conditions during the early half of the year; the phase shifted radically into a wet-year condition beginning in June. North America was in a weak dry condition, and most of the rivers in the region, including the Mississippi (U.S.) and Macken - zie (Canada), were facing a water deficit. Africa has been experiencing a persistent dry condition since the 1980s, and Australia has had near-neutral to dry conditions after a historic wet year in 2011. F ig . 2.27. Zonal mean terrestrial water storage anoma - lies (cm, equivalent height of water, 2005–10 base period) from GRACE. White areas indicate months when data were unavailable. | S34 AUGUST 2018

55 way (Dorigo et al. 2017b; Gruber et al. 2017; Liu et al. 2012). The latest dataset version (COMBINED v04.3) merges 11 different sensors between late 1978 and December 2017 into a single harmonized long-term dataset with reduced uncertainties and fewer data gaps compared to the single sensor products. The dataset has been validated against a large number of land surface models and in situ datasets used for a wide range of applications (Dorigo et al. 2017b). Based - on the ESA CCI SM the yearly and monthly anoma lies are computed here with respect to a 1991–2016 climatology. F ig . 2.28. Global average terrestrial water storage anomalies from GRACE (cm, equivalent height of For several regions, spatial soil moisture anomaly water, 2005–10 base period). patterns in 2017 (Plate 2.1g) were remarkably drier previous year, while conditions in the southeast were or wetter than normal. While after several dry years in a row (Blunden and Arndt 2017) soil moisture generally dry. TWS changes in North America were - conditions in the northeast of South Africa were also mixed. The heavy winter rains that led to f lood ing and mudslides in central and southern California partly alleviated in 2017, drought conditions in the aided in its recovery from long-term drought. Much region around Cape Town intensified in the course of the eastern and mountain regions of the U.S. also of the year (Online Fig. S2.15; see Section 7e4). Dry gained TWS, while Canada and Mexico were gener soil moisture conditions already observed in 2016 in - the Greater Horn of Africa (Blunden and Arndt 2017) ally dry, including drought in British Columbia that - persisted into 2017 and reportedly led to a displace contributed to its most extensive wildfire season on ment of more than 1 million people, according to a record (see Section 7b1). Central and southern Asia report from the World Meteorological Organization. exhibited patchy drying, while heavy rains in the Indochina peninsula increased TWS dramatically. On the other hand, soil moisture contents were higher than normal for most other parts of southern Africa, Eastern Siberia also was wetter than normal. While - particularly during the first half of the year, and con GRACE has measured significant reductions in TWS in Antarctica, Greenland, southern coastal Alaska, tributed to severe f looding, for example, in Botswana. Very dry soils were also observed in Morocco and and Patagonia (the latter two are apparent at 60°N and 46°–55°S, respectively, in Fig. 2.27) due to ongo - southern Europe. Italy suffered particularly severe rainfall deficits and had its driest January–September ing ice sheet and glacier ablation, these processes are period on record. not properly simulated by the model and the regions While soil moisture conditions in most parts of must be ignored in Plate 2.1h. Figures 2.27 and 2.28 plot zonal mean and global - Brazil were around average, some parts of northeast mean deseasonalized monthly TWS anomalies from ern Brazil showed strong anomalous negative soil GRACE (excluding Greenland and Antarctica). Re moisture conditions for the sixth consecutive year - State of the Climate duced dryness in the southern tropics (Fig. 2.27) is reports, e.g., Dorigo [see previous et al. (2017a)]. Wet conditions were observed for associated with the TWS increases in the Amazon, southern Africa, and northwestern Australia. While southern South America and the west coast of Peru, only five months of GRACE data are plotted in Fig. which strongly contrasts with the anomalously dry conditions that were observed in this region in 2016 2.28, GRACE data assimilation output (not shown) (Dorigo et al. 2017a). Also, most of the southern and indicate that recovery from the January 2016 global TWS minimum continued in 2017, owing largely to eastern United States were much wetter than normal. the increases in the Amazon. In particular, August was very wet (Online Fig. S2.15) with Hurricane Harvey making landfall in southern Texas. On the other hand, the Canadian Prairies and oil s —W. A. Dorigo, T. Scanlon, A. Gruber, moisture 8) adjacent northern border areas of the United States R. van der Schalie, C. Reimer, S. Hahn, C. Paulik, W. Wagner, were anomalously dry, mainly during the summer and R. A. M. de Jeu months (Online Fig. S2.15). The ESA Climate Change Initiative soil moisture In 2017, soils in large parts of Southeast Asia were (ESA CCI SM) product combines observations from much wetter than normal. The monthly anomaly a large number of historical and present-day passive images reveal that this pattern persisted throughout and active microwave instruments in a synergistic | S35 AUGUST 2018 STATE OF THE CLIMATE IN 2017

56 the year (Online Fig. S2.15). A much wetter-than- et al. 2013; Boening et al. 2012; Dorigo et al. 2017b; - Miralles et al. 2014). Although soil moisture condi average start to the year in many parts of western and northern Australia (Online Fig. S2.15) resulted tions in the Southern Hemisphere were on average in net average soil moisture conditions in 2017 that wetter than normal, deviations were far from being were wetter than usual for these areas. At the same as pronounced as in 2000 or 2010/2011 (Figs. 2.29 and 2.30), which were episodes associated with strong La time, most parts of eastern Australia were drier than Niña events. In the Northern Hemisphere, average average, ref lecting precipitation amounts that were well below average (see Section 7h3). soil moisture was close to normal in 2017 (Fig. 2.30). The year 2017 was mostly dominated by a neutral No evident large-scale long-term global soil mois - ture trends can be observed (Fig. 2.30). However, this state of ENSO (see Section 4b). ENSO anomalies are does not exclude the existence of long-term trends at known to potentially cause continent-wide deviations in terrestrial water storages (Bauer-Marschallinger the regional or local scale (An et al. 2016; Rahmani et al. 2016; Wang et al. 2016). However, anomalies and trends in average global soil moisture should be treated with caution, owing to dataset proper - ties changing over time and the inability to observe beneath dense vegetation, mountain areas, or frozen or snow-covered soils (cf. gray regions in Plate 2.1g and Online Fig. S2.15). 9) —T. J. Osborn, J. Barichivich, I. Harris, d rou Ght G. van der Schrier, and P. D. Jones Hydrological drought results from a period of abnormally low precipitation, sometimes exacerbated - by additional evapotranspiration (ET), and its occur rence can be apparent in reduced river discharge, soil moisture, and/or groundwater storage, depending on season and duration of the event. Here, an estimate of drought called the self-calibrating Palmer drought - F ig . 2.29. Time–latitude diagram of surface soil mois 3 3 − ture anomalies (m , base period: 1991–2016). Data m severity index (scPDSI; Wells et al. 2004; van der were masked as missing where retrievals are either Schrier et al. 2013) is presented, using precipitation not possible or of low quality (dense forests, frozen and Penman–Monteith potential ET from an early soil, snow, ice, etc.). (Source: ESA CCI Soil Moisture.) update of the CRU TS 3.26 dataset (I. Harris et al. 2014). Moisture categories are calibrated over the complete 1901–2017 period to ensure that “extreme” - droughts and pluvials relate to events that do not oc ig F . 2.30. Time series of average global surface soil 3 3 − m moisture anomalies for 1991–2017 (m , base pe - riod: 1991–2016). Data were masked as missing where . 2.31. Percentage of global land area (excluding ice F ig retrievals were either not possible or of low quality sheets and deserts) with scPDSI indicating moderate (dense forests, frozen soil, snow, ice, etc.). (Source: 2), severe (< − 3), and extreme (< − 4) drought for (< − ESA CCI Soil Moisture.) each month of 1950–2017. Inset: each month of 2017. | S36 AUGUST 2018

57 cur more frequently than in approximately 2% of the south-central Chile (Garreaud et al. 2017) continued for the eighth consecutive year, though the geographic months. This affects direct comparison with other extent of extreme drought decreased with a slight hydrological cycle variables in Plate 2.1s that use a different base period. increase in winter rainfall. Notably, severe drought in the semiarid northeastern Brazil (Jiménez-Muñoz After a notable peak in the overall area of drought across the globe in the second half of 2015 and all et al. 2016) continued in 2017 without much change - of 2016 (Osborn et al. 2017), drought area declined in intensity and extent (Fig. 2.32). Moderate, or oc sharply by early 2017 (Fig. 2.31) before increasing to casionally severe, drought was present across the above average once more (though still below the 2016 Northern Hemisphere part of the South American area). Extreme drought conditions affected at least continent (Plate 2.1s) though its intensity had eased compared with 2016 in most areas (Fig. 2.32). 3% of global land area in every month of 2017, which Many coastal countries in Africa experienced was matched only by 1984, 1985, and 2016, but the drought in 2017, with the exception of some in East geographical extents of moderate and severe droughts Africa (see Section 7e3). These droughts intensified were not so unusual. The area where scPDSI indicates - compared with 2016 especially in southern Madagas moderate or worse drought began at 24% in January, car and the Western Cape of South Africa, the latter fell below 22% by April, before rising to around 25% contributing to water supply restrictions in Cape in the latter months of 2017. Altogether, three months Town in early 2018 (Le Page 2018). The partial easing had moderate or worse drought affecting more than 25% of the global land area, which has been matched of drought farther north, including in the Zambezi basin, is important given the increasing concentration or exceeded in 34 other years since 1950. The area of hydropower in the region that increases the risk of of severe plus extreme droughts exceeded 10% for concurrent drought-related disruption to electricity ten months during 2016, which has been matched production (Conway et al. 2017). Conditions were or exceeded in 12 other years since 1950. The 2017 values should be interpreted cautiously because they drier in 2017 than in 2016 in a band across Eurasia may be modified by additional observations that will around 45°N (Fig. 2.32). This exacerbated drought in western and southern Europe, resulting in many become available in due course. Drought area is just one of several ways to measure drought conditions; impacts, including reduced agricultural yields and hydroelectric power production in the Balkans and for example, Heim (2017) shows that area-integrated drought severity or duration yields different rankings Albania, and wildfire and hydrological impacts in for the major droughts of the 20th and 21st centuries Iberia. over the contiguous United States since 1900. Parts of the Middle East remained in drought, Extensive severe or extreme droughts affected all and particularly severe drought developed in the continents except North America during 2017 (Plate southwestern peninsula of India (especially Kerala) 2.1s). Starting in the Western Hemisphere, persistent during 2017. Farther north in Asia, severe drought conditions were present in the Krasnoyarsk region moderate-to-severe drought conditions affecting of Russia, extending south to northern China. The severe drought in mainland Southeast Asia in 2016 was ended by much wetter conditions during 2017 (Plate 2.1s and Fig. 2.32). Much of Australia was drier than normal during 2017, with severe drought most notable in Tasmania. and evapor ation l 10) —D. G. Miralles, B. Martens, H. E. Beck, A. J. Dolman, C. Jiménez, M. F. McCabe, and E. F. Wood - Evaporation, the return f lux of water from ter restrial ecosystems to the atmosphere, modulates regional energy and water balances and affects . 2.32. Change in drought (mean scPDSI) from 2016 ig F precipitation, both locally and in remote locations. to 2017. Increases in drought severity are indicated by Estimating this variable in near real-time is impor - negative values (brown), decreases by positive values - tant for both agricultural and hydrological manage (green). No calculation is made where a drought index ment, while being able to monitor long-term trends is meaningless (gray areas: ice sheets or deserts with approximately zero mean precipitation). enables the identification of climatological impacts | S37 AUGUST 2018 STATE OF THE CLIMATE IN 2017

58 - on the global hydrosphere. Despite promising ad wet tropics this is typically associated with negative - anomalies of incoming radiation (due to cloudy con vances in the global sensing of evaporation from ditions, for example), while in dry tropics it ref lects space (e.g., Mallick et al. 2016; McCabe et al. 2017b), an abnormally low supply of rainfall (Miralles et al. and a potentially bright future as novel sensors are launched into space (McCabe et al. 2017a; Fisher et 2011). As such, the low evaporation in the semiarid eastern South America likely relates to the drought al. 2017), evaporation remains an elusive variable: in situ measurements are scarce and satellites can only - that started in 2011, intensified in 2012, and per sense it indirectly. As such, models that combine the sisted into 2017 (Brito et al. 2018). These conditions satellite-observed environmental and climatic drivers were particularly intense in the Caatinga shrubland of the f lux are often applied to yield global evapora ecosystems of Brazil. In fact, the strong anomaly in - tion estimates (Wang and Dickinson 2012). Ongoing evaporation, shown in Fig. 2.33 around 20°S in the efforts aim to reduce product latency and improve second half of the year, possibly relates to the shortage in plant-available water in this region. Likewise, the spatial resolution, which is essential for applications negative anomaly in Amazonia, shown in Plate 2.1t, - such as drought monitoring, seasonal extreme fore casting, or irrigation management (Ghilain et al. 2011; persisted throughout the year, ref lecting the impact of the meteorological drought that started in 2015, and Anderson et al. 2011; Mu et al. 2013; McCabe et al. 2017a). The results shown here ref lect recent simula - z o ñ ez-Mu n é was driven by the strong El Niño (Jim tions of the Global Land Evaporation Amsterdam et al. 2016). The legacy of such events on rainforest Model (GLEAM; Miralles et al. 2011) version v3.2a ecosystem functioning is known to extend over pro - by Martens et al. (2017). While GLEAM was not longed periods of time (Zemp et al. 2017). intentionally designed with an operational intent, The spatial patterns found in Africa also relate the long-term record is updated to near real-time on to anomalies in the supply of water to a large extent. an annual basis. Negative anomalies in the Sahel region and Horn of - Africa can be attributed to below -average rainfall The geographical patterns of evaporation anoma - lies shown in Plate 2.1t resemble those from El Niño (Mpelasoka et al. 2018), particularly during the sec - years (see Miralles et al. 2014), yet the ENSO condi ond half of the year. A low water supply also explains tion in 2017 was neutral on average. Consequently, the negative anomaly in the Congo basin. While the regional negative anomalies coincide with those in Congo rainforest is thought to be primarily energy 2016: eastern South America, Amazonia, southern - limited, recent studies have shown evidence of ecosys Africa, the Horn of Africa, and India (Plate 2.1t). tem water limitation (Zhou et al. 2014). Furthermore, In addition, other regions such as central-eastern the f lux of interception loss, i.e., the vaporization of Australia and Central America also experienced low the rainfall captured by the leaves and branches of plants, constitutes a large fraction of the evaporation values. A closer look at these patterns indicates that evaporation was below normal in most of the trop - in the Congo region (Miralles et al. 2010). Conversely, ics during the second half of the year (Fig. 2.33). In the positive anomaly in the Kalahari Desert (Plate 2.1t) relates to above-average rainfall in January and February, which was followed by a positive anomaly - in the atmospheric demand for water in March (Sec tion 2f). Finally, in the absence of particularly strong anomalies in water supply in North America, the positive anomaly in evaporation over the U.S. likely relates to the abnormally high temperatures during the first months of 2017 (see Section 7b2). Figure 2.34 shows the multiannual (1980–2017) variability in terrestrial evaporation derived from GLEAM v3.2a (Martens et al. 2017). A linear trend of −1 approximately 0.3 mm yr ( p = 0.002) for the entire continental surfaces is obtained. While the year-to- year variability is mostly dictated by the variability in the Southern Hemisphere—and particularly affected by the signature of ENSO (Miralles et al. 2014)—the - Zonal mean terrestrial evaporation anoma . 2.33. F ig 1 − lies (mm month ; relative to 1980–2017). (Source: multidecadal trend detected by GLEAM relates al - GLEAM.) most exclusively to the dynamics of evaporation in the | S38 AUGUST 2018

59 ing the 2014–16 episode. ENSO, arguably the most globally impactful mode of variability, encompasses a family of events and episodes. Individually, these exhibit wide-ranging effects across the Indo-Pacific region, with teleconnections to higher latitudes in both hemispheres (Capotondi et. al. 2015; C. Wang et. al. 2017). The sea level pressure derived Southern Oscillation index (SOI; Allan et al. 1996; Kaplan 2011) was primarily positive (the phase typically associated with La Niña conditions) from mid-2016 through the end of 2017 (Fig. 2.35). Nevertheless, the immediate 1 − F . 2.34. Land evaporation anomaly (mm yr ; 1980 – ig impacts of the 2014–16 El Niño episode have lingered 2017 base period) for the NH, SH, and the entire globe in the eastern Australian region, where its inf luence (blue, purple, and black solid lines, respectively). Lin - ear trends in evaporation (dashed lines) and the SOI was particularly profound (Allan and Folland 2017). from NOAA (right axis, shaded area) are also shown. This has taken the form of persistent above-average (Source: GLEAM.) eastern Australian SST anomalies from the Coral Sea Northern Hemisphere. This trend is qualitatively and - southwards via major extensions of the East Austra lian Current into the Tasman Sea region from 2014 quantitatively in agreement with Clausius–Clapeyron through 2017 (Australian Bureau of Meteorology expectations in a warming atmosphere (Miralles 2017; Oliver et al. 2017). Historically, periods of per - et al. 2014; Brutsaert 2017). The global average ter - restrial evaporation in 2017 was slightly below this sistent drought (widespread f looding) in this region trend and close to the 1980–2016 mean (Fig. 2.34). have been strongly amplified by protracted El Niño Notwithstanding the novel insights made available (La Niña) episodes (Murphy and Ribbe 2004; Allan et al. 2018, manuscript submitted to from remote platforms, trends in satellite-based ). Atmosphere Mean sea level pressure (MSLP) can also be used to evaporation should be interpreted with care, and derive indices of many regional modes of variability the weighted use of multiple retrieval approaches is that drive significant weather and climate events usually recommended (Miralles et al. 2016; McCabe et al. 2016). Unfortunately, as of today, algorithms dedicated to estimating evaporation using satellite observations at global scales are mostly intended for research applications and are not regularly updated in near- real time (Fisher et al. 2017). e. Atmospheric circulation sea level pressure m ean 1) r e l at e d a n d m o d e s o f variabilit y —R. Allan and C. K. Folland - Overviews of the most re cent El Niño have been made in papers such as L’Heureux et al. (2017), but the protracted nature of the El Niño from 2014 to 2016 should also be noted F ig . 2.35. Time series for modes of variability described using sea level (Allan and D’Arrigo 1999; pressure for the (left) complete period of record and (right) 2006–17. (a),(b) Allan et al. 2018, manuscript SOI (provided by the Australian Bureau of Meteorology); (c),(d) AO (NCEP Atmosphere ). submitted to Climate Prediction Center); (e),(f) AAO (NCEP Climate Prediction Cen - The climate system exhibited ter); (g),(h) winter (Dec–Feb) NAO average (NCAR; presented for winter weak La Niña (positive SOI) at the beginning of each year so winter 2017/18 is not shown); (i),(j) summer (Jul–Aug) SNAO average (Folland et al. 2009). - to neutral conditions follow | S39 AUGUST 2018 STATE OF THE CLIMATE IN 2017

60 (Kaplan 2011): the Arctic Oscillation (AO); North Atlantic Oscillation (NAO); summer NAO (SNAO); and the Antarctic Oscillation (AAO) (Fig. 2.35). In the Northern Hemisphere, the last six winters have displayed broadly positive NAO conditions but a diverse range of circulation patterns. During the early winter of 2015/16 the NAO oscillated between phases, with a deep trough over the North Atlantic - leading to an enhanced jet stream that directed a se ries of extratropical cyclones toward northern Ireland and Scotland–northern England (Fig. 2.36). By the mid-to-latter part of the 2015/16 winter the pattern had changed, with the NAO swinging from slightly negative in January 2016 to positive in February 2016 (Allan and Folland 2017). The 2016/17 boreal winter was marked by an increasingly positive NAO through - mid-December 2016, temporarily negative NAO val ues around the start of 2017, and then a f luctuation between phases for the rest of January (Fig. 2.36; Allan and Folland 2017). During the 2017/18 boreal winter, the NAO has been mainly positive (Fig. 2.36). As a consequence, temperatures in Europe were mild to warm, and the region experienced its fifth warm - est year on record, while Portugal in particular was strongly impacted, with its driest April to December period in its 87-year record (Section 2d9, Section 7f4). As in 2016/17, the Aleutian low was markedly weakened, leading to reduced rainfall and conditions conducive to major wildfires in the British Columbia region of Canada (Section 2h3; Figs. 2.36a–c; Section 7 b1). In 2017, the phase of the SNAO defined over July and August as in Folland et al. (2009) was on aver - age slightly negative (Figs. 2.37a,b). As in 2016 (Al - lan and Folland 2017), there was a rather persistent anticyclonic anomaly over southern Greenland in both months, but this was markedly less intense and smaller than in 2016. This feature is normally associ - ated with a negative SNAO. In fact, July (Fig. 2.37a) had a variable and overall negative SNAO as seen in the daily values (Fig. 2.37c). The most notable feature in summer 2017 was a mostly strong negative SNAO that lasted ten days from the end of July into early August. August overall showed a near-neutral SNAO despite the anticyclonic MSLP anomaly over southern Greenland (Fig. 2.37b) and the variable August daily SNAO series. The multidecadal tendency noted in . 2.36. Boreal winter sea level pressure anomalies ig F Allan and Folland (2017) toward a more negative (hPa; 1981–2010 base period) averaged over Dec–Feb SNAO index since 1970 continued to slow. Thus, for (a) 2015/16, (b) 2016/17, and (c) 2017/18. NAO daily the average level of the SNAO index in the last five time series (hPa) for winter (d) 2015/16, (e) 2016/17, and years is near the average observed over 1850–1960 (f) 2017/18. The 5-day running mean is shown by the solid black line. The data are from HadSLP2r (Allan but is considerably more negative than the positive and Ansell 2006). | S40 AUGUST 2018

61 ), and of a negative AAO submitted to Atmosphere (Fig. 2.35), there was a major reduction in the WAP sea ice margin centering on November 2016 and a slight recovery in extent through 2017 (see Section 6e) despite a return to positive AAO values (Fig. 2.35f; http://nsidc.org/data/seaice_index/). urface —C. Azorin-Molina, R. J. H. Dunn, winds s 2) C. A. Mears, P. Berrisford, and T. R. McVicar Over land, observations of globally averaged wind speed continued to “recover” (commencing in ~2013; Dunn et al. 2016a; Azorin-Molina et al. 2017a) from the previous slowdown of winds (from ~1960s onwards; McVicar et al 2012), termed “stilling” by Roderick et al. (2007). Surface wind speed increased in 2017 (Fig. 2.38a), showing a global (excluding Australia) average wind speed anomaly of +0.024 m −1 with respect to the 1981–2010 climatology (Table s 2.4). Regionally, this recent rebound was caused by −1 positive anomalies for central (+0.142 m s ) and East −1 −1 ) Asia, with Europe (+0.002 m s ) being (+0.108 m s −1 very close to average. North America (−0.068 m s ) showed a negative anomaly but less negative than its 2012 record lowest anomaly. In contrast, Australia F ig . 2.37. MSLP anomalies (hPa; 1961–90 base period) in (a) Jul and (b) Aug 2017 over the extratropical North Atlantic and Europe. (c) Daily SNAO index for Jul and Aug 2017, calculated from eigenvectors of the daily SNAO. index averaged over the two decades 1966–1985. Linderholm and Folland (2017) provide more detail on recent multidecadal changes in the SNAO index. In the Southern Hemisphere, the AAO has been predominantly in its positive phase since 2015/16 (Fig. 2.35). This favors reduced sea ice extent in the West Antarctic Peninsula (WAP) region, owing to enhanced westerly wind conditions (Stammerjohn F ig . 2.38. Global (excluding Australia) and regional et al. 2008). In the interplay between the protracted annual time series of land surface wind speed anomaly El Niño, which favors a weaker polar jet stream, and 1 − (m s ; relative to 1981–2010) using HadISD2 (1973– a positive AAO mode, with stronger westerly winds, 2017), an Australian dataset, and ERA-Interim (1979– the former appears to have dominated. With the ces - 2017), MERRA-2 (1980–2017) and JRA-55 (1970–2017). sation of the protracted El Niño episode in mid-2016 Occurrence frequencies (in %) for wind speeds (b) >3 1 − 1 − m s and (c) >10 m s do not include Australia. (Allan and Folland 2017; Allan et al. 2018, manuscript | S41 AUGUST 2018 STATE OF THE CLIMATE IN 2017

62 reanalyses to reproduce 2.4. Global and regional statistics for land surface wind speed using t le AB wind speed trends. This observational HadISD2 and Australian datasets for 1979–2017. is shown in Fig. 2.38a, as the long-term variability Trend 1979–2017 Mean Anomaly 1 1 − − (m s decade ) and Number of of reanalyzed land surface Region 1981–2010 2017 5th to 95th percentile Stations winds is almost stable as 1 − − 1 (m s ) ) (m s confidence range opposed to the decline in - the observations. The un Globe 0.066 − 3.332 +0.024 (excluding 2632 derestimation of the mag - − ( 0.074 0.058) − Australia) nitude of reanalysis wind trends is mainly due to the North − 0.088 3.728 598 0.068 − America ( 0.099 − 0.076) − shortcomings in the simu - lation of near-surface layer − 0.057 3.662 Europe +0.002 788 processes (e.g., McVicar et ( − 0.070 0.047) − al. 2008; Pryor et al. 2009; Central − 0.128 Vautard et al. 2010). 2.875 +0.142 263 Asia 0.144 − ( − 0.099) The global land wind - speed trend from obser − 0.036 +0.108 2.738 East Asia 474 −1 vations was −0.066 m s − 0.045 ( − 0.027) −1 for 1979–2017, decade Australia 28 2 .091 − − 0 . 311 0.092 which is slightly less nega - tive than the 1979–2016 −1 −1 −1 ; Azorin-Molina et al. ) had the lowest anomaly in its time se trends (−0.070 m s (−0.311 m s decade - ries. Excluding the latter, the 2017 anomalies continue 2017a). As shown in Table 2.4, the strongest 1979–2017 −1 negative trends are in Central Asia (−0.128 m s to support the reversal in the “stilling” detected over de - −1 −1 −1 decade the last few years. This rebound of wind speeds has ) and North America (−0.088 m s cade ), whereas the weakest ones are in East Asia (−0.036 m also been reported elsewhere (South Korea, Kim and −1 −1 −1 −1 ). For s decade decade ) and Europe (−0.057 m s Paik 2015; and Saudi Arabia, Azorin-Molina et al. 2018a). The recent strengthening in terrestrial wind all these regions, the magnitude of observed trends is −1 speed is much clearer for the moderate (>3 m s ) than also less negative than Azorin-Molina et al. (2017a), −1 −1 −1 ) winds (Figs. 2.38b,c), as the except for Australia (−0.092 m s decade ). Indi - the strong (>10 m s vidual station trends (Fig. 2.39) are 64.9% negative occurrence of moderate winds has slightly increased after a steady slowdown since records began. The from the HadISD2 dataset, and 96.4% negative for the Australian dataset. Even though a recent recovery recovery of surface winds is not detected for those of of terrestrial surface wind speeds is detected, when - strong intensity in 2017, which only showed a stabi lization in frequency recently. considering the past four decades “stilling” remains widespread (McVicar et al. 2012). Two observational databases from anemometer records were chosen for evaluating the spatio–tempo - ral variability of land-surface winds globally: (1) the HadISD2 (1973–2017; Dunn et al. 2012, 2016b) and (2) an Australian dataset (1979–2017; McVicar et al. 2008). As a result of unresolved differences for the wind run and wind speed data over Australia, this re - gion is treated separately (see Dunn et al. 2016a). Both data sources were subject to quality control checks resulting in 2660 series for 1979–2017. Additionally, three reanalysis products (MERRA-2, 1980–2017; Gelaro et al. 2017; ERA-Interim, 1979–2017; Dee et al. 2011a; and JRA-55, 1970–2017; Kobayashi et al. 2015) were used to assess wind speed variability − 1 1 − (m s . 2.39. - or the ob ) f decade F Wind speed trends ig across land and ocean surfaces. A global reanalysis servational HadISD2 and Australian datasets (circles) intercomparison (Torralba et al. 2017) has pointed over land for 1979–2017, and MERRA2 over land/ice and RSS over ocean for 1988–2017 (shaded areas). out the large uncertainty in the ability of atmospheric | S42 AUGUST 2018

63 3) pper air winds —L. Haimberger, M. Mayer, and u V. Schenzinger Figure 2.41 shows global (land + ocean) mean 850- hPa wind speed anomalies from reanalyses and in situ upper air (TEMP and PILOT) observations, for com - parison with surface wind speed anomalies in Section 2e2. There is a general tendency towards higher wind speeds at this level, at least in the reanalysis data, but only trends from ERA-Interim (Dee et al. 2011a) and MERRA2 (Gelaro et al. 2017) for 1979–2017 are statistically significant (95% confidence). Trends are larger over the oceans, particularly in the Pacific trade wind region, and weaker over land. At higher levels (200–300 hPa), the global wind trends turn negative 1 − (not shown) but remain weak. ; . 2.40. Global average surface wind anomaly (m s F ig The annual mean 850-hPa wind speeds for 2017 1981–2010 base period) over ocean from (a) satellite −1 radiometers and (b) reanalyses. are clearly above normal (0.22 m s in ER A-Interim), consistent with the overall increasing trend at this Satellite-borne microwave radiometers and the level and also with the recovery of the surface winds three above-mentioned reanalysis products were from wind stilling noted in Section 2e2. They appear anomalously high particularly in the tropics, as can be chosen for assessing surface wind variability over seen from Plate 2.1w, with stronger-than-normal east oceans. During 2017, global wind speed anomalies for - the satellite estimates (Fig. 2.40a) were close to zero, erlies over large regions. This result should be taken with reanalysis showing neutral to positive anomalies with care though, because Liu and Allan (2018) re - (Fig. 2.40b). In comparison to 2016, over ocean, glob cently have detected problems with reanalysis winds. - ally averaged wind speed anomalies tended to be less Over land (not shown), the 850-hPa trends from −1 m s reanalyses are only weakly positive (0.01 negative (or even positive) for all products; in agree de - - −1 ment with the observed recovery of terrestrial surface in ERA-Interim for the 1979–2017 period; the cade −1 winds. The strongest spatial anomalies for 2017 (Plate ). They are still slightly 2017 anomaly is 0.13 m s 2.1v) corresponded to: (1) strong negative anomalies more positive than the surface wind trends over land (see Section 2e2). The in situ upper air dataset dominating in the Gulf of Alaska and for much of the (GRASP; Ramella Pralungo et al. 2014) has negative Atlantic Ocean north of the equator, as well as in the −1 −1 trends (−0.03 m s decade ) in the period 1979–2016. southwest Pacific–Tasman Sea and western Indian Ocean; and (2) strong positive anomalies mostly The anomalies of this dataset in the most recent years observed over the South Pacific and South Atlantic were, however, also slightly positive, similar to the surface wind anomalies (see Section 2e2). Oceans and parts of the Southern and Arctic Oceans. The limited knowledge about the causes behind the stilling phenomenon and the recent recovery of - surface winds suggests the need for comprehensive at tribution analyses of wind speed variability over land and ocean and at different altitudes (i.e., including high-elevation stations; Azorin-Molina et al. 2017c). In the last few years the scientific literature has at - tributed the stilling over land to three major drivers: (1) increase of surface roughness (Vautard et al. 2010; Bichet et al. 2012; Wever 2012; Wu et al. 2016); (2) large-scale atmospheric circulation changes (Azorin- Molina et al. 2014, 2016); and (3) instrumental issues (Wan et al. 2010; Azorin-Molina et al. 2017b, 2018b). F ig . 2.41. Annual anomalies of global mean wind speed - The attribution analysis of the recent recovery of sur − 1 (m s ; base period 1981–2010) at 850 hPa from four face winds is also complicated by interplaying factors, reanalyses and one observational dataset (GRASP; and future research should fill this knowledge gap. Ramella Pralungo et al. 2014). The numbers in brackets 1 − 1 − are linear trends in m s decade ; valid for 1979–2017. | S43 AUGUST 2018 STATE OF THE CLIMATE IN 2017

64 To date, there is no independent satellite-derived product for upper air winds. Atmospheric motion vectors from AVHRR have been reprocessed recently at EUMETSAT and are ready to be as - similated (Schulz et al. 2017); however no gridded product has been generated. The 2017 Atlantic hurricane season - (see Section 4f2) deserves special atten tion since it was exceptionally intense, particularly at peak time (mid-August– F ig . 2.42. Aug–Sep 2017 average of velocity potential anomaly (× September). From an upper air circula - − 6 2 1 ) and divergent wind at 200 hPa (vector arrows) com s - m 10 tion perspective, one cause that may have pared to the 1979–2016 Aug–Sep climatology. Velocity potential favored the observed large number of anomaly minima indicate positive divergence anomalies. (Source: strong hurricanes is anomalously large ERA-Interim.) upper-level divergence, a parameter whose State of the importance has been stressed in previous Climate reports. A second factor may be the abun - dance of strong tropical easterly wave disturbances that can amplify under favorable conditions (Dieng et al. 2017; Russell et al. 2017). Figure 2.42 shows that upper level divergence averaged over August and September was anomalously positive throughout the western Atlantic. Together with negative (positive) values over the eastern (western) tropical Pacific, this is consistent with weak La Niña conditions establish - ing at that time (Mayer et al. 2013). However, it is difficult to separate cause and effect for the anomaly in the upper air circulation over the western Atlantic, because the strong hurricanes themselves potentially F . 2.43. Hovmöller diagrams of 850-hPa meridional ig contributed to the anomalies in that region. − 1 wind (m s ) averaged over 8°–18°N (the region with Tropical wave activity was also high, as shown strongest wave disturbances according to Dieng et in the Hovmöller diagram [similar, for example, al. 2017) for the peak hurricane season 15 Aug–1 Oct between 70°W and 0°. 6-hourly ERA-Interim wind to Seo et al. (2008)] in Fig. 2.43 for the period 15 fields at 1° resolution without any filter have been used. August–1 October, during which four major hur - Upper panels show standard deviation of meridional ricanes were observed. In particular the standard − 1 wind (m s ) as a function of longitude in individual deviation of meridional wind speed in the west years (blue) and for the 1979–2017 average (red). Se - central Atlantic was high compared with 2015, 2016, lected waves that developed into hurricanes in 2017 and the 1979–2017 climatology, which shows slowly are marked with lines and named. decaying wave activity from the maximum near the anomaly decayed in 2017 and the usual oscillation West African coast toward the west. The strong waves resumed with a relatively large, but not exceptional, in 2017 together with the anomalously high oceanic amplitude (Online Fig. S2.16). However, the westerly heat content (see Section 3c) likely fostered the quick formation of Irma, Jose, and Maria, which developed - wind regime at 20-hPa lasted for 24 months, com pared to the average duration of 13 months and the into major hurricanes already over the west central mean QBO period of 28 months (e.g., Schenzinger Atlantic. Hurricanes are visible in Fig. 2.43 as regions of extreme east–west wind gradients. et al. 2017). A new analysis by Watanabe et al. (2018) The quasi-biennial oscillation (QBO; see also points to interaction of extratropical Rossby waves Section 2b5) exhibited an unprecedented anomaly at with the mean equatorial f low as main reason for the anomaly. Comparison of this episode with results the beginning of 2016. It was characterized by highly - from historical CMIP climate model runs shows only unusual and strong upward propagation of equato rial wind regimes, particularly between 10-hPa and one similar event in the model data (Osprey et al. 2016; see also Schenzinger 2016). 40-hPa (Newman et al 2016; Dunkerton 2016). The | S44 AUGUST 2018

65 f. Earth radiation budget 1) e radiation bud Get at top - of - atmosphere — arth T. Wong, D. P. Kratz, P. W. Stackhouse, Jr., P. Sawaengphokhai, A. C. Wilber, S. K. Gupta, and N. G. Loeb The energetic state of the Earth–atmosphere sys - tem is defined by the balance of the incoming total solar irradiance (TSI) from the Sun with the ref lected shortwave (RSW) and the outgoing longwave radia - tion (OLR) from Earth. This balance characterizes Earth’s radiation budget (ERB) at the top of the at - mosphere (TOA) and drives weather processes and climate forcings as well as climate feedbacks. An analysis of all CERES ERB measurements (Table 2.5) shows that the 2017 global annual mean OLR remained approximately unchanged while the −2 relative to their cor - RSW decreased by ~0.05 W m responding values in 2016. Over the same timeframe, F ig . 2.44. Time series of global monthly mean de - −2 − 2 the global annual mean TSI declined by ~0.10 W m . seasonalized anomalies (W m ) of TOA Earth radia - The sum of these components amounts to a small tion budget for OLR (upper), absorbed shortwave (TSI- −2 RSW; middle), and total net (TSI-RSW-OLR; lower) reduction of ~0.05 W m in the global annual mean from Mar 2000 to Dec 2017. Anomalies are relative total net radiation into the Earth climate system for to their calendar month climatology (2001–16). Time 2017 as compared with 2016. Relative to the multiyear series shows the CERES EBAF Ed4.0 1Deg data (Mar - data average from 2001 to 2016, the 2017 global an 2000–Sep 2017) in red and the CERES FLASHFlux nual mean f lux anomalies (Table 2.5) are +0.50, −0.10, version 3C data (Oct–Dec 2017) in blue; see text for −2 for OLR, TSI, RSW, and total −0.80, and +0.20 W m merging procedure net f lux, respectively. These changes are at or within the corresponding 2-sigma interannual variability mostly positive during 2017, and the magnitudes of (Table 2.5) for this period. this anomaly were larger than the corresponding OLR anomaly. The absorbed shortwave anomaly The global monthly mean anomaly time series of started the year with a maximum value of +1.9 TOA f luxes (Fig. 2.44) reveals that the global monthly −2 −2 W m , decreased to a minimum value of −0.2 W m mean OLR anomaly stayed mostly positive through - out 2017. The OLR anomaly began 2017 with a value in October, then climbed back to a positive value at −2 of +0.9 W m , reached its maximum value of +1.2 year end. For the year as a whole, the 2017 global −2 annual mean absorbed shortwave anomaly is +0.7 W m in April, dropped to its minimum value of −0.2 −2 −2 −2 . The global monthly mean total net anomaly, W m W m in August, then oscillated around +0.4 W m which is calculated from absorbed shortwave anoma - for the rest of the year. The global monthly mean ab - ly minus OLR anomaly, began 2017 with a maximum sorbed shortwave (TSI−RSW) anomaly also remained t le 2.5. Global annual mean TOA radiative flux changes between 2016 and 2017, the AB global annual mean radiative flux anomalies relative to their corresponding 2001–16 mean climatological values, and the 2-sigma interannual variabilities of the 2001–16 2 − global annual mean fluxes (all units in W m ) for the outgoing longwave radiation (OLR), total solar irradiance (TSI), reflected shortwave (RSW) and total net fluxes. All flux − 2 values have been rounded to the nearest 0.05 W m . 2017 Anomaly Interannual Variability One Year Change (2017 minus 2016) (2001 to 2016) (Relative to Climatology) 0.00 +0.50 ±0.60 OLR TSI ±0.15 0.10 − 0.10 − RSW ±0.80 0.05 − 0.80 − Net +0.20 ±0.75 − 0.05 | S45 AUGUST 2018 STATE OF THE CLIMATE IN 2017

66 −2 value of +1.0 W m , remained mostly positive for ~278 ppm in 1750 (Etheridge et al. 1996). Since then, 15 ~430 Pg C (1 Pg C = 10 to eight months, declined to mostly negative in the last g C) were emitted as CO 2 four months of the year, and ended the year with a the atmosphere from fossil fuel burning and cement −2 - production (Boden et al. 2017). Based on observa value of −0.4 W m . The positive absorbed shortwave and N /O (Manning and tions of atmospheric CO anomaly in 2017 dominated the negative effect of OLR 2 2 2 anomaly and resulted in a slightly positive 2017 global Keeling 2006) and increased carbon in the oceans −2 . Long- annual mean total net anomaly of +0.2 W m (Sabine et al. 2004), most of the anthropogenic CO 2 not remaining in the atmosphere was taken up by the term trend analyses that include the last three months of the merged dataset are discouraged because of the oceans (Tans 2009). While the terrestrial biosphere natural f luctuation in ERB components, uncertainty , net is currently also a net sink for fossil fuel CO 2 from the data merging process, and potential for drift to the atmosphere from land use emissions of CO 2 in the FLASHFlux product. change prior to ~1940 offset recent terrestrial uptake (Tans 2009). These mass balance considerations The TSI data used in this study are provided by the Total Irradiance Monitor aboard the Solar Radiation overwhelmingly suggest that the observed increase in since 1750 is caused by combustion and Climate Experiment (SORCE) mission (Kopp and atmospheric CO 2 Lean 2011) and the Royal Meteorological Institute of of fossil fuels. This conclusion is further supported 12 14 12 13 Belgium composite dataset (Dewitte et al. 2004), both - C and C of at C/ C/ by measured decreases in renormalized to the SORCE Version 15. The RSW mospheric CO , and an increase in the north–south 2 gradient of atmospheric CO abundance (Tans 2009). and OLR data were obtained from the Clouds and 2 has risen from 0.6 the Earth’s Radiant Energy System (CERES) mission The global growth rate of CO 2 −1 . and (Wielicki et al. 1996, 1998) aboard in the early 1960s to an average of 2.3 Te r ra Aqua ± 0.1 ppm yr −1 ppm yr during the past ten years, with interannual The time series (Fig. 2.44) was constructed from −1 variability of ±0.5 ppm yr (1-sigma) (Fig. 2.45). The the CERES EBAF (Energy Balanced And Filled) Ed4.0 increase in global annual mean CO product (Loeb et al. 2009, 2012, 2018) for March 2000 from 2016 to 2017 2 to September 2017 and from the CERES Fast Long was 2.2 ± 0.1 ppm. In the two years prior to this (2015 - −1 wave and Shortwave Radiative Fluxes (FLASHFlux) and 2016), atmospheric CO increased by 3.0 ppm yr . 2 version 3C product (Kratz et al. 2014), for October to The strong El Niño that peaked in late-2015 contrib - uted to this strong CO increase (Betts et al. 2016). December 2017. The normalization of the FLASHFlux 2 data (Stackhouse et al. 2016) results in a 2-sigma monthly uncertainty of ±0.43, ±0.08, ±0.20 and ±0.55 −2 W m for the OLR, TSI, RSW, and total net radiation, respectively. g. Atmospheric composition l on G - lived Greenhouse Gases —E. J. Dlugokencky, 1) ü h B. D. Hall, S. A. Montzka, G. Dutton, J. M le, and J. W. Elkins The three long-lived greenhouse gases (LLGHGs) with the largest contributions to climate forcing are, ), methane in decreasing order: carbon dioxide (CO 2 ), and nitrous oxide (N O). Systematic measure - (CH 4 2 ments of CO began at Mauna Loa, Hawaii (MLO) in 2 1958, when the atmospheric CO abundance was ~315 2 - ppm (parts per million in dry air). In 2017, MLO an nually averaged CO reached 406.5 ± 0.1 ppm (www 2 .esrl.noaa.gov/gmd/ccgg/trends/; all uncertainties are 68% confidence intervals, unless noted otherwise), at Earth’s while preliminary globally averaged CO 2 surface was 405.0 ± 0.1 ppm (Fig. 2.45a, see www.esrl F ig . 2.45. Global mean surface mole fractions (in dry .noaa.gov/g md/ccgg /t rends/g loba l.ht m l). O (ppb) (ppm), (b) CH (ppb), and (c) N air) of (a) CO 4 2 2 prior to 1958 is The atmospheric history of CO 2 derived from the NOAA sampling network. Growth - determined from air extracted from ice in Green rates are shown on the right axis. (Measurements were not sufficient to calculate instantaneous growth rates land and Antarctica. From those measurements, it is for N O with reasonable certainty prior to 1995). 2 known that the abundance of atmospheric CO was 2 | S46 AUGUST 2018

67 The 2017 globally averaged methane mole fraction at Earth’s surface was 1849.7 ± 0.8 ppb (Dlugokencky 2018). The increase in annual mean CH from 2016 4 to 2017 was 6.9 ± 0.9 ppb, comparable to the average −1 ; growth rate over the past 10 years (+7.1 ± 2.6 ppb yr the uncertainty is the standard deviation of annual increases). Since 1750, CH has increased by ~1128 4 ppb from 722 ± 15 ppb. Atmospheric CH is inf luenced by a complex 4 mix of sources and sinks, with emissions from both anthropogenic (~60%) and natural (~40%) sources (Fung et al. 1991). Its main loss process, atmospheric − 2 F . 2.46. Direct radiative forcing (W m ig ) due to 5 oxidation initiated by reaction with hydroxyl radical major LLGHG and 15 minor gases (left axis) and the (OH), is the largest term in the atmospheric CH bud- 4 associated values of the AGGI (right axis). get of sources and sinks. Total global emissions of CH 4 are well-constrained by the atmospheric measure - ments and an estimate of its lifetime (Dlugokencky (CO O, CFC-11, and CFC-12) and 15 minor , CH , N 2 2 4 gases (Hofmann et al. 2006; Table 2.6; Fig. 2.47; www et al. 2011), but the magnitude and trend in emissions atmo - from individual sources and trends in CH .esrl.noaa.gov/gmd/aggi/). The AGGI represents the 4 spheric lifetime are still highly uncertain. In the past annual cumulative radiative forcing of these gases three decades, the CH relative to the Kyoto Protocol baseline year of 1990 growth rate has undergone 4 −2 (2.16 W m ). It does not include indirect radiative long- and short-term changes (red line in Fig. 2.45b). forcing (e.g., inf luences on ozone and water vapor). Analysis of these changes can be used to improve −2 understanding of processes that emit and remove In 2017, CO direct radiative contributed 2.01 W m 2 CH forcing, or about 66% of the combined forcing of , but so far, causes behind even large changes 4 −2 3.06 W m from LLGHGs. CH and N have not been unambiguously identified. Numerous O contributed 2 4 −2 −2 0.5 W m (6.5%) respectively, (16%) and 0.2 W m publications address the increase in growth rate that while the sum of halogenated gases, including CFCs, abundance and started in 2007; measurements of CH 4 - its isotopic composition strongly suggest increased HCFCs, and HFCs, among others (Table 2.6), con emissions from biogenic sources, both natural and anthropogenic (Nisbet et al. 2016; Schaefer et al. 2016; Schwietzke et al. 2016), rather than changes in fossil fuel–related emissions. Changes in other CH sources 4 loss rate (Prather (e.g., Worden et al. 2017) and CH 4 and Holmes 2017) have also been implicated, but because the problem is underconstrained by observa - tions, all explanations are uncertain. Nitrous oxide (N O) is both a greenhouse gas and 2 an ozone-depleting substance (Ravishankara et al. 2009). Sources include natural and agricultural soils as well as oceans. Anthropogenic activity is thought to contribute about one-third to total global emissions of −1 ~18 Tg yr (Ciais et al. 2013). Except for a brief period O has been increasing in the 1940s, atmospheric N 2 steadily throughout the industrial era (MacFarling O Meure et al. 2006). The mean global atmospheric N 2 mole fraction in 2017 was 329.8 ± 0.1 ppb, an increase of 0.9 ppb from 2016 (Fig. 2.45c). This 0.9 ppb annual change is similar to the average annual change over the last two decades (0.85 ± 0.17 ppb). F ig . 2.47. Global mean mole fractions at Earth’s surface The NOAA Annual Greenhouse Gas Index (ppt, dry air) for several LLGHG, many of which also deplete stratospheric ozone. See Table 2.6 for the 2017 - (AGGI) (Fig. 2.46) summarizes trends in the com global mean mole fractions of these gases. bined direct radiative forcing by five major LLGHGs | S47 AUGUST 2018 STATE OF THE CLIMATE IN 2017

68 t AB 2.6. Summary table of long-lived greenhouse gases for 2017 (CO le mixing ratios are in ppm, N O and 2 2 CH in ppb, and all others in ppt). 4 Radiative Mean Surface Mole Lifetime Chemical Industrial Designation 2017 Fraction AGGI Efficiency ODGI (years) Formula or Common Name b a − 2 − 1 (change from prior year) ) ppb (W m c – 5 CO Y 405.0 (2.2) N Carbon Dioxide 1.37 × 10 2 c – 4 N CH Methane 1849.7 (6.9) Y 9.1 3.63 × 10 4 c,d – 3 Y O N 329.8 (0.9) 123 N Nitrous Oxide 3.00 × 10 2 Chlorofluorocarbons c,d F Y Y 0.26 C F C -11 CCl 0.8) − 228.9 ( 52 3 c,d Y F 0.32 Y CFC-12 CCl 509.3 ( − 2.9) 102 2 2 c FCClF Y 0.30 Y CCl C F C -113 0.5) 70.9 ( − 93 2 2 Hydrochlorofluorocarbons CHClF Y Y HCFC-22 240.8 (3.3) 11. 9 0.21 2 0.16 CCl 9.4 F Y Y HCFC-141b CH 24.5 ( − 0.4) 3 2 CH 22.1 (0.2) CClF 0.19 18 Y Y HCFC-142b 2 3 Hydrofluorocarbons N FCF HFC-134a Y 0.16 CH 95.7 (6.1) 14 2 3 N Y CHF 0.10 CH HFC-152a 6.8 (0.2) 1.6 2 3 CF Y N 0.16 CH HFC-143a 20.6 (1.6) 51 3 3 0.23 N Y CF HFC-125 CHF 22.8 (2.7) 31 3 2 F N N 0 .11 CH HFC-32 13.0 (1.8) 5.4 2 2 Y N 0.18 CHF HFC-23 29.9 (1.0) 228 3 0.22 CF N CH N CF CH HFC-365mfc 0.93 (0.05) 8.7 3 2 2 3 0.26 N CHFCF N CF HFC-227ea 1.29 (0.12) 36 3 3 Chlorocarbons Methyl Chloroform CCl CH Y Y 0.07 5.0 − 0.4) 2.2 ( 3 3 c,d 33 Y Y 0.17 CCl Carbon Tetrachloride − 0.9) 80.2 ( 4 Cl N Y 0.01 0.9 CH Methyl Chloride − 12 .1) 5 47. 3 ( 3 Bromocarbons Methyl Bromide CH 0.004 Br N Y 0.8 0.2) − 6.6 ( 3 0.29 Y Y Halon 1211 CBrClF 16 3.43 ( − 0.09) 2 Y Y 0.30 3.26 (0.00) 72 CBrF Halon 1301 3 28 0.01) − 0.4 ( Y Y 0.31 CBrF CBrF Halon 2402 2 2 Fully fluorinated species SF 9.26 (0.34) Y N 0.57 Sulfur Hexafluoride >600 6 c 000 N N 0.09 83.6 (0.9) ~50 PFC-14 CF 4 c P F C -116 F 000 N N 0.25 4.66 (0.10) ~10 C 6 2 a Radiative efficiencies were taken from IPCC AR5 (Myhre et al. 2013). Steady-state lifetimes were taken from Myhre et al. (2013) (CH ), 4 , numerous removal processes ), Ko et al. (2013), Liang et al. (2016) (CCl ), and Carpenter et al. (2014). For CO Ray et al. (2017) (SF 4 6 2 complicate the derivation of a global lifetime. b Mole fractions are global, annual surface means for the indicated calendar year determined from the NOAA cooperative global air sampling network (Hofmann et al. 2006), except for PFC-14, PFC-116, and HFC-23, which were measured by AGAGE (Mühle et al., 2010; Miller et al., 2010). Changes indicated in brackets are the differences between the 2017 and 2016 means. c Preliminary estimate. d Global mean estimates derived from multiple NOAA measurement programs (“Combined Dataset”). | S48 AUGUST 2018

69 −2 tributed 0.34 W m -related production of (11%). C H 4 tropospheric O O contributed and stratospheric H 3 2 −2 ~0.3 W m indirect radiative forcing (Myhre et al. 2013). The combined direct forcing in 2017 represents a 41% increase since 1990 (2017 AGGI = 1.41). zone - depletin G Gases —B. D. Hall, S. A. Montzka, 2) o G. Dutton, B. R. Miller, and J. W. Elkins Chlorine and bromine from CFCs, HCFCs, halons, and other ozone-depleting substances (ODS) are - released in the stratosphere, causing ozone destruc tion. The emissions and atmospheric abundances of most ODS are declining as expected due to con - trols implemented in the Montreal Protocol and its Amendments (Carpenter et al. 2014). An exception is carbon tetrachloride, which has not decreased as expected for a number of years (Carpenter et al. 2014). Furthermore, it has recently been reported that the atmospheric abundance of CFC-11 has not declined as rapidly as expected, leading to concern that sustained . 2.48. (a) EESC (ppt) and (b) NOAA ODGI. The F ig increased emissions of CFC-11 would substantially ODGI represents the relative mole fractions of re - delay the recovery of stratospheric ozone (Montzka et active halogen in the midlatitude (open circles) and −1 Antarctic stratosphere (closed circles) scaled such that ± 0.3 ppt yr al. 2018). CFC-11 declined at a rate of 2.1 ODGI = 100 at maximum EESC and zero in 1980. Both from 2002 through 2011, but that rate slowed to 1.0 EESC and ODGI are derived from NOAA surface mea - −1 0.2 ppt yr from mid-2015 to mid-2017 (Montzka ± surements of long-lived ODS (circles) or, for earlier et al. 2018). The observed changes in CFC-11 are due years, WMO scenarios (dashed lines; N. Harris et al. to an increase in emissions, although some changes 2014). The EESC and ODGI values from 1992 forward in atmospheric transport also contributed in some correspond to Jan of each year. years. In addition, emissions and abundances of some short-lived chlorine-containing gases, which EESC and ODGI have been calculated since 1992 for two representative stratospheric regions—Ant are not controlled by the Protocol, have increased - recently and could delay ozone recovery if they were arctica and the midlatitudes—that differ in total to continue to increase at similar rates in the future. available reactive halogen (Fig. 2.48). EESC is larger in the Antarctic stratosphere than in the midlatitudes Cl For example, the atmospheric abundance of CH 2 2 has approximately doubled over the past fifteen years because more ozone-reactive halogen is released dur - ing the longer transit time to the Antarctic from mid (Fig. 2.47; Hossaini et al. 2017). - Equivalent effective stratospheric chlorine (EESC) latitude surface-based source regions. ODGI values is a measure of the ozone-depleting potential of the at the beginning of 2017 were approximately 80 and 56 for the Antarctic and midlatitudes, respectively. stratospheric halogen loading at a given time and - These represent 20% (100 minus 80) and 44% (100 place. As EESC declines, stratospheric ozone is show minus 56) reductions from the peak values in EESC ing signs of recovery (Kuttippurath and Nair 2017; Strahan and Douglass 2018; see Sections 2g4 and 6h). over Antarctica and the midlatitudes, respectively, EESC is calculated from global average surface mole toward the 1980 benchmark values. fractions of long-lived ozone-depleting gases and — weighting factors that include surface-to-stratosphere S. Rémy, N. Bellouin, A. Benedetti, and erosols 3) a transport times, mixing during transit, photolytic O. Boucher Atmospheric aerosols are a key component of air reactivity, and ozone-destruction efficiency (Montzka quality and are now recognized as a serious public et al. 1996; Newman et al. 2007). Short-lived gases Cl - such as CH are not included in EESC. NOAA health issue (WHO 2013). They also play an impor 2 2 tant role in the climate system, by scattering and tracks changes in EESC with an Ozone-Depleting Gas Index (ODGI; Hofmann and Montzka 2009; www absorbing short- and long-wave radiation, and by .esrl.noaa.gov/g md/odg i/). indirectly affecting the life cycle, optical properties, and precipitation activity of clouds. | S49 AUGUST 2018 STATE OF THE CLIMATE IN 2017

70 The Copernicus Atmosphere Monitoring Service and Taklimakan/Gobi deserts and seasonal biomass burning in Africa, South America, and Indonesia. (CAMS; http://atmosphere.copernicus.eu) runs a near Overall, the 2017 anomalies of biomass burning real time (NRT) global analysis of aerosols and trace gases. The CAMS project also produced a reanalysis aerosols are consistent with those of tropospheric ozone (Section 2g6), carbon monoxide (Section 2g7), of global aerosols and trace gases that spanned 2003 - and fires (Section 2h3). Seasonal burning was, in gen to 2015 (Flemming et al. 2017) named the CAMS interim reanalysis (CAMSiRA). This reanalysis was eral, less severe than usual in 2017 in the main regions extended to 2017. that are subject to large seasonal fires: Indonesia, the Amazon Basin, and parts of south equatorial Africa. Retrievals of aerosol optical depth (AOD) at 550 Negative anomalies in 2016 and 2017 over Indonesia nm (Remer et al. 2005) from the MODIS instrument and Aqua onboard NASA’s Te r ra (Collection 5) were may be explained by meteorological conditions as well as the government policies regarding land use used as observational constraints from 2003 to 2016. following the El Niño event of 2015 which contributed In 2017, MODIS Collection 6 (Sayer et al. 2014) was assimilated, which can lead to significant differences to severe drought and extreme fires in this region in between 2017 and the previous years in CAMSiRA. 2015. Large but isolated biomass burning events in 2017 are associated with positive anomalies in Chile - Aerosols are produced both by mechanical up lifting over ocean (marine aerosols) and dry areas (January 2017), Siberia (June 2017), and western (mineral dust) and by human activities (industries, Canada—where British Columbia experienced the worst fires in its recent history during July–August traffic, domestic heating, agricultural burning, etc.). 2017 (Plate 2.1ab, Section 7b). Generally, the variability of natural aerosols such as Global maps of the 2003–17 average total AOD dust is large and has high seasonality. Anthropogenic and statistically significant (95% confidence) linear aerosols are more localized but can have significant temporal variability as well. In CAMSiRA, the an trends over the period are shown in Fig. 2.50. The - - thropogenic emissions of black carbon, organic mat highly polluted areas of eastern Asia and India remain prominent features in the total AOD map, as are the ter, and sulfur dioxide were taken from the MACCity inventory (Granier et al. 2011). Open fire emissions dust-producing regions of the Sahara, Arabia, the were provided by the Global Fire Assimilation System Middle East, and the Taklamakan and Gobi deserts (GFAS) inventory (Kaiser et al. 2012) that estimates (Fig. 2.50a). Large AOD values over equatorial Africa are caused by seasonal biomass burning. The linear - fire emissions from MODIS observations of fire ra diative power. These emissions are similar between trend highlights the long-term decrease in anthropo - the NRT analysis and the CAMSiRA. Dust and sea genic aerosols over the eastern U.S., Europe, Japan, salt aerosol emissions are computed dynamically as and parts of southern China, while a significant in - crease occurred over most of the Indian subcontinent, a function of wind speed. - Time series of globally averaged total AOD dur possibly linked to increased industrial activity and, ing 2003–17 (Fig. 2.49) show strong seasonality, with hence, increased emissions in the area. The area of - yearly maxima in March–April and August–Sep decreasing trends in the southern Amazon Basin is associated with reduced deforestation there (Chen tember driven mainly by dust episodes primarily et al. 2013). The decreasing trends over the northern in spring and summer in the Sahara, Middle East, Sahara and western Mediterranean indicate lower frequencies or intensities of dust episodes in these regions or less transport; these were already present in 2016 so are not attributable to model changes. The positive trends over the Southern Ocean may be an artifact of the CAMS interim reanalysis in 2017 and 2016. Radiative forcing resulting from aerosol–radiation (RFari) and aerosol–cloud interactions (RFaci) for the period 2008–17 is shown in Fig. 2.51, as estimated using the methods described in Bellouin et al. (2013) using CAMSiRA data. Negative radiative forcings im - ply a cooling effect of the aerosols on the climate. Due - . 2.49. Global average of total AOD at 550 nm av F ig to a relatively large contribution of anthropogenic eraged over monthly (red) and annual (blue) periods aerosols to total aerosol optical depth, 2017 has been a for 20 03 –17. | S50 AUGUST 2018

71 strong year in terms of aerosol radiative forcing, with the third consecutive increase in RFari, estimated to −2 in 2017, stronger than the −0.55 W be −0.68 W m −2 estimated for 2015. The increase may be linked m to increased biomass-burning aerosols in the tropics. Trends remain statistically fragile, however, because of large uncertainties in the estimates. Absorbing anthropogenic aerosols exert positive RFari over bright surfaces, like the African and Arabian deserts, as shown in the upper panel of Fig. 2.51. RFaci, esti - −2 mated at −0.8 W m in 2017, was comparable to 2015 −2 −2 ). (−0.82 W m ) and 2016 (−0.77 W m —M. Weber, W. Steinbrecht, 4) ozone s tr atospheric R. van der A, S. M. Frith, J. Anderson, M. Coldewey-Egbers, S. Davis, D. Degenstein, V. E. Fioletov, L. Froidevaux, D. Hubert, J. de Laat, C. S. Long, D. Loyola, V. Sofieva, K. Tourpali, C. Roth, R. Wang, and J. D. Wild Throughout nearly the entire Southern Hemi - sphere annual mean total column ozone levels in 2017 were above the mean from the 1998–2008 reference period (Plate 2.1y). In particular, the Antarctic region showed values that were more than 10 DU (Dobson units) above the long-term mean (see also October - mean in Fig. 2.52e). The main cause was the weak po ig . 2.50. (a) Total 550-nm AOD averages for 2003–17. F - lar vortex (stratospheric cyclone) observed in south Note the regional differences, with much greater total ern winter/spring resulting in below-average polar AOD values over parts of northern Africa, the Arabian - Peninsula, southern Asia, and eastern China. (b) Lin ozone losses and a rather small ozone hole in size and − 1 ear trends of total AOD (AOD yr ) for 2003–17. Only depth (see Section 6h). In the second half of 2017 the trends that are statistically significant (95% confidence) quasi-biennial oscillation (QBO) was in the east phase are shown. (easterly f low in the tropical lower stratosphere), which had a global impact on the stratospheric circulation. During the QBO east phase - planetary waves are de f lected toward the pole (SH winter in 2017) and weaken the polar vortex (Baldwin et al. 2011). Associated with these planetary waves is an enhanced meridional - or Brewer–Dobson circu lation transporting more ozone into middle to high latitudes which, in addition to reduced polar losses, con - tributed to the overall SH increase (e.g., Salby 2008; Weber et al. 2011). In the − 2 Northern Hemisphere total F ig . 2.51. Radiative forcing (W m ) in the SW spectrum resulting from (a) ozone was generally near RFari and (c) RFaci from 2008–17. (b,d) The uncertainties of these estimates are shown in gray. | S51 AUGUST 2018 STATE OF THE CLIMATE IN 2017

72 average in 2017 with some regions with slightly lower ozone (Plate 2.1y). Figure 2.52 shows the annual mean total ozone time series from various merged datasets in the tropics, extratropics, and selected months in the polar regions as well as the near-global (60°N–60°S) average. For all time series, the average ozone levels from the 1970s, a time when ozone losses due to ozone-depleting substances were still very small, are also shown. Except for the tropics, total ozone levels have not yet recovered to the values from the 1970s. A recent study indicates that total ozone trends since −1 ) but only the late 1990s are positive (<1% decade reach statistical significance at a few latitudes (Weber et al. 2018). The small increase in global total ozone following the significant decline before the 1990s is regarded as proof that the Montreal Protocol and its Amendments, signed thirty years ago and responsible for phasing out ozone-depleting substances (ODS), works. ODS currently decrease at about one-third of the absolute increasing rate before the 1990s, but the re - cent increase in total column ozone is in comparison smaller than expected from the ODS change. Model studies show that the predicted ozone evolution is consistent in most regions outside the tropics with ODS changes and observed stratospheric ozone and total column observations (Shepherd et al. 2014; Chipperfield et al. 2017). The lack of observed ODS- related changes in tropical total ozone (but observed in climate models with stratospheric chemistry) may be due to a compensation by increases in tropospheric ozone that contribute to the total column (Shepherd et al. 2014). However, observed global tropospheric ozone trends from various studies are highly vari - able and often insignificant (Gaudel et al. 2018 and Figure 26 therein). Ball et al. (2018) suggest, based on an analysis of satellite measurements, that a near-continuous, near-global (< 60° in . 2.52. Time series of annual mean total ozone (DU) in (a)–(d) four zonal bands, ig F both hemispheres) de - and (e) polar (60°–90°) total ozone in Mar (NH; see also Section 5j) and Oct (SH), - cline in lower strato the months when polar ozone losses usually are largest. Data are from WOUDC spheric ozone since 1998 - (World Ozone and Ultraviolet Radiation Data Centre) ground-based measure was compensated by ments combining Brewer, Dobson, SAOZ, and filter spectrometer data (Fioletov et al. 2002, 2008); the BUV/SBUV/SBUV2 V8.6/OMPS merged products from NASA - observed upper strato (MOD V8.6, Frith et al. 2014, 2017) and NOAA (Wild and Long 2018, manuscript spheric increases and in preparation); the GOME/SCIAMACHY/GOME-2 products from University of tropospheric increases, Bremen (Weber et al. 2011; Weatherhead et al. 2017) and GTO from ESA/DLR resulting in rather small (Coldewey-Egbers et al. 2015; Garane et al. 2018). MSR-2 assimilates nearly all total ozone trends. A ozone datasets after corrections with respect to the ground data (van der A et al. - recent chemistry-trans 2015). All six datasets have been bias corrected by subtracting averages from the port model study by reference period 1998–2008 and adding the multiple data mean from the same Chipperfield et al. (2018) period. The horizontal dotted gray lines in each panel show the average ozone level for 1970–79 calculated from the WOUDC data. All data from 2017 are preliminary. shows that the observed | S52 AUGUST 2018

73 lower stratospheric and total column ozone changes Steinbrecht et al. 2017). Figure 2.53 shows that since are mostly explained by variability in atmospheric about 2000, ozone has generally been increasing in - dynamics and is not contradicting our current un the upper stratosphere, ending the previous period - derstanding of stratospheric ozone chemistry related of ozone decline. In 2017, ozone values in the up to ODS changes as otherwise suggested by Ball et al. per stratosphere were below the EESC curve both in the tropical belt and at northern midlatitudes. (2018). In the tropics a continuous decline in total This is somewhat surprising for the easterly phase ozone in the future is predicted by chemistry-climate of the QBO and may in part arise from the decadal models as climate change will enhance tropical up - welling and potentially thin ozone in the lowermost tropical stratosphere, thus increasing UV radiation in the equatorial region (WMO 2014; Chipperfield et al. 2017). While the expected slow recovery of stratospheric ozone has not yet resulted in substantial increases of total column ozone, ozone in the upper stratosphere has been showing clearer signs of increase and recovery over the last 10 to 15 years (WMO 2014; ig . 2.53. Annual mean anomalies of ozone (%; 1998– F 2008 baseline) in the upper stratosphere, near 42-km altitude or 2-hPa pressure for three zonal bands: 35°–60°N (NH), 20°N–20°S (tropics), and 35°–60°S (SH). Colored lines are for long-term records obtained by merging different limb (GOZCARDS, SWOOSH, . 2.54. Mean ozone trends in the upper atmosphere F ig 1 − - SAGE+OSIRIS, SAGE+CCI+OMPS-LP) or nadir view (% decade ) prior to 1997 and after 2000 as derived ing (SBUV, OMPS-NP) satellite instruments. Black from the CCMI REF-C2 models’ simulation (median in line is from merging ground-based ozone records at blue and mean in purple) and satellite data (black line) NDACC stations employing differential absorption in three zonal bands: (a) 35°–60°N (NH), (b) 20°N–20°S lidars, microwave radiometers, and/or Fourier Trans - (tropics), and (c) 35°–60°S (SH). Mean trends were form InfraRed spectrometers (FTIRs). Gray line is - averaged from trends of individual model runs and vari for ground-based Umkehr measurements. See Stein - ous merged datasets shown in Fig. 2.53. The shading brecht et al. (2017) for details on the various datasets. of the models’ mean trend. Same type of shows the 2 σ Orange line gives inverted EESC as a proxy for man- multilinear regression analysis was used to determine made ozone depletion. Ozone data for 2017 are not yet the trends in models and observations. Adapted from complete for all instruments and are still preliminary. LOTUS (2018, SPARC report under review). | S53 AUGUST 2018 STATE OF THE CLIMATE IN 2017

74 minimum of solar activity (e.g., Randel and Wu 1996; 2.55, 2.56c,d), about 40% of the average seasonal cycle Newchurch et al. 2003; WMO 2014). amplitude at 82 hPa in the tropics and 140% of the It is a challenge to accurately attribute observed climatological average difference between these two months. This steep increase in tropical lowermost stratospheric ozone changes, because changes due to SWV during the first half of 2017 and subsequent recovery are expected to be small and thus potentially - return to near-normal values by the end of the year masked by long-term natural variability and mea were observed by both the Aura Microwave Limb surement uncertainty. Substantial efforts, therefore, have gone into improving the available observational Sounder (MLS) satellite instrument (Fig. 2.55) and ozone profile records and into better ways to estimate balloon-borne frost point hygrometer soundings at tropical sites Hilo, Hawaii (20°N), and San José, Costa ozone profile trends and their uncertainties (LOTUS Rica (10°N) (Figs. 2.56c,d). 2018, SPA RC re por t under review). Figure 2.54 shows the resulting updated trend profiles from observa - Variations in cold-point temperatures (CPTs) in the tropical tropopause layer (TTL) on annual and in - tions and chemistry-climate models, both during the phase of ODS-driven ozone decline from the late terannual timescales provide the dominant control on water vapor entering into the lowermost stratosphere 1970s to the late 1990s, and during the beginning - in the tropics by freeze-drying tropospheric air dur recovery phase from 2000 to 2016. Observations are ing its slow ascent through the TTL. Thus, seasonal in generally good agreement with chemistry–climate model simulations. to interannual variability in tropical SWV around As a result of the Montreal Protocol and its 82 hPa is highly correlated with CPT variations. The dramatic swing in tropical lower SWV during 2017 Amendments, ODS have been declining in the is consistent with the substantial 2.5°C increase from stratosphere since the late 1990s. The model simula - November 2016 to May 2017 and subsequent 1.5°C de tions predict that ozone in the upper stratosphere - −1 crease in tropical CPT anomalies over the remainder , due to both should now increase by 2%–3% decade of 2017 (Fig. 2.56d). declining ODS and stratospheric cooling, the latter - Interannual variations in CPTs are partially re caused by increasing greenhouse gases (WMO 2014). lated to interannual variability in the phases of ENSO The right panels of Fig. 2.54 demonstrate that ozone increases are observed in the upper atmosphere after 2000, although they are not statistically significant at all latitudes and altitudes. Nevertheless, the good - agreement between model simulations and observa tions gives confidence that ozone trends in the upper stratosphere are well understood and that ozone in that region is on its continuing (slow) path towards recovery. 5) wat e r s t r at o s p h e r i c va p o r —S. M. Davis, K. H. Rosenlof, D. F. Hurst, H. B. Selkirk, and H. Vömel Stratospheric water vapor (SWV) is a radiatively important gas that can also impact stratospheric ozone chemistry. The second consecutive year of dramatic changes in lower SWV occurred in 2017. Following 2016, during which the tropical mean (15°N–15°S) water vapor anomaly in the lowermost stratosphere (at 82 hPa) dropped from a near record high in January (+0.5 ppm, parts per million mole −1 mol mol m fraction, equivalent to ) to a record low by December (−1 ppm), 2017 anomalies increased to near record high values by midyear. In January 2017 negative (dry) anomalies were ob - served in the tropics and subtropics, in stark contrast to the strong positive (wet) anomalies of June 2017. F . 2.55. Global stratospheric water vapor anomalies ig − 1 From January to June 2017, the tropical SWV anomaly (μmol mol ; 2004–17 base period) centered on 82 hPa in the lower stratosphere increased by 0.9 ppm (Figs. in (a) Jan and (b) Jun 2017 from the Aura MLS. | S54 AUGUST 2018

75 entered the stratosphere in the previous year and was subsequently transported upward. During late 2017, the tropical SWV anomalies at 30 hPa were negative (dry), due to the anomalously cold CPTs and correspondingly dry water vapor anomalies that entered the stratosphere in the latter half of 2016 and beginning of 2017. In general, lowermost SWV anomalies propagate quasi-isentropically from the tropics to the middle latitudes of both hemispheres, as is demonstrated by the “C”-shaped contours in Fig. 2.57b. The early 2017 dry anomaly and the mid-2017 wet anomaly in tropical lower SWV can be seen a few months later in the middle latitudes of each hemisphere. These midlatitude anomalies are also observed by balloon measurements at Lindenberg, Germany (52°N); Boulder, Colorado (40°N); and Lauder, New Zealand (45°S) (Fig. 2.56a,b,e). SWV anomalies over Lauder, New Zealand (Fig. 2.56e) increased during most of 2017, consistent with the poleward transport of the strong wet anomalies in SWV present in the tropics during mid-2017. SWV in the Southern Hemisphere midlatitudes can also ig . 2.56. Lower stratospheric water vapor anomalies F − 1 ) over five balloon-borne frost point (FP) (μmol mol hygrometer stations. Each panel shows the lower stratospheric anomalies of individual FP soundings (black squares) and of monthly zonal averages of MLS retrievals at 82 hPa in the 5° latitude band containing the FP station (red lines). High-resolution FP verti - cal profile data were averaged between 70 and 100 hPa to emulate the MLS averaging kernel for 82 hPa. Each MLS monthly zonal mean was determined from 2000–3000 profiles. Anomalies for MLS and FP data are calculated relative to the 2004–17 period for sites except for Lindenberg (2009–17) and Hilo (2011–17). Tropical CPT anomalies (K) based on the MERRA-2 reanalysis (d, blue curve), which were generally well correlated with the tropical lower SWV anomalies, are the driving force behind the variations in tropical SWV during 2017. and the quasi-biennial oscillation (QBO) in tropical stratospheric winds (Dessler et al. 2014). During 2017, the QBO was in a westerly (warm) phase at 70 hPa, and ENSO was in a neutral state. It is possible F ig . 2.57. (a) Time series of vertical profiles of tropi - that suppressed tropical upwelling due to the QBO cal (15°S –15°N) stratospheric water vapor anomalies − 1 westerly phase led to the warm CPT anomalies and (μmol mol ) and (b) latitudinal distributions of SWV − 1 anomalies (μmol mol ) at 82 hPa. Both are based on positive (wet) SWV anomalies in the tropical lower MLS data. Anomalies are differences from the Aura stratosphere in the first half of 2017. mean 2004–17 water vapor mixing ratios for each Water vapor entering the tropical lowermost month. In panel (b) propagation of tropical lower SWV stratosphere is transported vertically in a quasi- anomalies to higher latitudes in both hemispheres as coherent fashion, forming the well-known “tropical well as the influences of dehydrated air masses from tape recorder” phenomenon (Fig. 2.57a; Mote et the Antarctic polar vortex as they are transported al. 1996). In the tropical middle stratosphere, the towards the SH midlatitudes at the end of each year water vapor abundance is indicative of how much are clearly seen. | S55 AUGUST 2018 STATE OF THE CLIMATE IN 2017

76 be inf luenced by the springtime (October–Novem - Report (TOAR) further discusses the global ground ber) northward transport of air masses that were network including update issues (Schultz et al. 2017; dehydrated within the Antarctic vortex. The weak see Sidebar 2.2). The present update again relies anomalies at high southern latitudes in late 2017 (Fig. mostly on OMI/MLS satellite data. Plate 2.1x shows broad regions of positive anoma - 2.56b) indicate that the Antarctic dehydration in 2017 lies (relative to the 2005–16 average) of up to 1.2 DU was not unusual. Therefore, the positive anomalies - (4%) in tropospheric ozone columns for 2017 in the observed at Lauder in late 2017 are primarily attrib Northern Hemisphere lower midlatitudes and smaller uted to the southward transport of the strong tropical wet anomalies. anomalies of ~1 DU or less elsewhere. Hemispheric and global average tropospheric ozone burdens and 6) t ozone —J. R. Ziemke and O. R. Cooper ropospheric their 95% confidence level precision uncertainties for 2017 were 159 ± 6 Tg for 0°–60°N, 147 ± 8 Tg for - Tropospheric ozone is a surface pollutant, a green 0°–60°S, and 306 ± 7 Tg for 60°N–60°S (Fig. 2.58). house gas, and the dominant source of the hydroxyl radical (OH), which is the troposphere’s primary Each of these 2017 averages represents an increase oxidizing agent. Sources include transport from the from previous years, continuing the long-term posi - tive trend. Linear trends in hemispheric and global stratosphere along with photochemical production burdens from October 2004 through December 2017 from a number of precursor gases such as lightning- −1 in Fig. 2.58 all depict increases of ~0.6% to 0.7% yr . , methane, biogenic hydrocarbons, generated NO x and emissions generated from combustion of fossil Figure 2.59 shows the spatial distribution of tro - pospheric ozone trends on a 5° × 5° grid for October fuels and biomass burning (e.g., Sauvage et al. 2007; 2004 to December 2017. All trends with statistical Leung et al. 2007; Martin et al. 2007; Murray et al. significance depict increases, the strongest of which 2013; Young et al. 2013; Monks et al. 2015; Zhang et al. 2016; Lin et al. 2017). Tropospheric ozone is highly variable from small (urban) to large (hemispheric) scales due to variations in dynamical transport and photochemical production (i.e., heterogeneity of precursor gas emissions and sunlight) and sinks - including loss mechanisms such as through HO pho x tochemistry and through surface deposition (IPCC 2014). Transport phenomena that drive large-scale variability include the El Niño–Southern Oscillation (e.g., Chandra et al. 1998, 2009; Sudo and Takahashi 2001; Doherty et al. 2006; Koumoutsaris et al. 2008) and the Madden–Julian Oscillation (e.g., Sun et al. 2014; Ziemke et al. 2015). Relatively short lifetimes for ozone and ozone precursors and short-term variability of transport including convection drives much of the variability of tropospheric ozone on short . 2.58. Monthly averages of OMI/MLS tropospheric F ig ozone burdens (Tg) from Oct 2004 through Dec 2017. timescales including day-to-day changes. Variability Top curve (solid black line) shows 60°S –60°N monthly from daily to interannual timescales adds challenges averages with 12-mo running means (dashed black to quantifying decadal trends at hemispheric and line). Bottom two curves show monthly averages global scales (e.g., Neu et al. 2014; M. Lin et al. 2014; and running means for the NH (red) and SH (blue). Barnes et al. 2016). Slopes of linear fits to the data are presented with State of The tropospheric ozone summary in the - their 2-sigma uncertainties. All three trends are sta the Climate in 2012 was based on measurements by tistically significant at the 95% confidence level. Prior to our analyses, the data were evaluated for potential ground- and satellite-based instruments (Cooper and offset and drift by comparison with globally distributed Ziemke 2013). Since then the reports have primarily - ozonesonde profiles and OMI convective cloud differ Aura Ozone Monitoring Instrument/ relied on the ential (CCD) measurements (Ziemke et al. 1997). A Microwave Limb Sounder (OMI/MLS) satellite mea - 1 − small drift of about +0.5 DU decade was found and surements (Ziemke et al. 2006, 2015) because of insuf - an appropriate correction was applied to the OMI/MLS ficient updates of global ground-based observations data. OMI/MLS tropospheric ozone was also adjusted (Cooper and Ziemke 2014, 2015; Ziemke and Cooper everywhere by +2 DU to correct for mean offset rela - tive to the ozonesondes. 2016, 2017). The Tropospheric Ozone Assessment | S56 AUGUST 2018

77 above the North Atlantic Ocean, the equatorial Africa and Atlantic/Indian Oceans regions, and the South - ern Hemisphere extratropics. As noted above, updating global surface ozone measurements annually is difficult because most ground stations do not provide quality-assured final data soon enough for the timing of this report. How - ever, there are three remote monitoring sites with rap - idly updated data: 1) the high-elevation Mauna Loa Observatory (MLO), Hawaii (19.5°N, 155.6°W, 3397 m asl); 2) South Pole Observatory (SPO), Antarctica F . 2.59. Linear trends in OMI/MLS tropospheric ig (90°S, 59°E; 2840 m asl); and 3) Utqiaġvik (Barrow), − 1 column ozone (DU decade ) on a 5° × 5° grid for Alaska (71.3°N, 156.6°W; 11 m asl). Continuous Oct 2004–Dec 2017. Asterisks denote statistically UV-based measurements of ozone began at MLO significant trends at the 95% confidence level. Note in September 1973, at SPO in January 1975, and at that trends were calculated using a multivariate linear regression model (Ziemke et al. 1998, and references Barrow in March 1973. Reliable ozone observations therein) that included a seasonal cycle fit and the based on the Regener automatic wet-chemical method Niño-3.4 index as an ENSO proxy; trend uncertainties are also available at SPO for 1961–63 (Oltmans and included autoregressive adjustment via Weatherhead Komhyr 1976), and at MLO for 1957–59 (Price and et al. (1998). Pales 1963). These time series, the world’s longest at −1 −1 [~+3.3 DU decade (+1. 0 5% y r remote locations, are reported in Fig. 2.60 as monthly )] are located above India, Southeast Asia, and extend eastward across the medians, based on all 24 hours of the day at SPO and North Pacific Ocean. These upward trends are con Barrow, but the MLO data are restricted to nighttime - sistent with model estimates based on strengthening values when local winds are downslope, ensuring that emissions of ozone precursors from Southeast, East, the observations are representative of the lower free troposphere. The limited data at MLO and SPO from and South Asia, primarily due to fossil fuel combus - the 1950s and 1960s indicate that ozone levels at these tion (Zhang et al. 2016; Lin et al. 2017). The model remote high-elevation sites were similar during the simulations indicate that ozone produced in these mid-20th century despite being located in different densely populated areas is transported eastward in the free troposphere over the North Pacific Ocean as hemispheres. Ozone at SPO has changed little since suggested in Fig. 2.59. Positive trends are also found the 1960s with no significant trend. In contrast, F ig . 2.60. Monthly median surface ozone at Utqia ġ vik (Barrow), Alaska (Mar 1973–Dec 2017; green) and South Pole (Jan 1975–Dec 2017; black) using data from all hours of the day. Additional data from South Pole are shown for the early 1960s. Also shown are nighttime monthly median ozone values at the Mauna Loa Observatory (MLO) calculated with all available data for months with at least 50% data availability, Sep 1973–Dec 2017 (blue), with early observations from the late 1950s. Monthly median values associated with dry air masses (orange) at MLO are also included (dew point less than the climatological monthly 40th percentile, and a sample size of at least 24 individual hourly nighttime observations). Trends (solid straight lines) are based on least-squares linear regression fit through the monthly values (1970s–2017), and reported with 95% confidence intervals and p -values. MLO and South Pole trend lines are extrapolated back in time to the late 1950s (dashed lines). | S57 AUGUST 2018 STATE OF THE CLIMATE IN 2017

78 SIDEBAR 2.2: THE TROPOSPHERIC OZONE ASSESSMENT REPORT (TOAR) —O. R. COOPER Recognizing the need for a comprehensive tropospheric the TOAR database these ozone metrics are freely accessible ozone survey and the challenges associated with gathering for research on the global-scale impact of ozone on human and processing ozone observations from thousands of sites health, crop/ecosystem productivity, and climate. All ozone worldwide, the International Global Atmospheric Chemistry data submitted to the database have undergone quality control - (IGAC) Project developed the Tropospheric Ozone Assess procedures by the agencies or research groups that made the ment Report (TOAR): Global metrics for climate change, observations. The site locations are then cross-referenced human health, and crop/ecosystem research, released in with global gridded datasets of human population, satellite- ), a bottom-up October 2017. Initiated in 2014, TOAR’s mission is to provide detected tropospheric nitrogen dioxide (NO 2 emissions inventory, satellite-detected night-time lights NO the research community with an up-to-date scientific assess - x of the world, and land cover so that all sites can be objectively ment of tropospheric ozone’s global distribution and trends queried to determine if they meet predetermined criteria for from the surface to the tropopause. TOAR’s primary goals are: urban or rural classifications. The database is publicly available (1) produce the first tropospheric ozone assessment report and the ozone metrics can be downloaded from: https://doi based on all available surface observations, the peer-reviewed .pangaea.de/10.1594/PANGAEA.876108. literature, and new analyses; and (2) generate easily accessible The particular ozone metrics available from the database and documented ozone exposure metrics at thousands of were chosen for their relevance to research (Lefohn et al. measurement sites around the world. TOAR is an international 2018) related to human health (Fleming and Doherty et al. collaborative effort with participation from over 230 scientists 2018), vegetation (Mills et al. 2018, manuscript submitted to and air quality experts from 36 nations representing research ), and climate (Gaudel et al. 2018). The metrics are Elementa on all seven continents. also being used to evaluate global atmospheric chemistry Monitoring global trends of long-lived greenhouse gases such models (Young et al. 2018), to assess long-term global ozone as carbon dioxide and methane is relatively straightforward as trends from the early 20th century to the present (Tarasick their spatial and temporal variability is limited and relatively et al. 2018, manuscript submitted to Elementa ), and to develop few measurement sites are required to demonstrate global- new statistical methods for quantifying regional ozone trends scale changes. Quantification of global ozone trends is much (Chang et al. 2017). more difficult due to ozone’s short lifetime (days to weeks) An illustration of the database’s capabilities is provided in and multiple sources and sinks that have heterogeneous spatial Fig. SB2.3 which shows the warm season (April–September distributions and seasonal cycles. While over 5000 surface in the Northern Hemisphere and October–March in the ozone monitoring sites are presently established worldwide, their distribution is uneven with high densities in North America, Europe, and East Asia, and few or no sites in South Asia, the Middle East, Central Asia, Africa, and South and Central America. Monitoring is also limited across the oceans and the polar regions. Another barrier to producing a global survey of surface ozone trends is the logistical problem of gathering the data from dozens of air quality agencies and research groups across many nations, all with different data formats. To produce a wide range of ozone metrics at thousands of surface sites worldwide, TOAR built the world’s largest database 1 − of surface hourly ozone observa - F ig . SB2.3. 98th percentile ozone (nmol mol equivalent to ppb) at all avail - able (4792) surface sites for the 2010–14 warm season (Apr–Sep in the NH, and tions (Schultz et al. 2017). Through Oct–Mar in the SH). | S58 AUGUST 2018

79 Southern Hemisphere) surface ozone 98th percentile value at all available stations, averaged across the period 2010–14. The data quickly reveal that the most ex - treme ozone events are found in Southern California, Mexico City, northern Italy, northern India, eastern China, South Korea, and Japan. Ozone monitoring is sparse in the Southern Hemisphere, but in general concentrations are much lower. Figure SB2.4 depicts the trends of the 98th percen - tile surface ozone at all available stations showing widespread decreases across North America and much of Europe in response to emission controls of ozone pre - cursor gases (oxides of nitrogen, carbon monoxide, volatile organic compounds). In contrast, high ozone events increased in Hong Kong, South Korea, and parts of western Japan due to broad, regional scale ozone precursor emission increases. F . SB2.4. Trends of the O ig 98th percentile at the sites shown in Fig.SB 2.3, 3 Further information on -values on the linear trend for p during 2000–14. Vector colors indicate the TOAR can be found on the IGAC each site: blues indicate negative trends, oranges indicate positive trends, and webpage: www.igacproject.org green indicates weak or no trend; lower -values have greater color saturation. p /activities/TOAR/ - ozone at MLO has increased significantly at the rate that the site is inf luenced by ozone increases in up −1 −1 , resulting in an overall yr of 0.15 ± 0.05 nmol mol wind regions to the west and northwest, most likely −1 since 1973, or 17%. MLO Asia where limited in situ observations have shown increase of 6.5 nmol mol general ozone increases over the past two decades at experiences high interannual ozone variability due the surface (Cooper et al. 2014; Ma et al. 2016; Sun to its location in the transition region between tropi - et al. 2016; Xu et al. 2016; T. Wang et al. 2017) and in cal and extratropical air masses. The ozone trend in the dry air masses, which tend to originate at higher the free troposphere (Zhang et al. 2016). altitudes and latitudes to the west and northwest of MLO, while moist air masses tend to come from the 7) c monoxide arbon — J. Flemming and A. Inness east at lower latitudes and altitudes (Harris and Kahl Carbon monoxide (CO) plays a significant role as 1990; Oltmans et al. 2006; M. Lin et al. 2014). Ozone a chemical precursor in determining the abundance observations at MLO were divided into dry (<40th ), through of climate forcing gases like methane (CH 4 percentile) and moist (>60th percentile) air masses us - hydroxyl radical (OH) chemistry and tropospheric ing observed dew point temperatures and a long-term ozone (Hartmann et al. 2013). CO is therefore re - garded as an indirect climate forcing agent. Sources climatology. The trend in the dry air masses is 50% greater compared to the trend using all air masses (9.9 of CO include incomplete fossil fuel and biomass combustion and in situ production via the oxidation ppbv total increase since 1974, or 23%), which implies | S59 AUGUST 2018 STATE OF THE CLIMATE IN 2017

80 of CH and other organic trace gases. Combustion than in the periods before and after this year, and 4 and chemical in situ sources typically produce similar further investigation is necessary to determine the cause. The large increase in the global CO burden in amounts of CO each year. the second half of 2015 and the first half of 2016 was The Copernicus Atmosphere Monitoring Service (CAMS) produced a retrospective analysis of CO, caused by intensive biomass burning in Indonesia aerosols, and ozone for the period 2003–15 by assimi - in October 2015 (Huijnen et al. 2016). A piecewise calculation of linear trends for the periods 2003–07, lating satellite retrievals of atmospheric composition −1 with the ECMWF model (Flemming et al. 2017). This 2008, and 2009–17 revealed trends of −3.0 Tg yr −1 −1 −1 ), −20.0 Tg yr (−5.0% yr ), and −1.1 Tg dataset has been extended to the end of 2017 and is (− 0 .7 % y r −1 −1 used here. Version 5 total column retrievals of CO ), respectively. This means that a much yr (−0.3% yr stronger reduction of the global CO burden occurred from the MOPITT instrument (Deeter et al. 2013) in 2008 and in the period 2003–07 than after 2009. were assimilated from January 2003 until the end of February 2017. From March 2017 onwards MOPITT 2017 was the year with the lowest CO burden in version 7 data were used because the older version the CAMS interim reanalysis. The annual mean of 2017 was below the median of the annual means for was discontinued. The anthropogenic emissions the 2003–17 period almost everywhere, mostly in were taken from the MACCity inventory (Granier et the range from 0 to −10% (Fig. 2.62). This indicates al. 2011) that accounts for projected emission trends that no regional biomass burning event in 2017 had according to the representative concentration path - ways (RCP) 8.5 scenario (Riahi et al. 2011). Biomass a global impact on annual regional burdens. In general the relative decrease was more pro - burning emissions were taken from the Global Fire Assimilation System (v1.2, Kaiser et al. 2012, also nounced in the mid- and high latitudes of the North - ern Hemisphere than in the Southern Hemisphere. Section 2h3). The global tropospheric CO concentrations have The largest annual minima occurred over Indonesia, −1 - in the last decade accord decreased by about 1% yr where the annual CO burden was up to 20% lower ing to studies based on MOPITT and other observa - than the median values after the extreme fires of tions (Worden et al. 2013; Yin et al. 2015; Flemming et al. 2017; Gaubert et al. 2017). Model simulations (Flemming et al. 2017; Gaubert et al. 2017) result in weaker negative trends than the observation based estimates. This could point to an underestimation of anthropogenic emissions trends or to unaccounted chemical feedback in the CO-OH-O system of -CH 4 3 the models (Gaubert et al. 2017). The time series of the global CO burden obtained from the CAMS interim reanalysis (Fig. 2.61) shows an average reduction from 410 Tg in 2003 to 358 Tg in 2017. This is equivalent to a linear trend of −3.3 Tg −1 −1 yr (− 0 . 8% y r ) over the whole period. However, the global burden decreased more rapidly during 2008 F . 2.61. Time series (black solid line for 2003–16, ig red for 2017) of monthly global CO burdens (Tg) from . 2.62. Total column CO anomalies (%) for (a) Jan– ig F the CAMS interim reanalysis and a piecewise linear Jun 2017 and (b) Jul–Dec 2017 with respect to 2003–17 trend (dotted line) for the periods 2003–07, 2008, and median from the CAMS interim reanalysis. 20 09 –17. | S60 AUGUST 2018

81 2015. Also, fire activity in Central Africa was overall - for the negative anomalies reaching (or locally ex ceeding) −30 % in the visible and about −10 % in the lower than in previous years. The lower CO in the first half of 2017 (Fig. 2.62a) was the primary reason near-infrared domain. The fast decline of the snow for the negative annual anomalies. Intensive fires in cover extent and duration as early as February (Sec - tion 2c2) may be due to unusually warm and relatively Chile in January had only a localized effect on the dry conditions over western Europe from early spring CO burden in the first quarter of 2017. to June (https://climate.copernicus.eu/resources The more active fires occurred predominantly /data-analysis/average-surface-air-temperature in the second half of 2017 (Fig. 2.62b; Section 2h3). -analysis/monthly-maps/) Large boreal fires in Canada (British Columbia, The Northwest Territories) and Central Siberia increased A few snow-free regions show positive anomalies, the CO burden in the high northern latitudes in the especially in the visible domain, in northeast Brazil, third quarter of 2017 by over 10% and locally up to from southeast Somalia and Kenya to northern Tanza - nia, Anatolia, and Nigeria, and in some localized spots 20% with respect to the decadal median (not shown). around the Caspian Sea. These are generally associated In September and the final quarter of 2017, increased activity during the fire seasons in Brazil and in east - with less favorable vegetation growing conditions ern and central Africa caused the CO burden to rise compared to previous years (Section 2h2), although contamination of the albedo retrievals by clouds and up to 10% over the long-term seasonal mean in the aerosol load (especially in intertropical regions) may affected regions and in the adjacent outf low regions also induce some artifacts. Many snow-free regions over the central Atlantic. exhibit noticeable and spatially consistent negative - h. Land surface properties anomalies, in particular in the visible domain, and es pecially pronounced (up to 30%) across eastern China, dynamics —B. Pinty and 1) l albedo and surface N. Gobron Southeast Asia, parts of India, much of southern and The land surface albedo represents the fraction of central Africa, parts of Australia, and much of Argenti - solar radiation scattered backward by land surfaces. na. Consistent warmer-than-usual conditions persisted In the presence of vegetation, surface albedo results over most of these regions, sometimes associated with from complex nonlinear radiation transfer processes below-normal precipitation. A significant fraction of determining the amount of radiation that is scattered these variations is attributable to vegetation dynamics - by the vegetation and its background, transmitted (Pinty et al. 2011a,b) over these regions where vegeta tion is sensitive to stress from ambient conditions and, through the vegetation layer, or absorbed by the veg - in particular, water availability. Although weaker in the etation layer and its background (Pinty 2012). near-infrared domain, these negative anomalies are, The geographical distributions of normalized anomalies in visible and near-infrared surface albedo in some instances, spectrally correlated, for example, for 2017 calculated for a 2003–17 base period [for over India and northeast Brazil. The amplitude of these positive and negative anomalies often changes which two MODIS sensors are available (Schaaf et al. 2002)] are shown in Plate 2.1ac, ad, respectively. Note with seasons. The situation is thus globally analogous to 2016, with above-average temperatures and a few that MODIS collection 6 albedo products are used here. Mid- and high latitude regions of the Northern extreme precipitation and drought events (e.g., across Hemisphere are characterized by both positive (blue) southern Europe) occurring across the world. and negative (brown) anomalies mainly as a conse - Analysis of the zonally averaged albedo anoma - lies in the visible (Fig. 2.63a) and near-infrared (Fig. quence of interannual variations in snow cover (see 2.63b) broadband spectral domains indicates large Section 2c1), amount, and duration in winter and interannual variations related to the occurrence of spring seasons. snow events in winter and spring at mid- and high The positive anomalies especially in the visible northern latitudes as well as to vegetation conditions range over the U.S. Northwest and High Plains, during the spring and summer periods. Negative southwest and eastern Canada, Scandinavia, and anomalies are noticeable between 20° and 45°S in northern Russia are probably associated with above- 2017, featuring a deviation from average conditions average snow cover and extent in spring with the occurrence of snow storms in some of these regions. mainly over Latin America, southern Africa, and Australia. Consistent negative anomalies in the visible Below-average snow cover extent across most of Eu - rope, Turkey, Iran, southern Russia, and in parts of domain are discernible across midlatitude regions in the Northern Hemisphere in 2017. the U.S. Northern Plains and Rockies extending into the southern Canadian Prairies may be responsible | S61 AUGUST 2018 STATE OF THE CLIMATE IN 2017

82 ing to missing data. The year 2017 is characterized by a trend of the negative anomalies toward aver - age conditions in the visible domain that is driven by the dominant contributions from the Northern Hemisphere regions. These figures also indicate spectrally correlated multiannual variations during 2003–17 with positively biased values in the visible at the beginning of this period. errestrial ve —N. Gobron activity 2) t Getation Terrestrial photosynthesis activity is inferred from space on the basis of one land essential climate - variable (ECV) as defined by GCOS (2016): the frac tion of absorbed photosynthetically active radiation (FAPAR). The 2017 analysis has merged 20 years of global FAPAR products retrieved from three passive optical sensors at medium spatial scale from 1998 to 2017 (Gobron et. al. 2010; Pinty et al. 2011a,b; Gobron and Robustelli 2013). Note that Collection 6 MODIS albedo (Section 2h1) was used in this year’s report. Plate 2.1ae displays the annual FAPAR anomalies at global scale for which brown (blue) color indicates negative (positive) values. Large geographical varia - tions in vegetated surface conditions were present at the global scale. Negative and positive anomalies indicate less and more photosynthetic activities in green live vegetation. F . 2.63. Zonally averaged albedo anomalies (%; ig The most negative anomaly events (not favorable 2003–17 base period) in the (a) visible and (b) near- for vegetation) took place over eastern Brazil, Somalia, infrared broadband. and Kenya followed by the weakest negative ones in the western part of Russia. The major positive events occurred in the eastern part of China and Botswana and the weakest appeared over Coahuila (northern state of Mexico), India, and the Rio Negro region in Argentina. The strong negative FAPAR anomalies over eastern Brazil were mainly due to severe droughts occurring at the start of the year that impacted the annual results. Over Somalia the persistent precipita - tion deficit extended both the geographical area and its negative level in terrestrial activities, meaning that vegetation photosynthesis declined rapidly at the beginning of the year. The vegetation activities in northwestern Russia declined during spring, possibly due to heavy snow events. Terrestrial photosynthesis ig . 2.64. Global albedo anomalies (%; 2003–17 base pe F - activities continued to proliferate over the eastern riod) in the (a) visible and (b) near-infrared broadband. part of China as stronger positive FAPAR anomaly events were observed as both higher temperatures and heavy precipitation were favorable to vegetation - The amplitude of the globally and hemispheri growth in 2017. FAPAR anomalies were also positive cally averaged normalized anomalies resulting from in 2017 over Botswana as in 2014, meaning that after a 12-month running mean (Fig. 2.64) is within ±5% (3%) in the visible (near-infrared) domain. The droughts in 2015–16, sufficient precipitation helped vegetation recover. - anomalies are not estimated over Antarctica ow | S62 AUGUST 2018

83 ig . 2.66. Global, NH, and SH FAPAR anomalies from F 1998–2017, plotted in black, blue, and red, respectively. Dotted lines denote each monthly period; solid lines F . 2.65. Zonally averaged FAPAR anomalies from ig indicate the 6-mo running averaged mean. (Sources: 1998 –2017. SeaWiFS, MERIS, and MODIS sensors.) Spring 2017 revealed negative anomalies at higher Positive anomalies occurred also over smaller latitudes (~ 60°N), as was the case in summer around regions such as over Coahuila (Mexico) and the Rio 20°N and 20°S. Around 30°S recurrent and strong Negro region in Argentina; these have occurred each year since 2015 and may correspond to high spring positive anomalies have occurred since 2014, contra - temperatures. Australia was found to have positive dicting the strong negative anomalies from 2005–10. As shown in Fig. 2.66, there was a strong reversal FAPAR anomalies over several local regions. between anomalies over the Northern and Southern Figure 2.65 displays the longitudinal average Hemispheres during the past 20 years. The FAPAR - anomalies from 1998 to 2017. Strong seasonal in terhemispheric variations are depicted with mainly anomaly over the Southern Hemisphere in 2017 returned to a positive level (last evident in 2000) positive anomalies after 2014 over 20°N and negative anomalies from 2002–10 in the south latitudes. while it has continued to increase over the Northern Hemisphere since the 2008–10 minimum. PHENOLOGY OF TERRESTRIAL AND FRESHWATER SIDEBAR 2.3: —D. L. HEMMING, R. ABERNETHY, C. ARMITAGE, K. BOLMGREN, PRIMARY PRODUCERS R. MYNENI, T. PARK, A. D. RICHARDSON, T. RUTISHÄUSER, T. H. SPARKS, AND S. J. THACKERAY. Phenology is the study of recurring events in nature - Shifts in the growing season, for example, are more tan and their relationships with climate. The word derives gible and more readily conveyed to the general public than ō “appear” and logos “reason”, em - from the Greek phaín seemingly small increases in mean annual temperature. phasizing the focus on observing events and understand - Phenological monitoring thus plays an important role in ing why they occur (Demarée and Rutishauser 2009). understanding how our planet is changing. Phenological recording has a history that dates back Here, we describe just a fraction of the phenological many centuries (Linnaeus 1753; Aono and Kazui 2008). information currently available, highlighting northern More recently, advances in monitoring technologies have hemisphere records of phenology of primary producers enabled automated and remotely sensed observations, across a range of spatial and temporal scales. complemented by increasing citizen science participation Ground-based observations in monitoring efforts. Phenological information can also be derived from widespread environmental monitoring Long-term phenology monitoring network, stations around the globe. Deutscher Wetterdienst (DWD) maintains Germany: Phenological records clearly demonstrate the biologi - a dense national phenological observation network and cal effects of year-to-year variability in climate, as well as database (www.dwd.de/phaenologie/). Plant phenological longer-term trends associated with environmental change. | S63 AUGUST 2018 STATE OF THE CLIMATE IN 2017

84 CONT. PHENOLOGY OF TERRESTRIAL AND FRESHWATER SIDEBAR 2.3: —D. L. HEMMING, R. ABERNETHY, C. ARMITAGE, K. BOLMGREN, PRIMARY PRODUCERS R. MYNENI, T. PARK, A. D. RICHARDSON, T. RUTISHÄUSER, T. H. SPARKS, AND S. J. THACKERAY. nological observations, supported by the Woodland Trust (www .woodlandtrust.org.uk /visiting-woods/natures-calendar/). Currently, over 4000 members of - the public contribute regular obser vations, and the database includes over 2.7 million records, dating from - 1695. Early observations of “Indica tors of Spring” were made from 1736 to 1797 by Robert Marsham in Norfolk and continued by his descendants until 1958 (Sparks and - Lines 2008). In 1875, a national net work was launched by the (Royal) Meteorological Society, which ran until 1948, recording flowering, appearance of bird and insect spe - cies, and publishing unusual events and their climate relationships (Clark 1936). In 1998, the Centre - for Ecology and Hydrology resur rected this network, and in 2000, F ig .SB2.5. Time series of phenological changes in primary producers was joined by the Woodland Trust from records in Germany and UK, showing timing (by ordinal date) of (a) to promote phenology to a wider leaf unfolding of tree species in Germany from DWD national network: audience (Sparks et al. 1998; Sparks L, (b) budburst of 4 common tree species Pedunculate oak – Quercus robur and Smithers 2002). Figure SB2.3b L. Gaertn; horse Alnus glutinosa in U.K. from Nature’s Calendar: Alder - highlights the timing of budburst for Fagus Aesculus hippocastinum chestnut - L.; pedunculate oak; and beech - four tree species in this record. As L, and (c) long-term phenological changes in spring phytoplankton sylvatica a growth, indicated by the seasonal timing of maximum spring chlorophyll- with other plant species, budburst concentrations. Original chlorophyll data collected from the north basin of is significantly related to spring Windermere by the Centre for Ecology & Hydrology and the Freshwater average temperature (Online Fig. Biological Association, U.K. S2.21), with a 1°C rise in March or records dating to 1951, some available since 1925, are April temperature associated with earlier budburst of 3.5 openly accessible via the online archive (Kaspar et al. to 4.8 days, depending on species and region (Abernethy 2014). Currently, about 1100 observers contribute to the et al. 2017). database, recording phenological events in cover crops, Seasonal activity of primary Windermere, UK: - wild plants, and fruit trees. The data have many applica - producers is monitored in marine and freshwater en tions, including advice on current growing season for vironments. For example, at Windermere—England’s agricultural activities, pollen forecasts, and environmental largest lake— fortnightly measurements of chlorophyll- a change research. Figure SB2.5a highlights the record of concentrations, a proxy for primary producer biomass, leaf unfolding of pedunculate oak ( L.), which Quercus robur have been recorded since the 1960s. These data show a has advanced by about 10 days over the last 50 years. long-term shift toward earlier spring algal blooms (Fig. This species is referred to as an “indicator species”, and, 2.3c), which is correlated with both increasing spring due to its strong dependence on spring temperature, leaf water temperatures and changes in nutrient availability unfolding is used to mark the beginning of “full spring”. (Thackeray et al. 2013). Hence, large-scale climatic drivers Nature’s Calendar is a Nature’s Calendar, UK: act alongside more localized lake-specific influences to coordinated national “citizen science” network of phe - bring about phenological changes in this system. | S64 AUGUST 2018

85 Pan European Phenology (PEP) project - : The PEP project promotes and fa cilitates phenological research, education and environmental monitoring across Europe. It maintains the Pan European Phenology (PEP) database (www.pep725.eu), which provides unrestricted data access for science and education. This currently includes 12 million records, with contributions since 1868 from 32 European partners for 46 growing stages species and cultivars (Templ et and 265 plant al. 2018). Remote sensing Remote sensing provides some of the clearest records of regional, hemispheric, and global phenological changes by linking radiance measurements to photosynthetic indicators of terrestrial and marine primary producers (Park et al. 2016; Sapiano et al. 2012). Digi Near-surface remote sensing: - tal camera networks observe “ the rhythm of ” the seasons , from the tropics to the tundra. PhenoCam (http://phenocam.sr.unh.edu) is a collaborative network of over 400 cameras, most at research sites in the United States. Measures of canopy greenness (Richardson et al. 2018a) derived from camera imagery can be used to track vegetation activity and identify the start and end of season. At one temperate deciduous forest (Richardson et al. 2007), the 2017 growing season was markedly shorter than the decadal average because of late onset and early senescence (Fig. SB2.6a). At the same site, the seasonal cycle of canopy greenness follows that of gross primary productivity - . SB2.6. Phenocam records of canopy greenness (green chro ig F (GPP) estimated from eddy covariance mea - matic coordinate, GCC) and GPP from two deciduous forest sites in fluxes, confirming the role surements of CO 2 the U.S.: Bartlett Experimental Forest, NH, and Duke Forest, NC, of phenology in regulating ecosystem carbon showing: (a) Time series of day of year of “Greenup”, “Greendown” fixation (Richardson et al. 2010; Fig. SB2.6c). and (b) number of days of “Green canopy duration” at Bartlett, (c) The difference between this cooler forest comparison of seasonality of GCC and GPP (estimated from flux and a warmer forest (Fig. SB2.6d) illustrates measurements) at Bartlett during 2017, and (d) seasonality in GCC the role of climate in controlling phenology. between Bartlett (mean annual temperature = 6.6°C) and a warmer These data can therefore help improve under - site, Duke (mean annual temperature = 15°C) during 2017. Photos show both sites in Jul 2017. standing of relationships between phenology, - ecosystem processes, and environmental driv | S65 AUGUST 2018 STATE OF THE CLIMATE IN 2017

86 CONT. PHENOLOGY OF TERRESTRIAL AND FRESHWATER SIDEBAR 2.3: —D. L. HEMMING, R. ABERNETHY, C. ARMITAGE, K. BOLMGREN, PRIMARY PRODUCERS R. MYNENI, T. PARK, A. D. RICHARDSON, T. RUTISHÄUSER, T. H. SPARKS, AND S. J. THACKERAY. ers. Furthermore, pheno - cam data are valuable for ground truthing satellite observations, as they are continuous in time and require minimal correction - or screening for atmo spheric effects Richardson et al. 2018b). Satellite remote sensing: Satellite-de - rived phenology indices provide useful regional - to global-scale monitor ing for phenology studies (Zhang et al. 2003). Figure SB2.7 highlights Northern Hemisphere land surface phenology indices during 2000–17, derived from radiance observations from the MODIS sensor. It shows a widespread and continued earlier start- of-season (−1.5 days) and later end-of-season (+1.3 days) over this period (Park et al. 2016). In 2017, the start-of-season reveals a dramatic spatial contrast ig . SB2.7. MODIS-derived NH (>45°N) land surface phenology, showing 2017 F between North America anomaly (days), relative to 2000–17 average, for (a) start-of-season, (b) end-of- - and Eurasia. Northeast season, and (c) hemispheric average day of year of the start and end of season ern Europe and western for 2000–17. Russia showed a striking delay (+6.0 days) associated with an anomalous spring to monitor phenological changes across wide spatial cold spell (−2.4°C), whereas North America showed a scales—from global biomes to microscopic organisms. widespread earlier start-of-season (−5.1 days), due to Furthermore, phenology records exist that span multiple warmer than average spring temperatures (+0.5°C). The decades, even centuries, and these provide valuable ar - end-of-season across Eurasia was generally later than chives of long-term environmental change. There is now a average (+2.3 days), but earlier (−3.6 days) over southern - fundamental need for integrated analyses of multiple phe European temperate zones. nology and climate observations to help understand, and Many phenological events provide clear indicators of prepare for, the future impacts of climate variability and the influence of climate on our environment and natural change on environmental systems, and routine monitoring resources. Current observations apply diverse techniques to capture important changes as they occur. | S66 AUGUST 2018

87 3) b burnin G —J. W. Kaiser and G. R. van der Werf iomass In 2017, low fire activity led to the lowest global pyrogenic emissions since at least 2003 and probably The first vegetation fires, a.k.a. biomass burning, occurred shortly after the first land plants evolved. since the start of GFED in 1997. Emissions were 15% They have since become an integral part of many below the 2003–16 average (Table 2.7; Plate 2.1af). The year with lowest emissions prior to 2017 was 2013; natural and cultivated ecosystems and are largely those two years were relatively close in magnitude, modulated by climate. Conversely, fires are a ma - especially when considering the substantial uncer - jor source of climate-forcing atmospheric aerosols tainty associated with these estimates. The negative and trace gases. Today, human activity also exerts a strong inf luence on fire occurrence through land regional anomalies were particularly pronounced in cover change by providing a large number of ignitions tropical Asia, where high rainfall rates, among other things, led to emissions that were only about 5% of and by active fire suppression. Fires have substantial interannual variability, which mostly originates from those reached during the El Niño episode in 2015. Strong negative anomalies also occurred in Indochina the boreal region and the tropical deforestation zone. and in southern Siberia. The extent of vegetation fires is traditionally quanti - Stronger-than-usual fire activity occurred in fied in terms of burned area, which is around 500 North America and Europe, with anomalies of +36% million hectares worldwide each year (Giglio et al. 2013; Randerson et al. 2012). and +22%, respectively. The time series for North America (Fig. 2.68) shows that four out of the last Burned area and the thermal radiation released five fire seasons were exceptionally intense. European by active fires have been quantified on a global scale fire emissions were dominated by an unusually long with satellites since the late 1990s. The Global Fire - Emissions Database (GFED) estimates emissions burning season in Portugal and in Galicia in north - since 1997 based on burned area and fuel consump western Spain. tion (van der Werf et al. 2017). Satellite-observed fire radiative power is used by the Global Fire Assimila - tion System (GFAS) to estimate emissions since 2003 and in near-real time (Kaiser et al. 2012). GFAS is calibrated to partly match GFED. Here, an updated GFAS version (Kaiser et al. 2017) is used; it resolves the subdaily variability and uses MODIS Collection 6 products (Giglio et al. 2016) for the entire time period. The absolute values have been homogenized with earlier GFAS and GFED versions by removing a global average bias of −14%. The combined use of GFAS (2003–17) and GFED (1997–2016) indicates −1 (Fig. 2.67). that fire emissions were around 2 Pg C yr F . 2.68. Time series of fire activity during 1997–2017 ig 1 − ) for in terms of carbon consumption (Tg month North America. . 2.67. Global map of fire activity in 2017 in terms of F ig 1 − 2 − yr ). (GFASv1.4) carbon consumption (g C m | S67 AUGUST 2018 STATE OF THE CLIMATE IN 2017

88 Table 2.7. Annual continental-scale biomass burning budgets in terms of carbon emission 1 − (Tg C yr ) from GFASv1.4. 2017 2003 –16 Mean Value Anomaly 1 − Value Quantity in Tg C yr (%) (Range) 1973 1683 Global 290 ( − 15%) − (1690 –2272) 30°–75°N 84 113 +30 (+36%) North America 170°–30°W ( 5 6 – 112 ) 0°–30°N 85 Central America 72 12 ( − 14%) − 170°–30°W (65–122) 320 0°–60°S 285 SH America − 35 ( − 11% ) (190 – 473) 170°–30°W 30°–75°N 33 +7 (+22%) 41 Europe and Mediterranean (19 – 62) 30°W–60°E 0°–30°N 404 357 NH Africa − 47 ( − 12%) (353– 453) 30°W–60°E 0°–35°S 485 457 SH Africa − − 6%) 28 ( (444–528) 30°W–60°E 186 30°–75°N 139 Northern Asia − 48 ( − 26%) 60°E–170°W (99 – 418) 10°–30°N 122 81 Southeast Asia − 41 ( − 34%) (101–150) 60°E–190°E 10°N –10°S 143 23 Tropical Asia − − 120 ( 84%) (38–425) 60°–170°E 112 10°–50°S +3 (+3%) 115 Australia (47–219) 60°E–170°W | S68 AUGUST 2018

89 3. GLOBAL OCEANS - concen Over the long term, as atmospheric CO 2 G. C. Johnson a. Overview— trations have risen, the ocean has taken up more The global oceans transport, store, and exchange carbon and acidified. Also on the long term, the with the atmosphere vast amounts of heat, water, - 1993–2017 trends in OHC and sea level ref lect statis carbon dioxide, and other constituents vital to tically significant warming and sea level rise, espe - climate. This chapter describes, with a focus on 2017 cially in the Southern Hemisphere. Near the Antarctic conditions, seasonal to interannual variability of sea Circumpolar Current a warming trend is evident in surface temperature; ocean heat content; salinity; the Indian Ocean and western Pacific sectors of the air–sea f luxes of heat, freshwater, and momentum; Southern Ocean, but a cooling trend is discernible in 2 sea level; surface currents; the Atlantic meridional the central Pacific sector. circulation; phytoplankton; and ocean inorganic While the global average SST for 2017 was slightly carbon cycling. It also puts 2017 conditions in a below the 2016 value, the long-term trend is upward. longer-term context. The last three years have been the three highest Neutral to weak La Niña conditions that held for annual values observed and have been associated with - much of 2016 recurred in 2017, so sea surface tem widespread coral bleaching. Both global average sea peratures (SSTs), ocean heat content (OHC), and sea level and the global integrals of 0–700 m and 0–2000 level continued to rise in the western tropical Pacific m OHC reached record highs in 2017. Global integrals and fall in the eastern tropical Pacific. A zonal band of OHC and global averages of sea level exhibit of westward surface current anomaly north of and substantially less variable upward trends than that along the equator played a role in the redistribution of for SST. In haiku form: - warm near-surface waters from east to west. Sea sur face salinity (SSS) freshened in the west and became Surface f luctuates, saltier in the east. In the eastern equatorial Pacific ocean warms more steadily, a chlorophyll- and CO seas continue rise. f lux from ocean to atmosphere 2 - were both elevated. All these tropical Pacific varia tions were consistent with 2016 to 2017 tendencies in b. Sea surface temperatures— B. Huang, J. Kennedy, Y. Xue, regional wind stress and freshwater f lux. There was a and H.-M. Zhang Global sea surface temperature (SST) and its prominent band of anomalously high SST, OHC, and changes are assessed mostly based on the Extended sea level as well as low SSS across much of the North Reconstruction Sea-Surface Temperature version 5 Pacific between about 5° and 30°N in 2017. Effects - included nuisance f looding in Hawaii. (ERSSTv5; Huang et al. 2017) unless otherwise speci 1 fied. The global integral of annually averaged SST SST, OHC, SSS, and sea Southeast of Greenland anomaly (SSTA; relative to a 1981–2010 climatology) level all remained below average in 2017, as they have ) ±0.06 decreased slightly from a historic high of 0.40 ( since 2014. Along the east coast of North America °C in 2016 to 0.34 ( ) °C in 2017, although these SSTs, OHC, and sea level were all anomalously high ±0.06 values are not statistically distinguishable. The 95% in 2017, as they have generally been since at least 2009. confidence levels are estimated from a 1000-member In climate models this North Atlantic SST pattern is ensemble of ERSSTv4 (Huang et al. 2016a). associated with a reduction in the Atlantic meridional Annually averaged SSTA in 2017 exceeded +0.5°C overturning circulation. in the western tropical Pacific, subtropical North In the Indian Ocean SST and SSS anomalies were Pacific, western subtropical South Pacific, western both high in the west and below average in different parts of the east in 2017. Around the equator OHC tropical Indian Ocean, and eastern North Atlantic fell in the west owing to a shoaling thermocline there. (Fig. 3.1a). SSTA values exceeded +1°C adjacent to the Anomalous eastward surface velocities around the Arctic (in the Nordic, Barents, Chukchi, and Bering - Seas; see Section 5e for a full description of Arctic equator in 2017 were likely associated with that shoal ing thermocline and the high SSS anomalies there, SSTs), the northwestern North Atlantic southeast of with the latter owing partly to anomalous eastward Cape Cod, around the Korean Peninsula, the central advection of salty water. Sea level fell from 2016 to southern Indian Ocean, and South Atlantic near the 2017 north of ~10°S in the Indian Ocean and rose coasts of Argentina and Uruguay. In contrast, SSTA was slightly colder than average in the Southern south of that latitude. 2 1 Southern Ocean conditions and sea ice are discussed in Arctic and Nordic Seas SST and sea ice are discussed in Chapter 6. Chapter 5. | S69 AUGUST 2018 STATE OF THE CLIMATE IN 2017

90 from a shift from the strong 2015/16 El Niño to a weak La Niña in late 2016, with neutral conditions during most of 2017 and La Niña recurring later in 2017 (Fig. 3.2d; Ashok et al. 2003; Huang et al. 2016b; Xue and Kumar 2017; L’Heureux et al. 2017). In the eastern tropical Pacific near Peru, negative SSTAs reached −1σ (standard deviation derived from ERSSTv5 over 1981–2010) below average in SON 2017 (Fig. 3.2d), while SSTAs were extraordinarily high (+2σ) in DJF and MAM (Figs. 3.2a,b). In the western tropical Pa - cific, high SSTAs (+1 to +2σ) persisted throughout the year (Fig. 3.2), extending to the subtropical North Pacific over 10°–30°N. In the Indian Ocean, the SSTA pattern of warm (+1 to +2σ) in the west and cold (−1σ) in the east was sustained throughout all of 2017. This SST pattern resulted in a positive phase of the Indian Ocean dipole (IOD; Saji et al. 1999) in 2017 (Fig. 3.2), while the IOD was negative in 2016. The positive IOD in 2017 did not correspond with the development of the La Niña (Meyers et al. 2007). In the North Pacific near 45°N, SST was 1σ colder than average in DJF (Fig. 3.2a), cold SSTA weakened in MAM and JJA (Figs. 3.2b,c), and SSTA reached +1σ 3.1. (a) Annually averaged ERSSTv5 SSTA (°C) . F ig near the dateline in SON (Fig. 3.2d). The pattern of for 2017 relative to a 1981–2010 climatology and (b) +1σ SSTA near the dateline and weaker (< +1σ) SSTA difference of annually averaged SSTAs between 2017 east of the dateline resulted in a negative phase of the and 2016 (2017–2016). Pacific decadal oscillation (PDO; Mantua and Hare 2002) in JJA and SON (Figs. 3.2c,d), consistent with Hemisphere oceans, the central-eastern tropical Pacific, eastern North Pacific near 45°N, southeastern the development of La Niña conditions in the tropical tropical Indian Ocean, and Southern Ocean south of 40°S in 60°W–60°E and 180°–120°W. In comparison with 2016, SST in 2017 was mostly cooler in the tropi - cal/subtropical oceans while warming - in the midlatitudes (Fig. 3.1b). Spe cifically, the 2017 SST was 0.2 °–0.5°C higher than 2016 in the central-to- southwestern North Pacific, central South Pacific, eastern North Atlantic, Chukchi Sea of the Arctic, western Indian Ocean, and Southern Ocean south of 40°S. In contrast, SST was 0.5 °–1.0°C lower in the high-latitude North Pacific and 0.2 °–0.5°C lower in the central-eastern tropical Pacific, western subtropical North Pacific, western North Atlantic, Arctic in the Atlantic sector, subtropical South . F ig 3.2. Seasonally averaged SSTAs of ERSSTv5 (°C; colors) for (a) Dec Atlantic, and eastern Indian Ocean. 2016–Feb 2017, (b) Mar–May 2017, (c) Jun–Aug 2017, and (d) Sep–Nov Cooling SST in the tropical Pacific 2017. Normalized seasonal mean SSTA based on seasonal mean std. and Indian Oceans in 2017 in compar - dev. for 1981–2010 are indicated by contours of − 1 (dashed white), +1 (solid black), and +2 (solid white). ison with 2016 (Figs. 3.1a,b) resulted | S70 AUGUST 2018

91 Pacific. In the subtropical South Pacific, the SSTA plitude is typically smaller in the Southern Ocean. The variations associated with the Atlantic multi - was +1 to +2σ in DJF and MAM (Figs. 3.2a,b). The warm SSTA sustained in the west in JJA and SON, decadal oscillation (AMO; Wanner et al. 2001) in but diminished in the east (Figs. 3.2c,d) due to the the North Atlantic can clearly be identified with development of the La Niña. warm periods during the 1930s–50s (not shown) and SSTAs were +1 to +2σ in most of the Atlantic 1990s–2010s and a cold period during the 1960s–80s (Fig. 3.2). SSTA in the Chukchi Sea was near neutral (Fig. 3.3f). From 2016 to 2017, annually averaged SSTA decreased in the Indian Ocean from 0.58°C to in DJF and MAM, and warmed to +2σ in JJA and SON. SSTA south of Greenland was below normal 0.35°C and decreased in the Pacific from 0.45°C to in DJF, MAM, and JJA (de Jong and de Steur 2016) 0.39°C. However, annually averaged SSTA increased slightly in the Atlantic from 0.42°C to 0.43°C due to but above normal in SON. SSTA from 20°–30°S was strong warming in the eastern North Atlantic, and near neutral in DJF and MAM, and cooled in JJA and it increased slightly from −0.11°C to −0.08°C in the SON. Cold anomalies in the high-latitude Southern Ocean weakened from −1σ in DJF and MAM to near Southern Ocean south of 45°S. neutral in JJA and SON in the Atlantic sector (Figs. SSTs in ERSSTv5 are compared with those in 3.2a,b), but strengthened from near neutral in DJF the high-resolution satellite-based daily optimum and MAM to −1σ in JJA and SON in the Pacific sector (Figs. 3.2c,d). The long-term warming trend of globally averaged SST remained strong (Figs. 3.3a,b), although SST cooled slightly from 2016 to 2017. The linear trend of globally and annually averaged SSTA (for ERSSTv5) is 0.17 ( ) °C ±0.08 −1 decade from 2000 to 2017 and −1 f rom 0.10 (±0.01) °C decade 1950 to 2017 (Table 3.1), with 95% confidence levels. The higher trend for 2000 to 2017 indicates a potential accelerating warming in the modern period. The 2000 to 2017 warming trend for ERSSTv5 is consistent with those reported by Karl et al. (2015) and Hausfather et al. (2017) using ERSSTv4 data. Warming of the global oceans from 2000 to 2017 (Table 3.1) was fast in the North Pacific (Fig. 3.3d) and tropical Indian Ocean (Fig. 3.3e) compared with the other regions (Fig. 3.3). Warming of the global oceans from 1950 to 2017 (Table 3.1) was nominally faster in - the tropical Indian Ocean, tropi cal Atlantic, and North Atlantic; and slower in the tropical Pacific, - Southern Ocean, and North Pa std. σ 2 3.3. Annually averaged SSTAs from ERSSTv5 (white line) with ig . F cific. dev. (gr ay shading) of ERSSTv4, DOISST (green line), and HadSST.3.1.1.0 In addition to the long-term (red line) from 1950 to 2017 except for (b). (a) global ocean, (b) global SST trend, short-term SST varia - ocean from 1880 to 2017, (c) tropical Pacific, (d) North Pacific, (e) tropical tions can be seen in all global Indian, (f) North Atlantic, (g) tropical Atlantic, and (h) Southern Oceans. The year 2000 is indicated by a vertical black dotted line. ocean basins, although their am - | S71 AUGUST 2018 STATE OF THE CLIMATE IN 2017

92 slightly higher in DOISST from 2000 – 1 t 3.1. Linear trends (°C decade ) of annually averaged SSTAs le AB to 2017. These SSTA differences are from ERSSTv5 except for global averaged SSTAs from HadSST3 mostly attributed to differences in and DOISST. The uncertainty at 95% confidence level are esti - ship-based observation bias correc - mated by accounting for AR(1) effect on the degrees of freedom of annually averaged SST series. tions in different products (Huang et al. 2015; Kent et al. 2017). 1950 –2017 2000–2017 HadSST.3.1.1.0, Global 0.137 ± 0.078 0.083 ± 0.017 c. Ocean heat content— N/A 0.180 ± 0.067 DOISST, Global G. C. Johnson, J. M. Lyman, T. Boyer, L. Cheng, C. M. Domingues, J. Gilson, M. Ishii, R. Killick, ERSSTv5, Global 0.166 ± 0.082 0.099 ± 0.011 D. Monselesan, S. G. Purkey, and S. E. Wijffels 0.100 ± 0.024 Tropical Pacific (30°S–30°N) 0.188 ± 0.192 Storage and transport of heat in the North Pacific (30°–60°N) 0.062 ± 0.031 0.268 ± 0.135 ocean are central to aspects of climate 0.240 ± 0.084 0.143 ± 0.016 Tropical Indian Ocean (30°S–30°N) such as ENSO (Johnson and Birn - 0.118 ± 0.100 North Atlantic (30°–60°N) 0.102 ± 0.042 baum 2017), tropical cyclones (Goni et al. 2009), sea level rise (Section 3f), 0.109 ± 0.018 0.158 ± 0.104 Tropical Atlantic (30°S–30°N) variations in the global average surface 0.103 ± 0.055 0.097 ± 0.014 Southern Ocean (30°–60°S) - warming rate (Xie et al. 2016), melt ing of ice sheet outlet glaciers around Greenland (Castro de la Guardia et al. 2015) and interpolation SST (DOISST; Reynolds et al. 2007), Antarctica (Schmidtko et al. 2014), and coral bleach - and U.K. Met Office Hadley Centre SST version 3 ing (Sidebar 3.1). Ocean warming accounts for about (HadSST.3.1.1.0; Kennedy et al. 2011a,b), which uses 93% of the total increase in Earth’s energy storage a different algorithm from ERSSTv5 to correct ship from 1971 to 2010 (Rhein et al. 2013). SST observation bias. ERSSTv5 is a monthly SST product on a 2°×2° horizontal grid from 1854 to Maps of annual (Fig. 3.4) upper (0–700 m) ocean heat content anomaly (OHCA) relative to a 1993–2017 present based on in situ observations only (Huang et al. 2017). It builds upon ERSSTv4 (Huang et al. 2015) baseline mean are generated from a combination of with more ship and buoy observations and the added in situ ocean temperature data and satellite altimetry ingestion of near-surface Argo observations. Biases in data following Willis et al. (2004), but using Argo (Riser et al. 2016) data downloaded in January 2018. ship-based measurements are corrected using more Near-global average seasonal temperature anomalies - accurate buoy observations. ERSSTv5 provides a bet (Fig. 3.5) vs. pressure from Argo data (Roemmich and ter representation of spatial and temporal variations in high-latitude oceans and ENSO variability in the Gilson 2009, updated) since 2004 and in situ global tropical Pacific than ERSSTv4. DOISST is a daily estimates of OHCA (see Fig. 3.6) for three pressure layers from six different research groups (including 0.25°×0.25° SST product for the modern satellite era from September 1981 to present using both in situ and those responsible for the 2000–6000 - stimate) are m e also discussed. satellite observations. HadSST.3.1.1.0 is a monthly 5°×5° SST product from 1850 to present using in The 2017 minus 2016 tendency of 0–700-m OHCA (Fig. 3.4b) shows increases in the western subtropical situ observations only. All datasets are averaged to - and tropical North Pacific, with strong bands extend monthly 2°×2° grids for comparison purposes. ing east-southeastward from Papua New Guinea Comparisons (Fig. 3.3) indicate that SSTA devia - and east-northeastward from the Philippines. These tions of DOISST and HadSST.3.1.1.0 from ERSSTv5 bands are reminiscent of Rossby wave signatures, are largely within 2σ (gray shading in Fig. 3.3), as derived from a 1000-member ensemble analysis which propagate westward more quickly closer to of ERSSTv4 (Huang et al. 2016a) and centered on the equator (Chelton and Schlax 1996). Decreases ERSSTv5 SSTA. However, SSTAs are slightly higher are observed in portions of the central South Pacific in the 1950s–70s and 1920s–30s in HadSST.3.1.1.0 and eastern North Pacific, as well as in the eastern than in ERSSTv5 (Fig. 3.3b). Additionally, SSTAs are equatorial Pacific around the latitude of the ITCZ. slightly higher in the 2000s–10s in HadSST.3.1.1.0 Throughout much of the Pacific, the 2017 upper and DOISST than in ERSSTv5, particularly in the OHCA is generally above the long-term average, with the most prominent below-average region in Southern Ocean. Therefore, SST trends are slightly the central South Pacific (Fig. 3.4a). A prominent weaker in HadSST.3.1.1.0 for both 1950 to 2017 and 2000 to 2017 (Table 3.1). In contrast, SST trends are band of high OHCA from about 5° to 30°N in 2017 | S72 AUGUST 2018

93 is associated with anomalously high sea level (see Fig. tor. This reverses a pattern that had held since 2014 reports) and is also State of the Climate (see previous 3.16), with impacts that include nuisance f looding ref lected in sea level (see Fig. 3.16). in Hawaii (Sidebar 3.2). The Bering Sea, the Sea of Okhotsk, and especially the Yellow Sea and the Sea of In the Indian Ocean, the 2017 minus 2016 ten - Japan all exhibited anomalously high OHCA values dency of 0–700-m OHCA (Fig. 3.4b) exhibited strong in 2017. The 2017 0–700-m OHCA anomalies (Fig. cooling in the western equatorial region. This cooling 3.4a) are slightly above the 1993–2017 average in the was caused by a shoaling of the thermocline, which western tropical Pacific, and below that average in had little expression in SST (see Fig. 3.1), but was re - f lected in a reduction in SSH (see Fig. 3.16b) and likely - the eastern tropical Pacific on both sides of the equa effected by anomalous eastward f low in 2017 around the equator (see Fig. 3.18a). Near-surface salinity also increased in the western Indian Ocean (see Figs. 3.7b and 3.10c) during 2017, so the SSH reductions may be larger than would be expected from the cooling alone. There was also some warming south of 10°S and in much of the eastern Indian Ocean along with a large tendency toward higher values in a zonal band around 40°S in the Indian sector of the Southern Ocean. Upper OHCA values for 2017 were above the 1993–2017 mean in much of the Indian Ocean (Fig. 3.4a), with the notable exception of the formation of a large patch of low values in the western equatorial region discussed above. There was a 2017 minus 2016 tendency toward higher values of 0–700-m OHCA (Fig. 3.4b) in the western subtropical and eastern subpolar North At - lantic, along with a slight increase throughout much of the tropics and a strong increase in a patch at about 35°S adjacent to South America. Weak decreases from 2016 to 2017 were apparent in the western tropical North Atlantic and in the subpolar region around Greenland. Much of the western North Atlantic from the Tropic of Capricorn to about 50°N had anomalously high upper OHCA values in 2017 (Fig. 3.4a), as did much of the Gulf of Mexico, the western Caribbean Sea, and the Greenland–Iceland–Norwe - gian Seas. The warm conditions off the east coast of North America have generally been present since 2009 (see previous State of the Climat e reports). The only large region in the entire Atlantic with 2017 values well below the 1993–2017 mean was south of Greenland, especially in the Irminger Sea; a pattern that has persisted since at least 2014. F 3.4. (a) Combined satellite altimeter and in situ . ig - Regions around all the subtropical western bound ocean temperature data estimate of upper (0–700 m) 2 9 − ary current extensions: the Kuroshio and the East OHCA (× 10 ) for 2017 analyzed following Willis et J m al. (2004), but using an Argo monthly climatology and Australian Current in the Pacific, the Agulhas ret - displayed relative to the 1993–2017 baseline. (b) 2017 rof lection in the Indian Ocean, and the Gulf Stream minus 2016 combined estimates of OHCA expressed and the Brazil Current in the Atlantic all displayed 2 − as a local surface heat flux equivalent (W m ). For (a) anomalously high upper OHCA in 2017 (Fig. 3.4a). − 2 and (b) comparisons, note that 95 W m applied over A trend toward higher values is also statistically − 9 2 one year results in a 3 × 10 change of OHCA. (c) J m significant from 1993 to 2017 in all these boundary Linear trend for 1993–2017 of the combined estimates − 2 currents (Fig. 3.4c), consistent with previous analysis of upper (0–700 m) annual OHCA (W m ). Areas with (Wu et al. 2012). statistically insignificant trends are stippled. | S73 AUGUST 2018 STATE OF THE CLIMATE IN 2017

94 Other large-scale statistically significant (Fig. significant positive trends in much of the tropical Atlantic and North Indian Oceans, as well as all of 3.4c) regional patterns in the 1993–2017 local linear the marginal seas except the Red Sea. The strongest trends of upper OHCA ref lect a warming trend in negative trends are found in the North Pacific south much of the Southern Hemisphere (Roemmich et al. 2015; Wijffels et al. 2016). In addition, there are of the Kuroshio Extension and in the North Atlantic SIDEBAR 3.1: UNPRECEDENTED THREE YEARS OF GLOBAL CORAL C. M. EAKIN BLEACHING 2014–17 — , G. LIU, A. M. GOMEZ, J. L. DE LA COUR1, S. F. HERON, W. J. SKIRVING, E. F. GEIGER, B. L. MARSH, K. V. TIRAK, AND A. E. STRONG Continued ocean warming has taken a severe toll on tropical coral reefs worldwide as heat stress has caused repeated bleaching and disease outbreaks (Eakin et al. 2009). Bleaching occurs when stress to the coral–algal symbiosis causes corals to expel endosymbiotic algae and, if prolonged or particularly severe, can result in partial or complete coral mortality (Brown 1997). While many stressors can cause bleaching, “mass” coral bleaching (covering hundreds of kilometers or more) is primarily driven by prolonged anomalously warm ocean temperatures coupled with high subsurface light levels, exceeding corals’ physiological tolerances. Heat stress causing mass coral bleaching can be monitored accurately by satellites (G. Liu et al. 2014, 2017) and has increased in . SB3.1. Evolution of and maximum heat stress for (a) F ig frequency and severity with a warming climate (Hughes 2015 and (b) 2016. Black lines show the annual pattern of et al. 2018). heat stress from the South Pacific to the southern Indian Past mass bleaching events usually have been limited Ocean, northern Indian Ocean and Southeast Asia, then to to El Niño years: the first in 1983 (Glynn 1990; Coffroth the Caribbean. Severe coral bleaching was reported from et al. 1990) was followed by the first recorded global colleagues in all areas circled in white. CRW Bleaching Alert event in 1998 and the second in 2010 (Heron et al. Area categories are defined as: 2016b). The third global coral bleaching lasted three full N 0; ≤ o Stress, HotSpot W atch, 0 < HotSpot < 1; years during 2014 to 2017 (NOAA 2017). This third global < DHW < 4; Warning, 1 ≤ HotSpot and 0 bleaching event was the longest, most widespread, and 4 nd a otSpot H ≤ 1 , 1 Alert DHW < 8, coral ≤ almost certainly most destructive on record. NOAA’s b leaching likely; Coral Reef Watch (CRW) defines global coral bleaching 8 and HotSpot widespread ≤ ≤ 1 2, Alert DHW, events as those with coral bleaching spanning hundreds leaching and significant coral mortality likely; b of kilometers or more in all three ocean basins. where HotSpot is the positive SST anomaly compared to The third global bleaching event started in June 2014, the maximum monthly mean climatology and DHW is the when El Niño formation was expected but never fully Degree Heating Week heat stress accumulated from 12 preceding weeks of HotSpot values (G. Liu et al. 2014). Data materialized (Blunden and Arndt 2015). CRW satellite from CRW’s Daily Global 5-km Coral Bleaching Heat Stress monitoring first detected heat stress sufficient to cause Monitoring Product Suite v.3 (G. Liu et al. 2017). coral bleaching (Alert Level 1, defined in Fig. SB3.1) in Guam and the Commonwealth of the Northern Mariana and sequence of heat stress and bleaching (Fig. SB3.1a) seen Islands (CNMI; Heron et al. 2016a). An anomalously warm during the second year of an El Niño during prior global North Pacific Ocean brought severe heat stress to parts of bleaching events (Hoegh-Guldberg 1999). Papua New Guinea the Northwestern Hawaiian Islands (Couch et al. 2017) and and Fiji reported moderate heat stress and bleaching early in the Marshall Islands in the central Pacific (Fellenius 2014) and 2015 (Alert Level 1; Fig. SB3.1a); American Samoa reported moderate stress to the main Hawaiian Islands (DAR 2014; Bahr its worst bleaching ever with widespread Alert Level 2 condi - et al. 2015). Teleconnected warming then brought heat stress tions. CRW documented moderate heat stress in the Indian to Florida (FRRP 2015). Ocean, which spread to the Chagos Archipelago, Red Sea, Despite El Niño formation being delayed until March and western Indonesia (Eakin et al. 2016). With the El Niño’s (Blunden and Arndt 2016), 2015 followed the classic pattern onset, SST anomalies in the central and eastern tropical Pacific | S74 AUGUST 2018

95 south of Greenland. The apparent warming and cool - in 1993–2013 but had disappeared for 1993–2016 (see reports) returned for ing trends adjacent to Antarctica are located in both previous State of the Climate in situ and altimeter data-sparse regions and may not 1993–2017, emphasizing the strong inf luence of El Niño and perhaps the PDO on sea level trends in that be robust. A statistically significant warming trend in the western tropical Pacific that was quite strong region (Merrifield et al. 2012), as well as the sensitiv - resulted in heat stress in the Line Islands at record levels (Fig. ever seen in the Flower Garden Banks, located in the Gulf of SB3.1a). Hawaii saw its worst bleaching ever (Rosinski et al. Mexico (Johnston et al. 2017). Later, bleaching returned to 2017). Florida’s reefs experienced a second consecutive mass the Marshall Islands in the central Pacific (Eakin et al. 2017). bleaching (FRRP 2016), and bleaching was reported on reefs Some South Pacific bleaching was reported in early 2017: across the eastern and western Caribbean (Eakin et al. 2016). mild in Fiji and severe in Niue, Samoa, and American Samoa. With widespread bleaching observed in the Indian, Pacific, and A second year of bleaching struck the GBR; this first-ever Atlantic basins, NOAA declared in October 2015 the third consecutive bleaching there focused on the northern and global coral bleaching event was underway (NOAA 2015). By central sectors, killing another 22% of the GBR’s corals then, 41% of the world’s coral reefs had experienced heat stress (Hughes and Kerry 2017). Heat stress was limited in the Indian of 4°C-weeks (as defined in the caption of Fig. SB3.1a) or more, Ocean with moderate bleaching in southwestern Madagascar and almost all reefs had exceeded their normal warm-season (CORDIO-EA 2016) and mild bleaching elsewhere. With a temperatures. lack of widespread, severe Indian Ocean bleaching, NOAA With a strong El Niño in 2016, heat stress and bleaching declared the global event over as of May 2017 (NOAA 2017). returned to the Southern Hemisphere and followed the same However, bleaching still continued, with Guam and the CNMI global sequence as 2015—the first time this occurred in back- experiencing widespread heat stress again. This was Guam’s to-back years (Fig. SB3.1b). Heat stress in 2016 was much more worst documented bleaching and fourth widespread event widespread and intense than in 2015, encompassing 51% of in five years. coral reefs globally. Even more important was the severity. Lasting an unprecedented 36 months, the third global event Continuous El Niño-related heat stress in the central Pacific brought mass bleaching-level heat stress (Alert 1) to more for over a year caused the highest heat stress values CRW had than 75% of global reefs; nearly 30% also suffered mortality- ever documented. More than 25°C-weeks of heat stress in the level stress (Alert 2; Fig. SB3.2). More than half of affected Northern Line Islands killed 80% and bleached another 15% reef areas were impacted at least twice. This global event has of corals in Kiritimati (Harvey 2016) and killed 98% of corals - punctuated the recent acceleration of mass bleaching. Oc at Jarvis Island (Brainard et al. 2018). The first mass bleaching curring at an average rate of once every 25–30 years in the (85% bleached) of the northern and far-northern Great Barrier 1980s, mass bleaching now returns about every six years and Reef (GBR; Hughes et al. 2017) killed 29% of the GBR’s shallow- is expected to further accelerate as the oceans continue to water corals (GBRMPA 2018). Widespread heat stress brought warm (Hughes et al. 2018). Severe bleaching is now occurring bleaching to much of the western Indian Ocean (CORDIO- more quickly than reefs can recover, with severe downstream EA 2016), including 69%–99% of corals bleached and 50% consequences to ecosystems and people. dead in the Seychelles (SIF 2017). CRW’s forecasts of mass bleach - ing led Thailand to close many coral reef sites to diving (Agence France-Presse 2016). In the boreal summer, over 90% bleaching and 70% mortality were observed on Japan’s largest reef (Harvey 2017); widespread bleaching also hit Guam again. Heat stress in the western Atlantic caused extensive SB3.2. NOAA CRW Maximum Bleaching Alert Area map for Jun 2014–May . ig F - bleaching in the western Carib - 2017. Data from CRW’s Daily Global 5-km Coral Bleaching Heat Stress Monitor bean and the worst bleaching ing Product Suite v.3 (G. Liu et al. 2017). | S75 AUGUST 2018 STATE OF THE CLIMATE IN 2017

96 −1 ity of trends in relatively short records to choices of by 2000 below 180 dbar and about 0.01°C decade dbar. Removing a linear regression against the Niño- end points. 3.4 index (e.g., Johnson and Birnbaum 2017) results in Near-global average monthly (smoothed to sea - a decadal warming trend (Fig. 3.5b, blue line) that is sonal time-scale) temperature anomalies (Fig. 3.5a) ref lect both a long-term warming trend (Fig. 3.5b, somewhat less than the simple linear trend near the orange line) and ENSO redistributing heat (e.g., Ro surface and slightly larger from about 100 to 400 dbar. - The analysis is extended back in time from the emmich and Gilson 2011) from the upper 100 dbar Argo period (which started in the early 2000s) to 1993, to a roughly 300-dbar thick layer just below. Lower temperature values are evident in the upper 100 dbar and deeper, using sparser, and more heterogeneous, historical data collected mostly from ships (e.g., and higher values from 100 to 400 dbar during La Abraham et al. 2013). Six different estimates of glob - Niña (e.g., 2008/09), and vice versa during El Niño (e.g., 2015/16). Since the peak of El Niño near the start of 2016, mean temperatures in the upper 100 dbar declined, but throughout 2017 they still remained above the long-term average. Negative anomalies from 150 to 400 dbar peaking around the start of 2016 also abated, such that the entire water column from 0 to 2000 dbar was warmer in 2017 than the 2004–17 average. The overall warming trend (Fig. 3.5b, orange −1 line) from 2004 to 2017 exceeds 0.19°C decade near −1 the surface, declining to less than 0.03°C decade F ig . 3.6. (a) Annual average global integrals of in situ 21 estimates of upper (0–700 m) OHCA (ZJ; 1 ZJ = 10 J) for 1993–2017 with standard errors of the mean. The MRI/JMA estimate is an update of Ishii et al. (2017). The CSIRO/ACE CRC/IMAS-UTAS estimate is an update of Domingues et al. (2008). The PMEL/JPL/ JIMAR estimate is an update and refinement of Lyman and Johnson (2014). The NCEI estimate follows Levitus et al. (2012). The Met Office Hadley Centre estimate is computed from gridded monthly temperature anomalies (relative to 1950–2016) following Palmer et al. (2007). The IAP/CAS estimate is described in Cheng . F 3.5. (a) Near-global (65°S –80°N, excluding i g et al. (2017). See Johnson et al. (2014) for details on continental shelves, the Indonesian seas, the Sea uncertainties, methods, and datasets. For comparison, of Okhostk, the Caribbean Sea, and the Gulf of all estimates have been individually offset (vertically on Mexico) integrals of monthly ocean temperature the plot), first to their individual 2005–17 means (the anomalies [°C; updated from Roemmich and Gilson best sampled time period), and then to their collective (2009)] relative to record-length average monthly 1993 mean. (b) Annual average global integrals of in values, smoothed with a 5-month Hanning filter and situ estimates of intermediate (700–2000 m) OHCA contoured at odd 0.02°C intervals (see colorbar) vs. for 1993–2017 with standard errors of the mean, and a pressure and time. (b) Linear trend of temperature long-term trend with one standard error uncertainty anomalies over time for the length of the record in (a) shown from 1992–2010 for deep and abyssal (z > 2000 − 1 plotted vs. pressure in °C decade (orange line), and m) OHCA following Purkey and Johnson (2010) but trend with a Niño-3.4 regression removed (blue line) updated using all repeat hydrographic section data following Johnson and Birnbaum (2017). available from https://cchdo.ucsd.edu/ as of Jan 2018. | S76 AUGUST 2018

97 ally integrated in situ 0–700-m OHCA (Fig. 3.6a) all G. C. Johnson, J. Reagan, J. M. Lyman1, T. Boyer, d. Salinity— reveal a large increase since 1993, with 2017 being a C. Schmid, and R. Locarnini G. C. Johnson and J. Reagan — ntroduction 1) i record high value. A similar pattern is apparent from The ocean plays a large role in the global hydro - 700 to 2000 m (Fig. 3.6b). While the trend in globally logical cycle, with the vast majority of evaporation integrated ocean heat content is modulated by El Niño and is slowed or even reversed after El Niño peaks and precipitation occurring over the oceans (e.g., Schanze et al. 2010). Ocean freshwater storage and (Johnson and Birnbaum 2017), it, like globally aver - transport, and variations thereof, are ref lected in aged sea level (see Fig. 3.15a), exhibits a much steadier salinity patterns and their variations (e.g., Yu 2011). increase than globally averaged surface temperatures (see Fig. 3.3a). Globally integrated OHCA values vary Where evaporation exceeds precipitation, such as in the subtropics, relatively salty surface waters are more both from year-to-year for individual years and found. In contrast, where precipitation (and river from estimate-to-estimate in any given year prior to the achievement of a near-global Argo array around run off) is greater than evaporation, such as under 2005. Causes of differences among estimates are the ITCZs and in subpolar regions, fresher waters are present. In high latitudes, sea ice formation, ad - discussed in Johnson et al. (2015). vection, and melt also inf luences SSS (e.g., Petty et The rate of heat gain from linear trends fit to each of the six global integral estimates of 0–700 m OHCA al. 2014). Subsurface ocean salinity patterns ref lect from 1993 through 2017 (Fig. 3.6a) range from 0.36 the surface formation regions of water masses (e.g., −2 applied over the surface - Skliris et al. 2014), with fresher tropical waters overly (±0.06) to 0.40 (±0.18) W m area of Earth (Table 3.2). Linear trends from 700 to ing saltier subtropical waters, which in turn overlay 2000 m over the same time period range from 0.19 fresher subpolar waters. Below these water masses −2 . Trends in the 0–700- (±0.07) to 0.35 (±0.03) W m lie the saltier North Atlantic Deep Water and below m layer all agree within uncertainties, and all but that the fresher Antarctic Bottom Water (Johnson 2008). North Atlantic Deep Water temperature and one of the four trends in the 700–2000-m layer do as salinity vary over decades (e.g., Yashayaev and Loder well. For that layer the PMEL/JPL/JIMAR trend is larger than the others because it assumes the average 2016), whereas Antarctic Bottom Waters have been freshening in recent decades (e.g., Purkey and John anomaly in sampled regions applies globally (Lyman - and Johnson 2014). For 2000–6000 m, the linear trend son 2013). Salinity changes impact sea level changes −2 is 0.04 (±0.04) W m (e.g., Durack et al. 2014) as well as the thermohaline from 1992 to 2010. Summing circulation (e.g., W. Liu et al. 2017) and have been the three layers (with their slightly different time periods), the full-depth ocean heat gain rate ranges used to quantify changes in the hydrological cycle −2 (e.g., Skliris et al. 2014). . from 0.59 to 0.79 W m To investigate interannual − 2 applied over 2. Trends of ocean heat content increase (in W m 3. le AB t changes of subsurface salinity, 14 2 the 5.1 × 10 m surface area of Earth) from seven different research all available salinity profile groups over three depth ranges (see Fig 3.6 for details). For the 0–700-m data are quality controlled and 700–2000-m depth ranges, estimates cover 1993–2017, with 5%–95% following Boyer et al. (2013) uncertainties based on the residuals taking their temporal correlation and then used to derive 1° into account when estimating degrees of freedom (Von Storch and Zwiers monthly mean gridded salinity 1999). The 2000–6000-m depth range estimate, an update of Purkey and Johnson (2010), covers from 1992 to 2010, again with 5%–95% uncertainty. anomalies relative to a long- term monthly mean for years Global ocean heat content trends 1955–2012 [World Ocean Atlas − 2 (W m ) for three depth ranges Research group 2013 version 2 (WOA13v2); 0–700 m 2000–6000 m 700–2000 m Zweng et al. 2013] at standard depths from the surface to MRI/JMA 0.36 ± 0.06 0.22 ± 0.06 — 2000 m (Boyer et al. 2013). In — CSIRO/ACE/CRC/IMAS/UTAS — 0.40 ± 0.07 recent years, the largest source 0.40 ± 0.16 PMEL/JPL/JIMAR — 0.35 ± 0.03 of salinity profiles is the profil- NCEI 0.38 ± 0.07 0.19 ± 0.07 — ing f loats of the Argo program (Riser et al. 2016). These data — — 0.40 ± 0.18 Met Office Hadley Centre - are a mix of real-time (pre — ICCES 0.40 ± 0.06 0.19 ± 0.01 liminary) and delayed-mode Purkey and Johnson update — — 0.04 ± 0.04 (scientific quality controlled) | S77 AUGUST 2018 STATE OF THE CLIMATE IN 2017

98 observations. Hence, the estimates presented here cies is the increase in salinity in the equatorial Indian could change after all data have been subjected to Ocean. This increase is associated with anomalous - eastward currents, consistent with anomalous advec scientific quality control. The SSS analysis relies on Argo data downloaded in January 2018, with annual tion in the presence of mean SSS that decreases from maps generated following Johnson and Lyman (2012) west to east. Other prominent large-scale SSS changes as well as monthly maps of bulk (as opposed to skin) from 2016 to 2017 include freshening in the northeast SSS data from BASS (Xie et al. 2014). BASS blends in Pacific, the western tropical Atlantic, and around the Aquarius (Le Vine equator across much of the Atlantic (Fig. 3.7b). situ SSS data with data from the et al. 2014; mission ended in June 2015), SMOS (Soil Moisture and Ocean Salinity; Font et al. 2013), and recently SMAP (Soil Moisture Active Passive; Fore et al. 2016) satellite missions. BASS maps can be biased fresh around land (including islands) and at high latitudes. Despite the larger uncertainties of satellite data relative to Argo data, their higher spatial and temporal sampling allows higher spatial and temporal resolution maps than are possible using in situ data alone at present. Salinity is measured as a dimen - sionless quantity and reported on the 1978 Practical Salinity Scale, or PSS-78 (Fofonoff and Lewis 1979). 2) s — G. C. Johnson and J. M. Lyman ea surface salinity - Sea surface salinity anomalies outside of the trop ics are fairly persistent, so 2017 SSS anomalies (Fig. 3.7a, colors) include some extratropical large-scale patterns that largely held from 2004 to 2016 (previ - ous State of the Climate reports). Regions around the subtropical salinity maxima are generally salty with respect to WOA13v2, except in the North Pacific, where the salinity maximum is anomalously fresh in 2017. There are fresh anomalies relative to WOA13v2 in much of the high-latitude, low-salinity regions, primarily in portions of the subpolar gyres of the North Pacific and North Atlantic, and to a lesser extent around the Southern Ocean. These multiyear patterns are consistent with an increase in the hydro - logical cycle (e.g., more evaporation in drier locations and more precipitation in rainy areas) over the ocean, as expected in a warming climate (Rhein et al. 2013). A similar assertion could be made for some of the extratropical 2005–17 trends discussed below. Tropical sea surface salinity changes from 2016 to 2017 (Fig. 3.7b, colors) are anti-correlated with 2016 F 3.7. (a) Map of the 2017 annual surface salinity ig . - to 2017 tendencies in precipitation minus evapora anomaly (colors, PSS-78) with respect to monthly tion ( P − E ; see Fig. 3.12b). The freshening in the off- climatological 1955–2012 salinity fields from WOA13v2 equatorial western tropical Pacific and salinification [yearly average (gray contours at 0.5 intervals), PSS- 78]. (b) Difference of 2017 and 2016 surface salinity around the equator in the west and under the ITCZ − 1 maps (colors, PSS-78 yr ). White ocean areas are too − P in the east are all well anti-correlated with E ten - data-poor (retaining < 80% of a large-scale signal) to dencies, and associated with the transition from an map. (c) Map of local linear trends estimated from El Niño that peaked around early 2016 to the neutral annual surface salinity anomalies for 2005–17 (colors, or weak La Niña conditions since then. A prominent 1 − PSS-78 yr ). Areas with statistically insignificant example of the role of advection by anomalous ocean trends are stippled. All maps are made using currents (see Fig. 3.18) in the 2016 to 2017 SSS tenden - Argo data. | S78 AUGUST 2018

99 North Pacific, as well as patches - in the Pacific and Atlantic sec tors of the Southern Ocean. In the tropics, they include the cen - tral Pacific, at the eastern edge of the western fresh pool, and in the warm fresh pool of the north - eastern equatorial Pacific. There are also freshening trends in the already fresh Bay of Bengal, and a large patch west of Indonesia and Australia that has been present since at least 2009 (see previous State of the Climate reports). The regions to the northwest of the Gulf Stream and in the northern Gulf of Mexico are also trending F ig . 3.8. Seasonal maps of SSS anomalies (colors) from monthly blended strongly saltier, as well as warmer maps of satellite and in situ salinity data (BASS; Xie et al. 2014) relative (Sect ion 3c). to monthly climatological 1955–2012 salinity fields from WOA13v2 for (a) Dec–Feb 2016/17, (b) Mar–May 2017, (c) Jun–Aug 2017, and (d) Sep–Nov 2017. 3) s ubsurface salinity — J. Reagan, Areas with maximum monthly errors exceeding 10 PSS-78 are left white. T. Boyer, C. Schmid, and R. Locarnini For the first time in the past decade, nearly all Strong seasonal variations of BASS (Xie et al. 2014) 2017 Atlantic Ocean basin-average monthly salinity SSS anomalies (Fig. 3.8) are evident near the Amazon anomalies were positive from 0 to 1500 m (Fig. 3.9a). and Orinoco River plumes. While there is almost no fresh signal in December–February, a strong fresh The year 2017 continued the same Atlantic salinity anomaly pattern that has been evident since 2008 anomaly extends north of the river mouths in March– with strong positive (> 0.05) near-surface salinity May and grows to the north and extends eastward in June–August, with a strong fresh anomaly extend - anomalies that weaken with depth (~0.005 at 700 ing across much of the northern equatorial Atlantic m) (Fig. 3.9a). Salinity increased at nearly all depths within 0–1500 m from 2016 to 2017 (Fig. 3.9b) with in September–November. Other factors, including the highest increase between 100 and 125 m (~0.017). stronger-than-usual precipitation in the region (see Fig. 3.12), likely contributed to the fresh anomaly as The 2017 Pacific Ocean basin-average salinity well. In the tropical Pacific, fresh anomalies in the anomalies continued the same pattern that began in eastern Pacific warm pool diminished throughout the mid-2014 with fresh anomalies from 0 to 75 m, salty year while the western Pacific warm pool freshened, anomalies from 100 to 200 m, and fresh anomalies again consistent with the shift of precipitation to the from 200 to 600 m (Fig. 3.9c). This marks the third straight year (2015–17) in which the upper ~75 m of western tropical Pacific after the 2015/16 El Niño the Pacific Ocean has been fresher than the long-term and persistence of neutral and La Niña conditions average. Previously, this layer had been saltier than throughout 2017. the long-term average for five straight years (2009–13; Sea surface salinity trends for 2005–17 (Fig. 3.7c) are estimated by local linear fits to annual average Fig. 3.9c). These basin-average multiyear near-surface SSS maps from Argo data. (The starting year is 2005 salinity shifts may be related to in-phase transitions because that is when Argo coverage became near- of both ENSO and the PDO and their associated pre - global.) Regions with statistically significant increas cipitation (Lau and Yang 2002) and equatorial wind - stress/Ekman upwelling changes (Wang et al. 2015). ing salinity trends are found near the subtropical salinity maxima in all the ocean basins, although From 2016 to 2017 the upper 125 m of the Pacific - the eastern subtropical North Atlantic is freshen became fresher (max of ~ −0.018 at 0 m), while the 150–400-m layer became saltier (max of ~0.013 at ing, even near the salinity maximum. In the higher 200 m; Fig. 3.9d). latitudes and the tropics, where mean salinity values From mid-2016 through 2017 the upper 200 m are lower, there are some regions where the trend is of the Indian Ocean became very salty (> 0.05 near toward freshening. In high latitudes, these freshen - ing regions include the subpolar North Atlantic and the surface) when compared to the long-term mean | S79 AUGUST 2018 STATE OF THE CLIMATE IN 2017

100 3.12a) over the Atlantic ITCZ may have also played a role in the freshening over this area. North of this freshening, there was salinification (> 0.03) between 2016 and 2017 from 20° to 52°N, which expanded and deepened from the surface to 100 m at 20°N to 400 m at 50°N with maximum salinification (> 0.09) occurring at 50 m at 47.5°N. The zonally averaged Pacific salinity changes from 2016 to 2017 are primarily concentrated in the upper 150 m (Fig. 3.10b). Near-surface (0–50 m) freshening (< −0.03) at 22°S extends equatorward and deepens to 150 m at 8°S with maximum freshening (< −0.12) at the surface near 15°S. This is a reversal of the sa - linification that took place from 2015 to 2016 in this area, and it is likely due to the transition from the strong El Niño in early 2016 to the neutral and weak La Niña conditions that dominated 2017 and its asso - ciated precipitation tendencies from 2016 to 2017 (see . F ig 3.9. Average monthly salinity anomalies (PSS-78) relative to the long-term WOA13v2 monthly salinity climatology for years 1955–2012 (Zweng et al. 2013) from 0 to 1500 m for the (a) Atlantic for 2008–17 and (b) the change from 2016 to 2017; (c) Pacific for 2008–17 and (d) the change from 2016 to 2017; (e) Indian for 2008–17 and (f) the change from 2016 to 2017. Data were smoothed using a 3-month running mean. (Fig. 3.9e). Much of the surface of the Indian Ocean experienced salty anomalies (Fig. 3.7a) that were not (see Fig. 3.12a) P – E driven directly by changes in but were due to anomalous ocean currents (see Fig. 3.18a and Section 3d2). Unsurprisingly, there was a large (~0.05 at 0 m; Fig. 3.9f) salinification of the near-surface from 0 to 100 m between 2016 and 2017, with freshening from 100 to 200 m, and salinification from 200 to 1000 m (Fig. 3.9f). - Most of the large (> |0.09|) zonally averaged sa linity changes from 2016 to 2017 in the Atlantic oc - curred in the North Atlantic (Fig. 3.10a). There was freshening in the upper 50 m from 0° to 20°N, with maximum freshening (< −0.12) at ~10 m depth. This is in contrast to the salinification that was observed in this region between 2015 and 2016 (see Fig. 3.10a in Reagan et al. 2017). The freshwater discharge from i g . 3.10. Difference between the 2017 and 2016 F the Amazon and Orinoco Rivers are likely the source zonal average monthly salinity anomalies (PSS-78) of this freshening (Figs. 3.7b and 3.8a–d) with a stron - relative to the long-term WOA13v2 monthly salinity ger 2017 North Brazil Current (when compared to climatology for 1955–2012 (Zweng et al. 2013) from 0 2016) from March through August (see Figs. 3.19b,c) to 500 m for the (a) Atlantic, (b) Pacific, and (c) Indian helping advect the freshwater river discharge farther Ocean basins. Contours are multiples of ±0.03 with a P – E (see Fig. to the north and northwest. Increased bold 0 contour. | S80 AUGUST 2018

101 Fig. 3.12d). Near the equator, salinity increased between 2016 and 2017, likely due to the aforementioned ENSO transition and upwelling of higher salinity water caused by increased zonal wind stress from 2016 to 2017 (see Fig. 3.13b). Farther north there was freshening (< −0.03) from 10° to 27°N reaching a depth of ~140 m and freshening (< −0.03) between 38° and 50°N from 0 to 150 m. Finally, there was salinification (> 0.03) between 50° and 60°N in the upper 100 m. – 2 ) anomalies (W m ig . ) for 2017 relative to F 3.11. (a) Surface heat flux (Q net The largest changes in the 2010–14 mean. Positive values denote ocean heat gain. 2017 minus 2016 zonally averaged salinity difference for (b) Q , (c) surface radiation (SW+LW), and (d) turbulent heat net between 2016 and 2017 in the fluxes (LH+SH), respectively. Positive differences denote more ocean heat Indian Ocean occurred in the gain in 2017 than in 2016. LH+SH are produced by the OAFlux high-resolution upper 100 m between 10°S (HR) satellite-based analysis, and SW+LW by the NASA FLASHFlux project. and 15°N (Fig. 3.10c). In this region, there was broad-scale salinification (> 0.03) (< −0.03) from 2016 to 2017 from 17° to 24°N extends with large increases in salinity (> 0.20) in the upper down to 350 m and is primarily caused by the near- coast freshening along India’s west coast (Fig. 3.7b). 50 m between 10°S and the equator. As was discussed previously in this section, this salinification was primarily due to the advection of salty water via 2017 e. Global ocean heat, freshwater, and momentum anomalous ocean currents (see Fig. 3.18a). Freshening fluxes— L. Yu, X. Jin, S. Kato, N. G. Loeb, P. W. Stackhouse, R. A. Weller, and A. C. Wilber The ocean and the atmo - - sphere communicate via in terfacial exchanges of heat, freshwater, and momentum. These air–sea f luxes are the primary mechanisms for keeping the global climate system nearly balanced with the incoming insolation at Earth’s surface. Most of the - shortwave radiation (SW) ab sorbed by the ocean’s surface is vented into the atmosphere by three processes: longwave radiation (LW), turbulent - heat loss by evaporation (la tent heat f lux, or LH), and conduction (sensible heat − 1 3.12. (a) Surface freshwater ( ) flux anomalies (cm yr ) for 2017 rela F ig . - P – E f lux, or SH). The residual tive to a 1988–2014 climatology. 2017 minus 2016 tendencies for (b) , (c) P – E heat is stored in the ocean and ). Green colors denote anomalous ), and (d) precipitation ( P evaporation ( E - transported by ocean circula ocean moisture gain and browns denote loss, consistent with the reversal of tion, forced primarily by the the color scheme in (c). P is computed from the GPCP version 2.3 product, momentum transferred to and E from OAFlux-HR satellite-based analysis. | S81 AUGUST 2018 STATE OF THE CLIMATE IN 2017

102 wave And Shortwave Radiative Fluxes (FLASHFlux; https: //ceres.larc.nasa.gov/products .php?product=FLASHFlux) Ed3A product (Stackhouse et P al. 2006). Global is from the Global Precipitation Clima - tology Project (GPCP; http: //gpcp.umd.edu) version 2.3 products (Adler et al. 2003). The CERES Energy Balanced and Filled (EBAF) surface SW and LW version 4.0 products (http://ceres.larc.nasa.gov; Loeb et al. 2018) are used in the time series analysis. urface 1) s heat fluxes The dominant feature in 2 – 3.13. (a) Wind stress magnitude (colors) and vector anomalies (N m ) . ig F anomalies (Fig. the 2017 Q net for 2017 relative to a 1988–2014 climatology, (b) 2017 minus 2016 differences 3.11a) is the broad-scale ocean 1 – in wind stress, (c) Ekman vertical velocity (W ) anomalies for 2017 ; cm day EK heat gain anomalies (positive relative to a 1988–2014 climatology, and (d) 2017 minus 2016 differences in ) generally exceeding Q net W . In (c) and (d), positive values denote upwelling favorable differences, and EK −2 10 W m in the tropical values are not presented within ±3° of the equator. Winds are computed from Pacific and Indian Oceans. the OAFlux-HR satellite-based vector wind analysis. The anomaly pattern in the the ocean by wind stress. Evaporation connects heat equatorial Pacific is associated with La Niñas at and moisture transfers, and the latter, together with the end of both 2016 and 2017 bracketing neutral - precipitation, determines the local surface freshwa conditions in between, so the central and eastern ter f lux. Identifying changes in the air–sea f luxes is tropical Pacific were cooler than normal both at the essential in deciphering observed changes in ocean beginning and end of 2017. Convection was located circulation and its transport of heat and salt from the more in the far western Pacific and less in the central tropics to the poles. Pacific, and SSTA was lower in the equatorial central Air–sea heat f lux, freshwater f lux, and wind stress and eastern Pacific (see Fig. 3.1a). The 2017 minus in 2017 and their relationships with ocean surface 2016 net downward radiation (SW+LW) differences variables are examined here. The net surface heat (Fig. 3.11c) were positive (ocean heat gain) in the , is the sum of four terms: SW + LW + LH f lux, Q central Pacific and negative in the far west. The net + SH. The net surface freshwater f lux into the ocean maximum SW+LW differences were centered near (neglecting riverine and glacial f luxes from land) is the dateline and extended both to the east along the simply precipitation ( ) minus evaporation ( E ), or the P ITCZ location and to the southeast into the South – P E f lux. Wind stress is computed from satellite wind Pacific. This difference pattern is consistent with retrievals using the bulk parameterization of Edson P difference pattern (Fig. 3.12d), showing that the et al. (2013). The production of the global maps of SW+LW increased in area of reduced ITCZ rainfall P E – Q (Fig. 3.12), and wind stress (Fig. (Fig. 3.11), and, conversely, SW+LW reduced in area of increased net 3.13) and the long-term perspective of the change of ITCZ rainfall. the forcing functions (Fig. 3.14) integrate multigroup - The ocean turbulent heat loss (LH+SH) tenden E , and wind stress are efforts. Ocean-surface LH, SH, cies from 2016 to 2017 increased (positive anomalies, from the Objectively Analyzed air–sea Fluxes (OAFlux; blue colors) in the equatorial and southeast tropical http://oaf lux.whoi.edu) project’s newly developed Pacific (Fig. 3.11d). The enhanced turbulent heat loss satellite-derived, high-resolution (hereafter OAFlux- - tendencies were related primarily to the strengthen HR) products (Yu and Jin 2012, 2014, 2018). Surface SW ing tendencies of the southeast trade winds from 2016 and LW radiative f luxes are from the Clouds and the to 2017 (Fig. 3.13b). Winds facilitate the efficiency of Earth’s Radiant Energy Systems (CERES) Fast Long - evaporation; stronger winds usually generate stronger | S82 AUGUST 2018

103 eastern Pacific and an increase in the Indo-Pacific warm pool (Fig. 3.12a–d). This pattern is primarily in response to the far westward location of tropical convection in association with La Niñas at the ends of both 2016 and 2017. The 2016 to 2017 ITCZ rainfall tendencies (Fig. 3.12d) were reduced considerably in the central equatorial Pacific, with major deficit tendencies (brown) of 1 m and greater centered near the dateline, stretching both eastward and southeast - ward. At the same time, the far western Pacific and the eastern equatorial Indian Ocean received more rainfall (green) in 2017 than 2016. The pattern ref lects the enhanced regional deep convection associated with La Niña. The central tropical Pacific along the ITCZ had a net freshwater deficit by about 0.5 m and the Indo-Pacific warm pool region had a freshwater gain by about 0.3 m relative to climatology (Fig. 3.12a). The tropical Indian Ocean in 2017 experienced a basin-wide reduction in evaporation, leading to less . F ig 3.14. Annual-mean time series of global average moisture loss from the region, or equivalently, more – 2 ; W m anomalies of (a) net surface heat flux (Q ) net freshwater gain (Fig. 3.12c). There were widespread from the combination of CERES EBAF4.0 SW+LW and sea surface cooling tendencies, which appear to be OAFlux-HR LH+SH, relative to 2000–14. The 2017 Q net tendencies. E a leading factor causing the reduced (denoted by star) is estimated from FLASHFlux and Wind may contribute too, as the cooling was also – 1 OAFlux-HR. (b) net freshwater flux ( E P – ) from ; cm yr accompanied by weakened surface winds in the , and (c) E and OAFlux-HR P the combination of GPCP – 2 western Indian Ocean. Away from the Indian Ocean, wind stress magnitude (N m ) from OAFlux-HR. Both and wind stress magnitude anomalies are relative – P E however, tendencies were governed mostly by wind E to 1988–2014. Gray error bars denote one std. dev. of tendencies. For instance, the enhanced E tendencies annual-mean variability. in the tropical and South Pacific were associated with enhanced wind tendencies, and the weakened - latent heat loss. In the South Pacific, LH+SH tenden E - tendencies in the North Pacific were associated cies increased in places where wind speeds strength with weakened wind tendencies. A similar correla - ened and vice versa. On the other hand, wind is not tion pattern is also observed in the Atlantic basin the only factor controlling LH+SH. The pronounced basin-wide reduction of LH+SH in the tropical In - from 45°S to 75°N. The increase of in the tropical E dian Ocean was governed by the basin-wide surface and subpolar North Atlantic and the reduction of E cooling tendencies (see Fig. 3.1b). The surface cooling in both the north and south subtropical basins were primarily the work of wind. was also responsible for the reduction of the LH+SH −2 tendencies (by ~10 W m ) in the North Pacific be - 3) w ind stress tween 30° and 60°N. The 2017 wind stress anomalies were most pro - Except for the equatorial Pacific, the 2016 to 2017 nounced in the extratropical regions (Figs. 3.13a,b). Q tendencies were predominantly determined by net Large negative anomalies were generally associated LH+SH tendencies. During 2017, the ocean gained - with the weakening of westerly winds in the midlati more heat in the tropical Indian Ocean, the equato - rial Pacific, and the extratropical Pacific and Atlantic tude North Pacific (30 °–50°N), the subpolar North Atlantic, and along the Antarctic Circumpolar Cur between 30° and 60° north and south, while the ocean - - - °–60°S). In the equa rent in the Southern Ocean (30 lost heat in the western Pacific, the south and south torial region, enhanced deep convection associated east Pacific, and the subpolar North Atlantic. Net with La Niña conditions strengthened the surface heat loss tendencies south of 60°S may be inf luenced by sea ice edge effect on f lux estimates. branch of the Walker circulation (Rasmusson and Carpenter 1982) from 2016 to 2017, with stronger 2) s easterly anomalies in the central equatorial Pacific. freshwater fluxes urface E Meanwhile, there was a strengthening of the south P The 2017 - – anomaly pattern is characterized by a reduction of net freshwater input in the central and east trade winds in the South Pacific and a weakening | S83 AUGUST 2018 STATE OF THE CLIMATE IN 2017

104 of the northeast trade winds in the North Pacific from by a period of no trend in the 2000s. The time series 2016 to 2017. and showed also a dip in 2009, lagging behind Q net – Winds vary considerably in space. These spatial P by one year, and the mean level of the global E - winds was slightly lower thereafter. variations of winds cause divergence and conver gence of the Ekman transport, leading to a vertical f. Sea level variability and change— P. R. Thompson, velocity, denoted by Ekman pumping (downward) or suction (upward) velocity W , E. Leuliette, W. Sweet, D. P. Chambers, M. A. Merrifield , at the base of the EK W Ekman layer. Computation of - B. D. Hamlington, S. Jevrejeva, J. J. Marra, G. T. Mitchum, R. S. Nerem, and follows the equa EK = 1/ f W tion: ρ∇× ( M. J. Widlansky / ρ i ), where s the density and τ f EK Global mean sea level (GMSL) during 2017 became tendencies W the Coriolis force. The 2016 to 2017 EK the highest annual average in the satellite altimetry (Fig. 3.13d) reveal stronger downwelling (negative) record (1993–present), rising to 77 mm above the 1993 °– anomalies in the vicinity of the ITCZ location (3 5°N) and stronger upwelling (positive) anomalies in average (Fig. 3.15a). This marks the sixth consecutive year (and 22nd out of the last 24) that GMSL increased the eastern equatorial Indian Ocean. Outside of the relative to the previous year. The new high ref lects tropical region, the weakened westerly band in the - midlatitude North Pacific induced a band of upwell the ongoing multidecadal trend in GMSL during the −1 satellite altimetry era, 3.1 ( (Fig. 3.15a), 2) mm yr ing anomalies (positive) to its south and a band of ±0. downwelling anomalies (negative) to its north. In the which continues unabated despite relaxing to neutral anomalies were characterized by W North Atlantic, to weak La Niña conditions following the 2015/16 El EK a tripole pattern, with positive upwelling anomalies at Niño (see Section 4b). °–60°N) and negative downwelling midlatitudes (40 An early section of the GMSL time series from anomalies at subpolar latitudes (poleward 60°N) as satellite altimetry (1993–99) was updated during 2017 °–30°N). Negative to ref lect a reevaluation of an algorithm designed to well as the subtropical latitudes (5 detect drift in the altimeter measurements due to downwelling anomalies dominated the ocean basins thermal or other changes in the internal hardware in the Southern Hemisphere (Fig. 3.13d). of the instrument (Beckley et al. 2017). Applying this correction lowers the global mean rate of change over 4) l G - term perspective on −1 , and wind , P – E and increases Annual-mean time series of Q the altimeter record by 0.3 mm yr net quadratic acceleration (i.e., two times the quadratic stress anomalies averaged over the global ice-free coefficient in a second-order polynomial fit) to 0.1 oceans (Figs. 3.14a–c) provide a decadal perspective −2 ) mm yr - (±0.04 . When effects of the Pinatubo vol on the ocean surface forcing functions in 2017. The canic eruption and ENSO are subtracted from GMSL Q time series from 2001 to 2016 were constructed net variability, the estimated climate-change-driven ac from the CERES EBAF4.0 surface radiation and - celeration in GMSL over the altimeter record is 0.084 OAFlux-HR turbulent heat f luxes, and the Q in net −2 ) mm yr (±0.025 2017 was diagnosed from FLASHFlux and OAFlux- (Nerem et al. 2018). forcing functions were down slightly in 2017. Variations in GMSL (Fig. 3.15a) result from chang - HR. Q net es in both the mass and density of the global ocean Recalling that the ocean has absorbed heat at an −2 average rate of about 0.7 W m (Leuliette and Willis 2011; Chambers et al. 2017). from 1993 through From 2005 to present, 2017 (see Section 3c), the time series of Q anomalies increasing global ocean mass net indicate that the heat budget over the global ocean observed by the NASA Gravity Recovery and Climate Experiment (GRACE) accounted for approximately displays contrasting patterns before and after 2008. −1 . The two thirds of the GMSL trend, 2.5 (±0.4) mm yr Q was relatively constant from 2001 to 2007, but net became after a sharp increase from 2008 to 2010, Q positive trend in ocean mass primarily resulted from net more volatile on interannual time scales. The time melting of glaciers and ice sheets (see Sections 2c, 5e, series of – E anomalies and wind stress anomalies P and 5f), but these contributions from land ice were partially offset by increased hydrological storage of - are about 30 years long, starting from 1988 when Spe fresh water on land, accounting for −0.7 (±0.2) mm cial Sensor Microwave Imager observations became −1 experienced (Reager et al. 2016) of sea level change. Steric (i.e., available. Over the 30-year period, yr P – E −1 , as density-related) sea level rise of 1.3 (±0.2) mm yr a slight decrease during the 1990s but had no obvious E observed by the Argo profiling f loat array, is mostly trend in the 2000s. The P – dip in 2008 coincided well with that of Q , suggesting a coherence between owing to ocean warming and accounts for the balance net P over the global ocean. A strengthening – E and Q of the GMSL trend since 2005. net of the global winds in the 1990s is observed, followed | S84 AUGUST 2018

105 ig . 3.15. (a) Monthly averaged global mean sea level (cm; black line) observed by satellite altimeters (1993–2017) F from the NOAA Laboratory for Satellite Altimetry relative to the start of the altimeter time series in late 1992. Monthly averaged global ocean mass (blue line; 2003–Aug 2017) from GRACE. Monthly averaged global mean steric sea level (red line; 2004–17) from the Argo profiling float array. Mass plus steric (purple line). All 1 − time series have been smoothed with a 3-month filter. (b) Linear sea level trends (cm yr ) from altimetry 1 − during 1993–2017. (c) Linear sea level trends (cm yr ) from altimetry during 2012–17. GMSL from altimetry observations increased of decadal oscillations in the process—the long-term from 2016 to 2017 by 0.9 mm (Fig. 3.15a). A majority trend map has become more uniform (Fig. 3.15b). of this moderate increase resulted from warming of During the altimetry era, rates of sea level change in the Indian Ocean and western tropical Pacific (3 to the upper 400 m of the global ocean (see Fig. 3.5) −1 ) generally exceed the global average rate and a 0.6 mm increase in global mean steric sea 7 mm yr −1 ) are level observed by Argo (Fig. 3.15a). Unfortunately, while rates in the eastern Pacific (1 to 3 mm yr below the global mean. East–west trend differences the mass contribution to the year-over-year increase cannot be directly ascertained due to failure of an across the Pacific result from f luctuations in trade winds, which strengthened during a multidecadal accelerometer on board one of the GRACE satellites, which resulted in only five months of valid GRACE trend toward the La Niña-like phase of the PDO during the first 15–20 years of the satellite record observations during 2017 (January and March–June). (e.g., Merrifield 2011). More recently, a trend toward Global ocean mass averaged over these five months of El Niño-like conditions during 2012–17 resulted in a data (after removing the mean seasonal variability) decreased relative to the 2016 average equivalent to 4.8 reversal of the Pacific zonal trend pattern (Hamling - (±2.4) mm in sea level change, where the uncertainty ton et al. 2016) with a majority of the eastern Pacific represents one standard error accounting for the experiencing rates of sea level rise greater than 10 −1 (Fig. 3.15c). Over the same period, positive mm yr different number of data points in the 2016 and 2017 averages and the substantial monthly uncertainties in sea level trends in the North Atlantic subtropical gyre opposed negative sea level trends in the subpolar gyre the 2017 data (Fig. 3.15a). Closing the sea level budget (Fig. 3.15c). The recent North Atlantic trends suggest requires a small (~0.3 mm) sea level equivalent for increasing gyre strength, which is coincident with year-over-year increase in global ocean mass. Thus a positive trend in the North Atlantic Oscillation we can infer an approximate increase in global ocean (NAO). The tendency toward a positive NAO state in mass equal to roughly 5 mm of sea level rise over the second half of 2017 to recover mass lost from the recent years represents a reversal of decadal trends in the Atlantic observed in leading modes of wind-stress ocean early in the year. curl over the basin throughout the 1990s and 2000s Regional sea level trends can differ substantially (Häkkinen et al. 2013). - from the global mean trend, but as the altimetry re cord has grown in length—capturing complete cycles | S85 AUGUST 2018 STATE OF THE CLIMATE IN 2017

106 south of the equator (see Fig. 3.13d) and are consistent with the tendency toward enhanced westward geostrophic surface currents in the region of the Indian Ocean South Equato - rial Current (see Fig. 3.18b). Increasing trades in the region create more negative wind- stress curl near the equator, which in turn enhances sub - tropical and cross-equatorial overturning cells (e.g., Miyama et al. 2003) that can lead to near-uniform variability in sea - level across the ENIO (Thomp ig 3.16. (a) Annual average sea level anomaly during 2017 relative to the . F son et al. 2016). average sea level at each location during 1993–2017. (b) Average 2017 sea Ongoing trends and year- level anomaly minus 2016. (c) Average sea level anomaly during Dec 2016– to-year changes in sea level Feb 2017 (DJF) relative to the DJF average during 1993–2017. (d) Same as (c), impact coastal communities by but for Sep–Nov 2017. GMSL was subtracted from (c) and (d) to emphasize increasing the magnitude and regional, non-secular change. Altimetry data were obtained from the gridded, frequency of positive sea level multimission product maintained by the Copernicus Marine and Environment extremes that cause f looding Monitoring Service (CMEMS). and erosion. In many areas, Positive annual sea level anomalies spanned most coastal infrastructure is currently exposed to nuisance- level (i.e., minor-impact) f looding when water levels of the global ocean during 2017 (Fig. 3.16a), which primarily ref lects the global mean trend relative to exceed a threshold defined by the top 1% of observed daily maxima from a global network of tide gauges the 1993–2017 climatology. The change in annual (Sweet et al. 2014). These thresholds vary geographi - mean sea level from 2016 to 2017 (Fig. 3.16b; similar to OHC, see Fig. 3.4b) shows an increase in the western cally (Fig. 3.17a) but are typically around 0.5 m above tropical Pacific and decrease in the eastern tropical mean higher high water (MHHW)—the average of ob - served daily maxima—and are expected to be exceeded - Pacific. This zonal pattern is consistent with the con tinuation of ENSO-neutral to weak La Niña condi 3–4 times per year. Most locations along the U.S. East - tions during most of 2017 in contrast to 2016, which Coast experienced greater-than-expected numbers of began in a strong El Niño state. Despite decreased exceedances during 2017 (Fig. 3.17b), due in part to trade wind strength over the North Pacific from mean sea level trends in the region that exceed the glob - al mean rate at multiple time scales (Figs. 3.15b,c). The 2016 to 2017 (see Fig. 3.13b), sea level increased in the western North Pacific due in part to downwelling (i.e., number of events over most of the U.S. East Coast in positive sea level) oceanic Rossby waves propagating 2017 decreased relative to 2016 (Fig. 3.17c), however, as mean sea levels decreased year-over-year in the western east to west across the basin during the year (Figs. North Atlantic (Fig. 3.16b). Year-over-year increases 3.16c,d). These propagating anomalies were likely - reinforced by a tendency toward downwelling Ek in threshold exceedances occurred across the tropical North Pacific and along the western coast of South man pumping anomalies at low latitudes north of the America (Fig. 3.17c). The increase in the North Pacific, equator (see Fig. 3.13d). A similar tendency toward including Hawaii (see Sidebar 3.2), is at least partially downwelling Ekman pumping occurred south of the equator, which likely accounts for the increase in sea related to the positive sea level anomalies propagating across the region during 2017 (Figs. 3.16c,d). In con - level in the western South Pacific. trast, increased exceedances along the western coast of Tendencies in the Indian Ocean show decreased South America cannot be readily attributed to mean sea level from 2016 to 2017 across the equatorial and north Indian Ocean (ENIO) while sea levels increased sea level change, as neither trends nor annual sea level anomalies are anomalous in the region. in the south Indian Ocean subtropical gyre region (Fig. 3.16b). The Indian Ocean tendencies are most likely linked to an increase in trade wind strength | S86 AUGUST 2018

107 g. Surface currents— R. Lumpkin, G. Goni, and K. Dohan This section describes ocean surface current changes, transports derived from ocean surface cur - rents, and features such as rings inferred from surface currents. Surface currents are obtained from in situ (global array of drogued drifters and moorings) and - satellite (altimetry, wind stress, and SST) observa tions. Transports are derived from a combination of sea height anomaly (from altimetry) and climatologi - cal hydrography. See the State of the Climate in 2011 report for details of these calculations. Zonal surface current anomalies are calculated with respect to a 1992–2007 climatology and are discussed below for individual ocean basins. acific o 1) p cean In 2017, the Pacific basin exhibited annual mean −1 zonal westward current anomalies of 5–20 cm s at 1 °–7 °N across much of the basin (Fig. 3.18a), −1 with peak values of 20 cm s at 6 °–145 °W. °N, 125 This anomaly had two proximate causes: a north - ward shift of the North Equatorial Countercurrent (NECC), normally centered on 7 °N but in 2017 on 8°N; and a strengthened northern South Equatorial −1 Current (nSEC) at 2 westward °N, which is 50 cm s −1 in the annual climatology but averaged 60 cm s in F ig . 3.17. (a) Nuisance-level flooding thresholds defined by the level of the top 1% of observed daily maxima during 1998–2016 from tide gauge records. Units are in meters above mean higher high water (MHHW) calculated over 1998–2016. (b) Number of daily maxi - mum water levels during 2017 above the thresholds in (a). Small, black circles in (b) and (c) indicate a value of zero. (c) Same as in (b), but for 2017 minus 2016. Daily maximum water levels were calculated from hourly tide gauge observations obtained from the University of Hawaii Sea Level Center Fast Delivery database. Only records with at least 80% completeness during 1996–2016 and 80% completeness during 2017 were F ig . 3.18. Annually averaged geostrophic zonal surface analyzed. –1 ) for (a) 2017 relative to a current anomalies (cm s 1992–2007 climatology and (b) 2017 minus 2016 zonal –1 currents (cm s ) derived from a synthesis of drifters, altimetry, and winds. | S87 AUGUST 2018 STATE OF THE CLIMATE IN 2017

108 A KAI: FLOODING IN HAWAII CAUSED BY A ‘ NU SIDEBAR 3.2: — “STACK” OF OCEANOGRAPHIC PROCESSES H. YOON, M. J. WIDLANSKY, AND P. R. THOMPSON Hawaiian Islands recorded water levels that were 20–30 cm The Hawaiian Islands experienced record-high sea levels above the astronomical tidal predictions (levels expected based during 2017 (Fig. SB3.3; also see Fig. 3.17b), which caused only on the Earth–Moon–Sun’s gravitational pull upon the nuisance flooding in vulnerable coastal communities and ex - ocean). On 21 August 2017, the tide gauge in Honolulu Harbor acerbated beach erosion, especially around times of highest observed the highest hourly water level, 99 cm above mean tides. During April and May, tide gauges throughout the main lower low water (MLLW), since records began in 1905. In addition to this new maximum, the gauge registered an unprecedented number of high-water events. In the 112-year record, sea level in Honolulu Harbor exceeded 90 cm above MLLW—the approximate level of nuisance flooding—on just 40 days; 15 of those days occurred during 2017. The highest observed water levels generally occurred under clear skies with wind and rain playing little to no role in the flooding impacts. - The repeated flooding events on fair-weather days dur ing 2017 generated substantial scientific and media interest. - Near-real-time analyses of tide gauge observations and sat ellite-based altimetry measurements of sea surface heights around Hawaii allowed the above-normal sea levels to be broken down into five primary contributions: 1) long-term - sea level rise, 2) seasonally high astronomical tides, 3) oce anic planetary waves, 4) oceanic mesoscale eddies, and 5) an inverse-barometer effect of low atmospheric pressure on the ocean. This combination of processes contributing ‘ a to high sea levels in Hawaii has become known as Nu Kai, which means “piled ocean” in the Hawaiian language and helps to convey the sense of multiple components of sea level variability stacking together to produce tangible impacts at the coast. Any one component of Nu ‘ a Kai would not have caused extreme high sea levels on its own. Long-term sea level rise relative to the National Tidal Datum Epoch (NTDE; 1983– SB3.3. Daily maximum hourly sea levels (cm) for ig F . 2001), on which tidal datums are based, is approximately 4 Honolulu, HI. Tide predictions (blue) are based on harmonic cm near Hawaii. However, sea level rose during the 20th analysis of the Honolulu Harbor sea level recorded during century more than 12 cm prior to the current NTDE, and the NTDE (1983–2001). Predictions and observations none of the 2017 events would have reached nuisance- (orange) are with respect to the MLLW datum. Residuals flood levels without this long-term trend. Large-scale between the observed sea levels and the tide prediction are shown by the green line. For reference, the 90-cm water (> 500 km in zonal width) anomalies raised background sea level is shown (dashed horizontal line), which is a threshold level around the Hawaiian Islands during most of 2017 (see for coastal impacts near Honolulu. Figs. 3.16c,d), including 13 and 16 cm above normal during °N. On the equator, eastward anomalies of 2017. Because 2016 on average had similar westward 8°–12 −1 anomalies over a broader area, the 2017 minus 2016 10–15 cm s were at 0 °−1 °S, associated with the pres - −1 ence of a ~10 cm s equatorial surface countercurrent tendencies (Fig. 3.18b) do not show this feature but not present in the 1992–2007 climatology. instead exhibit narrow bands of eastward anomaly From the start of 2017, all of the major anomalies differences. The northward shift of the NECC also −1 produced eastward anomalies of ~10 cm s at present in the annual average anomaly map were | S88 AUGUST 2018

109 the highest water level events of April and August, respectively the highest tides of July, a cyclonic eddy near Honolulu lowered (highlighted in Fig. SB3.3). These anomalies were associated sea level by 6 cm (Fig. SB3.4), which cancelled much of the with an oceanic Rossby (planetary) wave, which propagated large-scale contribution to Nu a Kai (the inverse-barometer ‘ slowly westward across the tropical North Pacific over the effect was negligible). As a result, at no time during July did past year. In addition, a series of energetic mesoscale eddies the Honolulu tide gauge record water levels above 90 cm and impinged on Hawaii, either elevating or lowering sea levels no significant coastal impacts occurred, unlike during April depending on whether they were anti-cyclonic or cyclonic. and August when astronomical tide cycles were less extreme. Anti-cyclonic eddies near Honolulu during flood events in There is a pressing need to understand the causes and April and August raised sea levels by 10 and 5 cm, respectively implications of the unprecedented number of high sea level (Fig. SB3.4). Low atmospheric pressure made additional, non- events across Hawaii during 2017, because increasing global ‘ negligible contributions to Nu a Kai via the inverse-barometer mean sea level rise will increase the frequency of such periods effect during April and August, increasing water levels by an of coastal flooding in the future (see Fig. 3.15a). Whereas the additional 4 and 2 cm. In total, the sum of all non-tidal compo - a Kai can be reasonably magnitude of each component of Nu ‘ nents of the highest levels during April and August was close well-quantified in hindsight, a number of questions remain con - to 30 cm above MLLW, but the relative contributions of each cerning the forcing, duration, and probability of reoccurrence. component differed in each case. ‘ a Most importantly, the largest non-tidal components of Nu The coastal impacts of the non-tidal contributions to Nu ‘ a Kai are oceanic Rossby waves and mesoscale eddies. Eddies Kai described above (e.g., overwash and erosion of the famed can be difficult to predict, which makes it challenging for those Waikiki Beach) occurred during seasonally large astronomical a Kai to prepare for the events. Researchers ‘ affected by Nu high tides, which contributed another 7 and 16 cm above the at the University of Hawaii are working toward high sea level - average tidal range in April and August, respectively. Astro forecast products, which will be improved by more complete a Kai, but the ‘ nomical tides are a necessary component of Nu dynamical understanding. By the end of 2017, Honolulu sea level highest sea levels of 2017 did not occur during the month with anomalies decreased relative to earlier in the year, but levels the highest high tides of the summer (July; Fig. SB3.3). During remained above normal. F ig . SB3.4. Daily mean sea surface height anomalies (cm) from satellite altimetry (CMEMS/AVISO). The spatial mean has been removed over the domain of each map to highlight mesoscale structure. - already established. The evolution of these features countercurrent. These anomalies intensified in Feb −1 through the year (Fig. 3.19) was as follows: ruary to maximum values of 40 cm s and persisted −1 through March. While the countercurrent did not Eastward anomalies of 20–33 cm s were pres - weaken significantly in April, anomalies were weaker ent in the central and eastern equatorial Pacific in °–155 °S, 115 °−1 January 2017, with peak values at 0 because the climatological f low reverses to eastward °W, - associated with the anomalous surface equatorial at these latitudes in April. This situation was main | S89 AUGUST 2018 STATE OF THE CLIMATE IN 2017

110 Kuroshio has remained close to its climatological latitude. These shifts ref lect a decadal stable/unstable oscillation of the Kuroshio jet (Qiu and Chen 2005), which shifts to the north when it intensifies and be - comes stable, thus lowering eddy kinetic energy (EKE). Averaged in the downstream °N Kuroshio jet region 32 °–38 141 °–153 °E, (Qiu and Chen 2005) EKE was low in 1994/95, elevated in 1999–2001, low in 2002/03, reached a peak in 2005, and then decreased from 2009 to 2015. Since 2015, EKE has remained relatively F . 3.19. Seasonally averaged zonal geostrophic surface current anomalies ig –1 steady (at interannual time (cm s ) with respect to a 1992–2007 seasonal climatology, for (a) Dec 2016– scales) and somewhat lower Feb 2017, (b) Mar–May 2017, (c) Jun–Aug 2017, and (d) Sep–Nov 2017. than the 1993–2017 average in the downstream Kuroshio jet region, while exhibiting tained through June. In July, when the climatological equatorial current reverses again to westward, the intra-annual variations such as a short-lived increase - f low was near zero (an eastward anomaly of 5–10 cm in mid-2016. During 2017, EKE in the region aver −1 2 −2 ). By August, the equatorial current was westward aged 0.10 m s s compared to the 1993–2017 average 2 −2 . of 0.18 m s (no longer a countercurrent) and remained close to Changes in the equatorial Pacific current system climatology for the remainder of the year. that advect surface waters across the basin result in The northward shift of the NECC was already anomalies in the SST fields. The behavior of the sur present in January 2017. By March, core speeds of - −1 the NECC were a maximum of 25–30 cm s face current anomalies in this region is an indicator at 8 °N, −1 compared to 20 cm s of upcoming SST anomalies with surface current - °N in the March climatol at 7 ogy. Eastward velocity associated with the NECC ex - anomaly behavior leading SST anomalies by several °N in climatol - months. This leading nature can be seen in the first tended north to 11 °N, compared to 9.5 ogy. Anomalies with respect to the April climatology were dramatic, as the NECC maintained its intensity through the month when it weakens in climatology. - However, the NECC then rapidly weakened, be coming anomalously weak (westward anomalies of −1 ~5 cm s ) in May. For the remainder of the year, the NECC was close to its climatological strength, with a core latitude also matching climatology in June–August and a shift to the north with respect to climatology in September–December. In all months of 2017, the northern branch of the SEC was faster to the west than in climatology, with little seasonal variation in the magnitude of this anomaly as the current followed typical sea - − 1 ig . 3.20. Principal EOF of surface current (SC; m s ) F sonal variations, weakening from March to May and and of SST anomaly (°C) variations in the tropical strengthening from June to October. Pacific from the OSCAR model (Bonjean and Lagerloef State of the Climate As noted in earlier reports (e.g., 2002; www.esr.org/enso_index.html). (a) Amplitude Dohan et al. 2015), the Kuroshio was shifted anoma - time series of the EOFs normalized by their respective lously northward in 2010–14, although this shift di - std. dev. (b) Spatial structures of the EOFs. minished during 2014. During 2015 through 2017, the | S90 AUGUST 2018

111 principal empirical orthogonal function (EOF) of The changes in transport and location of several surface current (SC) anomaly and separately the first key surface currents and mesoscale rings associated EOF of SST anomaly in the tropical Pacific basin (Fig. with them in the Atlantic Ocean basin are continu - ously monitored using satellite altimetry observations 3.20). The maximum correlation between SC and SST anomalies is R (www.aoml.noaa.gov/phod/altimetry/cvar/index = 0.65 for 1993–2017, with SC leading .php). During 2017, the number of rings shed by the SST anomalies by 76 days. The year 2017 began with Agulhas and North Brazil Currents, which are partly a continued lessening of the dramatic negative SC - - indicative of Indian–Atlantic and South–North At anomalies of 2016, approaching zero values in Janu ary. Although the EOF amplitude for SC anomalies lantic water mass exchanges, respectively, remained - within their mean 1993–2017 values. The altimetry- was negative throughout 2017, this lessening coin cided with an increase in positive SST anomalies, derived transports of the Agulhas, Malvinas, Brazil, with a maximum SST EOF amplitude in March of North Brazil, and Florida Currents did not exhibit 0.9 standard deviations. As the year progressed the 2017 variations beyond one standard deviation from SC EOF amplitude decreased to a minimum of –1.9 their mean 1993–2017 values. In the southwest At - standard deviations in October. The SST EOF fol - lantic Ocean, the separation of the Brazil Current °S in lowed this trend with a steady decrease after the peak from the continental shelf break (located at 37.6 in March to a minimum of −1.2 standard deviations the mean) reveals the intrusion of subtropical waters into the subpolar region. Since 1993, this current has in December. The year ended with SC anomalies again approaching zero. separated farther to the south from the continental ° latitude (c.f., Lumpkin and Garzoli shelf break by 3 o ndian 2) i 2011; Goni et al. 2011). Compared to its mean value cean The annually averaged near-equatorial current in 2016, the separation moved to the south by about in the Indian Ocean basin is eastward, ref lecting the 2° latitude (see www.aoml.noaa.gov/phod/altimetry dominance of the Southwest Monsoon Current in /cvar/mal/BM_ts.php), the largest southward shift in the altimeter time period 1993–present. the annual average. During 2017, the mean current −1 , near the equator had peak values of 33–35 cm s h. Meridional overturning and oceanic heat transport somewhat elevated from its climatological average −1 of 25–27 cm s circulation observations in the North Atlantic Ocean— (Fig. 3.18a). Because these anomalies , were much stronger in 2017 than in 2016, the 2017 M. O. Baringer, J. Willis, D. A. Smeed, B. Moat, S. Dong minus 2016 tendencies (Fig. 3.18b) are negative on W. R. Hobbs, D. Rayner, W. E. Johns, G. Goni1, M. Lankhorst, and U. Send the equator. An examination of the month-by-month The Atlantic meridional overturning circulation development of these anomalies reveal that they re - - f lect a much stronger-than-average Southwest Mon (AMOC) and the Atlantic meridional heat transport soon Current during July–October 2017. Maximum - (AMHT) carry warm near-surface water north −1 eastward anomalies of 30–35 cm s ward, provide heat to the atmosphere at northern were observed at °S, 65 latitudes, and carry colder deep water southward. E, in August. °–85° 1°–2 Buckley and Marshall (2016) present a summary of the dynamical forcing mechanisms of the AMOC o tlantic 3) a cean Annual mean anomalies in the Atlantic Ocean - and AMHT and the role they play in regulating cli −1 mate variability around the Atlantic sector. Owing strengthening of (Fig. 3.18a) indicate an ~15 cm s to the large amounts of heat, carbon, and fresh water the eastward NECC at 5 °–6 °N, 33 °–50 °W and a 20–25 −1 cm s transported by the AMOC, climate models suggest °–1°N, strengthening of the westward nSEC at 0 25 °. The year began with the NECC anomaly °W– 0 accurate estimation of its rate of change is critical to understanding and predicting our changing climate established but with an anomalously weak (by 5–10 −1 ) nSEC (Fig. 3.19). In February, strengthening (e.g., W. Liu et al. 2017; Rahmstorf et al. 2015). Even cm s on short time scales the AMOC/AMHT can impact of the nSEC had developed east of 24 °W and spanned climate (e.g., Duchez et al. 2016). These recognitions the basin in March. These anomalies weakened through April and May, and in June–August the nSEC have led to the implementation of enhanced observing was close to climatology. Westward anomalies again systems of the strength of the AMOC in the subpolar North Atlantic (Lozier et al. 2017) and the subtropi developed in the nSEC in September and persisted - through November. In December, the nSEC was close cal South Atlantic (Ansorge et al. 2014). These new to its climatological December strength. observing systems will eventually provide a more complete spatial picture of the state of the AMOC. | S91 AUGUST 2018 STATE OF THE CLIMATE IN 2017

112 In general, estimating the AMOC/AMHT amounts to summing ocean-spanning measurements of the velocity/heat transport horizontally and vertically over the full water column. As all relevant time and space scales cannot be simultaneously measured, all the current AMOC/AMHT time series estimates include trade-offs between one quantity and another and can have errors and biases that are dependent on observing system design (e.g., Sinha et al. 2018). The systems described herein include the AMOC/ AMHT observing systems at 41°N, 26°N, and 16°N and AMHT at 41°N, 26°N, and 35°S, which represent the most complete, longest time AMOC/AMHT series currently available. Studies have shown that density anomalies along the western boundary in particular are essential predictors of the strength of the AMOC (e.g., Le Bras et al. 2017; Yashayaev and Loder 2016), and subpolar density anomalies precede those in the subtropical gyre by 8–10 years; hence, observing systems that measure western boundary variability are particularly essential. i g F 3.21. (a) Daily estimates of FC transport . The Florida Current (FC; as the Gulf Stream is called at 1 6 − 3 ( × 10 m ) during 2017 (orange solid line), 2016 s 26°N) observing system is one such example that can (dashed purple line), and 1982–2015 (light gray lines) provide a longer time perspective of possible AMOC with 95% confidence interval of daily transport values variations (e.g., Frajka-Williams 2015). Providing data computed from all years (black solid line) and the long- since 1982, this is the longest open ocean transport term mean (dashed black line). Actual sea level minus time series (Fig. 3.21). Additionally, FC and AMOC predicted sea level at the Lake Worth tide gauge station (dark green line). (b) Daily estimates of FC transport transport variations at all time scales are inversely 1 − 3 6 10 ( × s ) for the full time series record (light gray), m linked to sea level variations along the east coast (e.g., smoothed using a 12-month second-order Butterworth Domingues et al. 2016). The median FC transport filter (heavy black line), mean transport for the full ) Sv (one ±0.25 from the full record (1982–2017) is 31.9 ( record (dashed black line), and linear trend from 1982 standard error of the mean assuming a 20-day integral through 2017 (dashed blue line). Two-year low-passed time scale) with a small downward trend of −0.29 AMO (orange line) and NAO (red dashed line) indices −1 (errors estimating 95% significance ) Sv decade (±0.23 are also shown. as above). The 2017 median FC transport was 32.3 (±1.6 The FC time series contributes to the estimate ) Sv, slightly above the long-term average and the of the AMOC at 26°N (Figs. 3.22, 3.23), where the 2016 annual average. Daily FC transports compared to those of all previous years (Fig. 3.21a) indicate that AMOC is measured with full-water column moorings that span the full basin and include direct transport 2017, unlike previous years, had few unusual transport measurements in the boundary currents as part anomalies (extremes defined as outside the 95% confi - of the large RAPID-MOC/MOCHA/WBTS 26°N dence limits for daily values). During 2017 there were no high transport events and the only low transport mooring array (Smeed et al. 2017). The data from - anomaly that was sustained for more than a day oc these moorings are collected every 18 months, with AMOC data presented in this section extending from curred during 10–11 December 2017 (averaging 23.6 April 2004 to February 2017. In the latest update, Sv). The FC lagged by 8 years has its maximum positive adding data from October 2015 through February correlation with the NAO (Fig. 3.21b). The FC lagged by 2017, the AMOC has increased slightly with average 5 years has its maximum negative correlation with the AMOC of 17.2 Sv in 2016 and 17 Sv in part of 2017. Atlantic multidecadal oscillation (AMO). It continues This seeming stabilization of the downward trend in to be inversely correlated with the difference between observed and astronomically predicted sea level at the the AMOC has resulted in a statistically insignificant −1 Lake Worth tide gauge station (Fig 3.21a, significant at , ) Sv decade ±2.01 downward trend estimate of −1.99 ( −1 over the 99% significance level, correlation coefficient half the −5.3 Sv decade trend first noted in Smeed −0.57 and 28% of variance explained). et al. (2014). This trend is entirely due to the increase | S92 AUGUST 2018

113 ocean temperature and salinity for the upper 2000 m on broad spatial scales, as well as velocity at 1000 m) and altimetry-derived surface velocity (Willis 2010; Hobbs and Willis 2012) are used to estimate the AMOC (Fig. 3.23) and AMHT (not shown). This time series has been updated since last year’s report (Baringer et al. 2017), extending from January 2002 to October 2017. Near 16°N, the AMOC is estimated using a mooring array of inverted echo sounders, current meters, and dynamic height moorings (Send et al. 2011) that measure the f low below 1000 m (the southward f lowing part of the AMOC “conveyor belt”); hence, the AMOC estimate at this latitude (Fig. 3.23) is a negative number (southward deep f low) to distinguish these observations from the full water column systems. These data have not been updated since last year’s report and remain from February 2000 to September 2016. At 35°S in the South Atlantic, the AMOC and AMHT are estimated 3.22. (a) Daily estimates of the volume transport F ig . 6 3 − 1 - m ( × 10 ) of the meridional overturning circula s tion (blue line) and its components, the FC (green line), wind-driven Ekman transport (red line), and the geostrophic interior (black line), as measured by the UK National Environmental Research Council (NERC) Rapid Climate Change Program (RAPID-WATCH), the National Science Foundation’s Meridional Overturning and Heat transport Array proposal, and the NOAA Western Boundary Time Series project (WBTS). The volume transports have a 10-day low-pass filter applied to the daily values and the annual median transports for each year are shown in the associated color text (Sv). (b) The deepest part of the MOC can be divided into upper deep water (1000–3000 m; orange) and 3.23. Estimates of AMOC transports . i g F lower deep water (3000–5000 m; purple) transports − 6 1 3 10 × (1 Sv = - ) from the Argo/Altimetry esti s m − 3 1 6 ( × 10 m ). s mate at 41°N (black; Willis 2010), the RAPID-MOC/ in the southward near-surface interior f low of −2.05 MOCHA/WBTS 26°N array (orange; Smeed et al. −1 ) Sv decade (±1.35 (Fig. 3.22, black dashed line), while 2017), and the German/NOAA MOVE array at 16°N the decrease in FC transport balances the increase in (purple; Send et al. 2011) shown vs. year. All time series have a 3-month second-order Butterworth Ekman transport. The decrease in the AMOC at this low-pass filter applied. Horizontal lines are mean latitude can be explained by the decreased export, transports during similar time periods as listed in the −1 ±1.07 1.65 ( , of upper North Atlantic Deep ) Sv decade corresponding text. Dashed lines are trends for each Water in the 1–3 km depth range, while the lowest series over the time period with data available for all layer remains fairly stable. Change-point analysis three series (Apr 2004 through Aug 2016). For MOVE shows that the AMOC time series trend is not linear data, the net zonal and vertical integral of the deep but rather consists of a significant break or jump in circulation represents the lower limb of the AMOC (with a negative sign indicating southward flow), and the time series in 2008 (Smeed et al. 2018), and the hence a stronger negative (southward) flow represents - baseline shift toward decreased AMOC was concur an increase in the AMOC amplitude. Light gray lines rent with changes of a southward shifting Gulf Stream show ECCO2-derived transports (Menemenlis et al. and large-scale changes of sea surface temperature, 2008): (top) thin gray is the 41°N transport, thick gray sea surface height, and heat content. - is the 26°N transport, (bottom) the negative meridi AMOC estimates are also provided at 41°N, where onal overturning circulation in the model shown for a combination of profiling Argo f loats (that measure ease of comparison with the 16°N data. | S93 AUGUST 2018 STATE OF THE CLIMATE IN 2017

114 using a combination of high-density (closely spaced) expendable bathythermograph (XBT) and broader- scale Argo profiling f loat data (Dong et al. 2014, not shown; www.aoml.noaa.gov/phod/soto/mht/ax18/ report.php). These data are collected and analyzed in near-real time, with values spanning July 2002 to September 2017. Similar to 26°N, at 41°N the AMOC and AMHT are decreasing less rapidly (Fig. 3.23), changing −1 ±2.7 ) Sv decade and −0.03 ( ±0.04 ) PW to −0.08 ( –1 −1 ) Sv decade as compared with −1.2 ( decade ±3.0 −1 − 3 ) PW decade ±0.21 reported last year. and −0.09 ( a distribution mg m F ig de - . 3.24. Annual mean Chl Farther south, the MOC/MHT trends are positive, for 2017. Also shown is Aqua rived from MODIS on the location of the mean 15°C SST isotherm (black but decreasing in the past three years as the annual a lines) delineating the boundary of the PSO. Chl data means at 16°N increased from −29.2 Sv in 2014 to are from NASA Reprocessing version 2018.0. Data −27.8 Sv in 2015 to −23.8 in 2016. This recent reduc - are averaged into geo-referenced equal area bins of tion in southward f low has led to a reduced estimate 2 approximately 4.6 × 4.6 km and mapped to an equi- of the long-term trend of the AMOC from February rectangular projection centered at 150°W. ±2.4 ) Sv 2000 to September 2016 at 16°N to be +3.4 ( −1 . While the 35°S AMOC transport estimate decade has remained fairly constant for the last three years (median AMOC of about 20 Sv), during 2017 it was dominated by the Ekman component whereas in pre - vious years it had been dominated by the geostrophic component. The variability at all latitudes in the Atlantic is not well correlated and, therefore, data from more than one latitude are needed to describe the state of the ocean. i. Global ocean phytoplankton— B. A. Franz , E. M. Karaköylü1, D. A. Siegel, and T. K. Westberry Marine phytoplankton contribute roughly half the net primary production (NPP) on Earth, fixing atmospheric CO into food that fuels global ocean 2 ecosystems and drives biogeochemical cycles (e.g., Field et al. 1998; Falkowski et al. 1998). Phytoplank - ton growth is dependent on availability of light and nutrients (e.g., iron, nitrogen, phosphorous) in the upper ocean euphotic zone, which in turn is inf lu - enced by physical factors such as ocean temperature (e.g., Behrenfeld et al. 2006). SeaWiFS (McClain 2009) and MODIS (Esaias et al. 1998) are satellite - ocean color sensors that provide observations of suf F ig . 3.25. Spatial distribution of average monthly ficient frequency and geographic coverage to globally (a) MODISA Chl a anomalies and (b) SST anomalies monitor changes in the near-surface concentration where monthly differences were derived relative to the MODISA 9-year climatological record (2003–11). ; a (Ch l a of the phytoplankton pigment chlorophyll- −3 Chl is expressed as % difference from climatology, a - mg m ), which serves as a proxy for phytoplank while SST is shown as an absolute difference (°C) . ton abundance. Here, global Chl a distributions for (c) identifies relationships between the sign of SST and 2017 are evaluated within the context of the 20-year anomalies from panels (a) and (b), with colors Chl a continuous record provided through the combined differentiating sign pairs and missing data masked observations of SeaWiFS (1997–2010) and MODIS in black. Also shown in each panel is the location of data a (MODISA, 2002–present). All Chl Aqua on °C SST isotherm (black lines) delineating the mean 15 used in this analysis correspond to NASA process - the PSO. | S94 AUGUST 2018

115 ing version R2018.0 (https://oceancolor.gsfc.nasa.gov light by decreasing their cellular chlorophyll levels /reprocessing/), which utilizes common algorithms (Behrenfeld et al. 2016). A secondary consequence of and calibration methods to maximize consistency in decreased MLD is a decrease in the vertical transport the multi-mission satellite record. of nutrients to the surface layer, but coupling between the MLD and nutricline depths throughout much The spatial distribution of MODISA annual a of the PSO is known to be weak (Lozier et al. 2011). for 2017 (Fig. 3.24) is consistent with the mean Chl Modestly depressed Chl a concentrations (< 10%) well-established, physically driven distribution of were also observed throughout the Sargasso Sea and nutrients (Siegel et al. 2013) and surface mixed-layer light conditions (Behrenfeld et al. 2016). Chl a values in the Mediterranean in 2017. Strongly elevated Chl a concentrations were observed in the northern reaches during 2017 ranged over three orders of magnitude, −3 in the central ocean gyres to > 20 of the North Atlantic (>30%), with weaker increases from < 0.02 mg m −3 mg m observed throughout much of the South Atlantic and in nutrient-rich coastal and subpolar waters. To assess changes in this distribution during 2017, Indian Ocean regions. Within the boundaries of the mean values for MODISA Chl in each month of the a PSO, an inverse relationship was generally observed a between Chl year were subtracted from monthly climatological and SST anomalies (light blue and gray colors in Fig. 3.25c), with some notable exceptions of means for MODISA (2003–11). These monthly fields and SST anoma were then averaged to produce the global chloro positive correlations between Chl - a - phyll anomaly map for 2017 (Fig. 3.25a). Identical lies in the South Atlantic and southwestern Pacific (dark blue colors in Fig. 3.25c). In regions outside calculations were performed on MODISA sea surface the PSO, no clear relationship was observed between temperature (°C) data to produce an equivalent SST a Chl annual mean anomaly (Fig. 3.25b), used to illustrate and SST anomalies (Fig. 3.25c), consistent with previous studies (e.g., Behrenfeld et al. 2006; Franz the relationships between Chl and SST anomalies a et al. 2017). (Fig. 3.25c). Here the permanently stratified ocean Over the 20-year time series of spatially integrated (PSO) is defined as the region where annual average monthly mean Chl a values for the PSO (Fig. 3.26a) surface temperatures are > 15°C (black lines in Figs. −3 ) mean concentrations varied by ~20% ( 3.24 and 3.25) and is characterized by surface mixed ±0.03 mg m −3 around a long-term average of ~0.14 mg m layers that are typically low in nutrients and shallower . This variability includes significant seasonal cycles in than the nutricline (Behrenfeld et al. 2006). Consistent with the establishment of weak La Niña conditions through a concentrations much of 2017, Chl - along the equatorial Pacific were neu tral to slightly elevated (<10%) above the climatological mean (Fig. 3.25a), ref lecting the return of cooler, more nutrient-rich waters conducive to phytoplankton growth. Chl a concen - trations throughout much of the tropi - cal Pacific, however, were generally diminished relative to climatological values (10%–30%) and inversely related to SST anomalies (gray areas above and - below the equator in Fig. 3.25c). An nual mean SST anomalies (Fig. 3.25b) a . ig F averaged over 3.26. 1998–2017, multimission record of Chl generally coincide with surface mixed the PSO for SeaWiFS (black) and MODISA (blue). (a) Independent layer depth (MLD) anomalies in the record from each mission, with horizontal black line indicating the PSO, with warmer temperatures as - 3 − multimission mean Chl a concentration for the region (mg m ). (b) sociated with shallower mixing, such Monthly anomaly (%) for SeaWiFS and MODISA after subtraction that phytoplankton spend more time of the 9-year MODISA monthly climatological mean (2003–11) from near the ocean’s surface and thus have each mission record. The gray region in (b) shows the averaged dif - higher daily sunlight exposures than ference between SeaWiFS and MODISA over the common mission lifetime. Green diamonds show the MEI, inverted and scaled to match deeper mixing populations. Phyto - the range of the Chl a anomalies. plankton respond to this increased | S95 AUGUST 2018 STATE OF THE CLIMATE IN 2017

116 a Chl distributions and responses to climatic events. has risen from pre-industrial levels of about 278 ppm (parts per million) to about 405 ppm (see Section The time series also demonstrates the high level of 2g1). The atmospheric concentration of CO is now consistency between the overlapping periods of the 2 SeaWiFS and MODISA missions, lending confidence higher than has been observed on Earth for at least 00 years (Lüthi et al. 2008). As discussed 0 the last 800 to interpretation of the multimission record. a Chl monthly anomalies within the PSO (Fig. reports, the global State of the Climate in previous −3 ocean is a major sink for C ) over . Here the discussion is 3.26b) show variations of ±15% (±0.02 mg m anth updated to include recent estimates of that sink. Over - the multimission time series. For 2017, these anoma lies were relatively constant and slightly elevated the last decade the global ocean has continued to take −3 up a substantial fraction of the anthropogenic carbon , on average) relative to the long-term (+0.005 mg m ) emissions and therefore is a major mediator (C mean, consistent with the weak La Niña conditions anth −1 - as discussed previously. The link between ENSO vari of global climate change. Of the 10.7 (±0.9) Pg C yr C - response in the PSO is dem a ability and mean Chl released during the period 2007−16, about 2.4 anth −1 onstrated by the correspondence of anomaly trends (±0.5) Pg C yr (26%) accumulated in the ocean, 3.0 −1 (30%) accumulated on land, and 4.7 (±0.8) Pg C yr with the multivariate ENSO index (MEI; Wolter and −1 (43%) remained in the atmosphere Timlin 1998; Fig. 3.26b, green diamonds, presented (±0.1) Pg C yr −1 (Le Quéré et al. in the inverse to illustrate the covariation). From 1997 with an imbalance of 0.6 Pg C yr 2018). This decadal ocean carbon uptake estimate is a through 2017, monthly anomalies in Chl a concentra - consensus view based on a combination of measured tion within the PSO continue to track large-scale cli - decadal inventory changes, models, and global air– mate oscillations as captured by the MEI (Fig. 3.26b), f lux estimates based on surface ocean partial sea CO with some notable deviations in the 2002–06 period. 2 - ) measurements. Using ocean CO p ( pressure of CO Variability and trends in Chl a ref lect both adjust 2 2 ments in phytoplankton biomass and physiology (or general circulation models that include biogeochemi - health). Both of these properties are mechanistically cal parameterizations (OBGCMs) and inverse models that are validated with observations-based air–sea linked to physical properties of the upper ocean, as - well as ecological relationships between phytoplank exchange f luxes and basin-scale ocean inventories, Le Quéré et al. (2018) have demonstrated that the ton and their zooplankton predators. Unraveling the diversity and covariation of factors that inf lu oceanic anthropogenic carbon sink has grown from - −1 1.0 (±0.5) Pg C yr concentrations is essential for correctly ence Chl in the decade of the 1960s to 2.6 a −1 f lux studies (±0.5) Pg C yr in 2016. Air–sea CO a anomalies on interpreting the implications of Chl 2 of 2.6 reported here indicate an ocean uptake of C ocean biogeochemistry and food webs. For example, anth −1 Pg C yr inverse relationships between Chl a f o r 2 017. and SST can emerge from changes in either mixed-layer light levels ir sea – 1) a fluxes or vertical nutrient f lux, but these two mechanisms dioxide carbon Ocean uptake of C have opposite implications on phytoplankton NPP can be estimated from the anth - f lux derived from the bulk f lux for net air–sea CO (Behrenfeld et al. 2016). An additional complication 2 mula with air–sea differences in CO partial pressure is that measured changes in ocean color often contain 2 ) and gas transfer coefficients as input. A CO (∆ p a contribution from colored dissolved organic matter 2 steady contribution of carbon from riverine runoff, (Siegel et al. 2005) that can be mistakenly attributed (Siegel et al. 2013). Thus, while the a originating from organic and inorganic detritus from to changes in Chl −1 (Jacobson et al. 2007) satellite record of ocean color continues to provide land, estimated at 0.45 Pg C yr is included to obtain the C critical insights on global processes, ongoing effort uptake by the ocean. anth p CO are annual updates of sur - and new approaches are needed to fully understand The data sources for 2 the story these data are telling regarding relationships CO p face water observations from the Surface Ocean 2 Atlas (SOCAT) composed of mooring and ship- between climate and marine ecosystems. CO 2 based observations (Bakker et al. 2016) and the LDEO database with ship-based observations (Takahashi et R. A. Feely, R. Wanninkhof, j. Global ocean carbon cycle— al. 2018). The increased observations and improved B. R. Carter, P. Landschützer, A. Sutton, and J. A. Triñanes mapping techniques such as neural network methods As a consequence of humankind’s collective emissions into the atmosphere from release of CO and self-organizing maps (Landschützer et al. 2013, 2 fossil fuel burning, cement production, and land 2014; Rödenbeck et al. 2015) provide global p CO 2 fields on a 1° latitude × 1° longitude grid at monthly use changes over the last 250 years, referred to as time scales annually. This allows investigation of (C Anthropogenic CO ), the atmospheric CO 2 2 anth | S96 AUGUST 2018

117 variability on subannual to decadal time scales. The the coast of Mauritania, and the Peruvian upwelling ∆ and a parameterization of the gas transfer system. Large sinks are observed poleward of the sub - p CO 2 tropical fronts, and the frontal position determines with wind described in Wanninkhof (2014) are used fluxes. the location of the maximum that is farther south in to calculate the air–sea CO 2 the Pacific sector of the Southern Ocean compared p The monthly 2017 ∆ maps are based on CO 2 to the other basins. an observation-based neural network approach of Landsch In the Northern Hemisphere, there is a significant zer et al. (2013, 2014). The 2017 values t ü are projections based on surface temperature, sea asymmetry in the sub-Arctic gyre with the North surface salinity, climatological mixed-layer depth, Atlantic being a large sink while the North Pacific , and the . Ocean carbon uptake anomalies is a source of CO satellite chlorophyll-α, atmospheric CO 2 2 CO neural network for developed from the data (Fig. 3.28b) in 2017 relative to a 1995–2015 average p 2w are attributed to the increasing ocean CO uptake from the previous decade. Moreover, winds from 2 with time (Fig. 3.27) and to variations in large-scale 2016 are used. A comparison of the 2016 air–sea climate modes. The air–sea f lux trend since 2000 is estimate using wind speeds from 2015, as presented −1 −0.8 Pg C decade State of the Climate report, and the 2016 , which leads to predominantly in last year’s p CO a nd 2016 global f lux using measured 2016 negative f lux anomalies (greater ocean uptake). 2w Despite this strong trend there are several regions winds show agreement on a global scale to within 0.1 −1 Pg C yr . Changes in winds over time have a small effect on global air–sea CO f luxes (Wanninkhof and 2 f luxes from 1982 to 2017 Triñanes 2017). The C anth suggests a decreasing ocean sink in the first part of the record and a strong increase from 2001 onward that continued into 2017 (Fig. 3.27). The amplitude of seasonal variability is large (≈ 1 Pg C) compared to the long-term trend with minimum uptake in the air–sea f lux of June–September timeframe. The C anth −1 in 2017 is 36% above the 2005–15 average 2.6 Pg C yr −1 of 1.9 (±0.5) Pg C yr . The average f luxes in 2017 (Fig. 3.28a) show the characteristic pattern of eff luxes in the tropical regions and uptake at higher latitudes. The region with largest eff lux is the equatorial Pacific. Localized hotspots of upwelling include the Arabian Sea, off fluxes for 3.28. Global map of (a) net air–sea CO . ig F 2 2017, (b) net air–sea CO - flux anomalies for 2017 rela 2 tive to a 1995–2015 average, and (c) net air–sea CO 2 3.27. Global annual (red line) and monthly (blue . ig F flux anomalies for 2017 minus 2016 values following − 1 line) C ) for 1982 to 2017. Negative fluxes (Pg C yr the method of Landschützer et al. (2013), all in mol anth − 1 − 2 values indicate CO uptake by the ocean. C m yr . 2 | S97 AUGUST 2018 STATE OF THE CLIMATE IN 2017

118 showing positive anomalies for 2017, notably the rise but it comes at a cost of increased acidification of surface and subsurface waters. eastern equatorial and western tropical Pacific. The cean 2) o increased eff luxes in the eastern equatorial Pacific are acidification - related to a predominant positive sign of the ENSO To date, the global oceans have absorbed ap index and the associated switch from the 2015/16 El proximately 150 (±20) Gt C of the total anthropogenic carbon dioxide emissions (Le Quéré et al. 2018). This Niño to the weak La Niñas at the ends of 2016 and uptake has caused an increase of ocean acidity in a 2017. Weaker eff luxes in the western tropical Pacific process referred to as anthropogenic ocean acidifi are related to strongly positive PDO values over the - past three years that have persisted into the first half cation (OA). Models indicate that over the last two- and-a-half centuries, the pH in open-ocean surface of 2017 and associated warmer SSTs. waters has decreased by about 0.11 units, equivalent f luxes in The differences between the air–sea CO 2 2017 compared to 2016 (Fig. 3.28c) are relatively small. to about a 29% increase in the hydrogen ion (H+) The increase in CO eff luxes in the western tropical concentration (Gattuso et al. 2015). This absorption 2 - of anthropogenic carbon is beneficial in slowing the Pacific from 2016 to 2017 is associated with anoma lously warm temperatures in this region in 2017. The rise of atmospheric carbon dioxide. However, the Southern Ocean (south of 40°S) shows a bimodal ocean’s uptake of carbon dioxide is having negative pattern with increasing f luxes in the Pacific and impacts on ocean chemistry and biology. Time series decreasing f luxes in the Atlantic for 2017 compared measurements, hydrographic surveys, and modeling to 2016. This broadly corresponds to the temperature studies have revealed that the changes in seawater are chemistry resulting from the absorption of CO anomalies in this region, with greater uptake in 2017 2 associated with warmer temperatures attributed to lowering seawater pH. For example, the time series data at Ocean Station ALOHA shows an average pH less upwelling of cold high-CO waters in the west - 2 −1 in the decrease of approximately 0.02 units decade ern and central Pacific sector of the Southern Ocean northeast Pacific (Fig. 3.29). and overall colder patterns in the eastern Pacific, Indian, and Atlantic sectors of the Southern Ocean The long-term trend at Ocean Station ALOHA associated with increased ventilation and associated of 2.0 shows an increasing rate of increase of p CO 2 −1 (Fig. 3.29a) while pH of ocean surface (±0.1) μatm yr eff luxes. The alternating patterns of stronger and −1 waters has already decreased by about 0.0016 yr , weaker uptake in the Southern Ocean are in accord with no apparent long-term change in annual CO with an asymmetric distribution of the atmospheric 2 pressure systems moving around the Southern Ocean inf lux (Fig. 3.29b; Sutton et al. 2017). The increase in surface ocean CO over recent decades is consistent associated with the Antarctic Circumpolar Wave 2 with the atmospheric increase within the statistical (Landschützer et al. 2015). - limits of the measurements. Many of the CO p and f lux anomalies can be at 2 tributed to variations in large-scale climate modes and associated physical anomalies, notably tempera - inventories arbon 3) c The Global Ocean Ship-based Hydrographic In - ture, but the causality is often complex. For example, with respect to temperature vestigations Program (GO-SHIP) is providing new the behavior of p CO 2 information about the uptake and storage of carbon includes competing processes: thermodynamics with decreasing SST, but p dictate decreasing within the ocean interior by determining the change CO 2 - waters originating from the deep with a cold tem in measured dissolved inorganic carbon (DIC) and p perature signal will have a high concentrations between decadal cruise reoc C . Moreover, the - CO anth 2 cupations. During the 2017/18 timeframe, a new p CO drawdown of due to biology is often associated 2 with increasing temperature, but this depends on set of measurements, including DIC, were finalized region and season. along the P18 line extending from San Diego south to The strong trend of increasing CO uptake since Antarctica and collected along the P06 line extending 2 2002 has continued through 2017 with an increase from Australia east to Chile. A synthesis of estimates of C storage along these sections and other recently of 0.1 Pg C above the 2016 estimate. This increase anth is well within the uncertainty of the estimate, but it measured Pacific sections is currently underway. is within the overall expectation that the ocean will While results are preliminary, the dominant signal remain an increasing sink as long as atmospheric CO is a clear and continuous increase in C storage, 2 anth especially in the least-dense and most well-ventilated by levels continue to rise. The sequestration of CO 2 the ocean partially mitigates the atmospheric CO shallower parts of the ocean (Fig. 3.30a). This C anth 2 storage is increasing ocean acidity, decreasing ocean | S98 AUGUST 2018

119 growth and changes in mixing rapid atmospheric CO 2 and ventilation within the ocean interior. The largest storages per unit area are found in high-latitude deep water formation regions such as the North Atlantic (Woosley et al. 2016), though the majority of the C anth inventory is stored in the subtropics due to the vast size of that region. Upwelling regions near the equa - tor, in the North Pacific, and in the Southern Ocean south of the Antarctic Circumpolar Current have lower decadal storages per unit area (e.g., the dark colors on the lower right Section in Fig. 3.30a). In these regions, upwelling of deep waters that have been isolated from the atmosphere for all or some of the industrial era displace the better-ventilated, higher - C intermediate depth waters. A preliminary esti anth mate of the decadal changes suggests that the Pacific basin stored 8.2 Pg C between 1995 and 2005 and 9.8 Pg C between 2005 and 2015. Ocean acidification, or the impact of C on pH, anth has a similar global pattern to the net C storage, anth though the pH decrease is amplified in seawater with 3.29. (a) Time series of atmospheric CO at Mauna F ig . 2 p Loa (ppm), surface ocean CO (μatm) and pH at 2 Ocean Station ALOHA in the subtropical North Pacific Ocean. Mauna Loa data: (ftp://aftp.cmdl.noaa .gov/products/trends/co2/co2_mm_mlo.txt); HOTS /ALOHA data: University of Hawaii (http://hahana .soest.hawaii.edu/hot/products/HOT_surface_CO2 p CO .txt). (b) Surface ocean (μatm) and rates of 2 change from Station ALOHA 1988–2013 (blue), 1989–2001 (orange), 2004–13 (red), and the adjacent WHOTS buoy 2004–13 (black) and (shaded inset) CO 2 − 2 − 1 flux (g C m ) from WHOTS buoy observations yr 2004–15 (after Sutton et al. 2017). pH (Fig. 3.30b), and decreasing carbonate mineral saturation states. For this comparison, anthropo - genic CO storage rates—estimated from decadal 2 measurements using methods described by Carter et al. (2017)—are used directly and, in some regions, extrapolated in time to estimate patterns of C anth storage since 1994. These storages are added to the climatology of Sabine et al. (2004) 1994 global C anth as gridded by Key et al. (2004). As also found in recent studies in the Atlantic (Woosley et al. 2016) and Pacific (Carter et al. 2017), these preliminary results suggest that Pacific storage rates have been increasing since ~2005 despite the tendency of water with more CO to absorb smaller 2 F ig . 3.30. Preliminary estimates of (a) anthropogenic 1 − increases (due to the fractions of atmospheric C carbon (C ; μmol kg ) along hydrographic sections anth anth in the Pacific interpolated (or extrapolated) in time to decreasing buffering capacity of seawater). The ob - the year 2015 and (b) the net impact of this C on pH. anth served storage increases are attributable to continued | S99 AUGUST 2018 STATE OF THE CLIMATE IN 2017

120 naturally high accumulated DIC from respiration by marine organisms. Such waters have a reduced buffer capacity due to their naturally high carbon concentrations, so the ongoing C storage has an anth enhanced impact on acidification. This effect can be seen in the nearly global subsurface ΔpH magnitude maximum, which is especially notable off the US West Coast where unusually high-DIC waters upwell near the surface (Fig. 3.30b). | S100 AUGUST 2018

121 4. THE TROPICS — H. J. Diamond and C. J. Schreck, Eds. Pacific, South Indian, and Australian basins were all particularly quiet, each having about half their a. Overview— H. J. Diamond and C. J. Schreck median ACE. The Tropics in 2017 were dominated by neutral El - Three tropical cyclones (TCs) reached the Saf Niño–Southern Oscillation (ENSO) conditions dur - ing most of the year, with the onset of La Niña con - fir–Simpson scale category 5 intensity level—two in the North Atlantic and one in the western North ditions occurring during boreal autumn. Although Pacific basins. This number was less than half of the the year began ENSO-neutral, it initially featured cooler-than-average sea surface temperatures (SSTs) eight category 5 storms recorded in 2015 (Diamond in the central and east-central equatorial Pacific, and Schreck 2016), and was one fewer than the four along with lingering La Niña impacts in the atmo recorded in 2016 (Diamond and Schreck 2017). - spheric circulation. These conditions followed the The editors of this chapter would like to insert two personal notes recognizing the passing of two giants abrupt end of a weak and short-lived La Niña during in the field of tropical meteorology. 2016, which lasted from the July–September season Charles J. Neumann passed away on 14 November until late December. Equatorial Pacific SST anomalies warmed con 2017, at the age of 92. Upon graduation from MIT - in 1946, Charlie volunteered as a weather officer in siderably during the first several months of 2017 and by late boreal spring and early summer, the the Navy’s first airborne typhoon reconnaissance unit in the Pacific. Later, as head of research and - anomalies were just shy of reaching El Niño thresh development at the National Hurricane Center, he olds for two consecutive, overlapping seasons of developed techniques for statistical tropical cyclone April–June and May–July. Thereafter, SSTs cooled track forecasting, error and risk analysis, and the through the remainder of the year and exceeded La compilation of a complete set of historical Atlantic Niña thresholds during September–November and October–December. hurricane tracks and intensities dating from the 1800s. These data were prototypes for the modern For the global tropics, land and ocean surfaces day best track datasets upon which so much of our combined (measured between 20°S and 20°N), the science relies. Charlie was known for his friendliness 2017 annual average temperature was 0.31°C above the 1981–2010 average. This makes 2017 the third and for his generosity in readily sharing data and warmest year for the tropics since records began in his expertise, and he was the recipient of numerous national and international awards. Please visit www 1880, behind only 2016 (+0.55°C) and 2015 (+0.53°C). .hurricanecenterlive.com/charles-newman.html for Precipitation over land for the same latitudes was above the 1981–2010 average for three major datasets more information. Professor Tiruvalam Natarajan Krishnamurti (GHCN, GPCC, GPCP), with anomalies ranging from (“Krish” to all who knew and worked with him) 45 to 125 mm above average. The dataset analyzed for tropical rainfall over the oceans (GPCP; Adler et al. passed away on 7 February 2018, at the age of 86. He was Professor Emeritus and Lawton Distinguished 2003) measured tropical precipitation 14 mm above Professor of Meteorology at Florida State University’s the 1981–2010 average. (FSU) Department of Earth, Ocean, and Atmospheric - Globally, 85 named tropical storms (TS) were ob Science. Krish, along with Bill Gray, is considered served during the 2017 Northern Hemisphere storm one of the fathers of modern tropical meteorology. season and the 2016/17 Southern Hemisphere storm season, as documented in the International Best For more than a half-century, Krish was a pioneer Tracks Archive for Climate Stewardship (IBTrACS; in tropical meteorology and numerical weather prediction, including high-resolution forecasting of Knapp et al. 2010). Overall, this number was slightly hurricane tracks, landfall, and intensities; short- and more than the 1981–2010 global average of 82 TSs. By long-range monsoon prediction; and interseasonal comparison, Diamond and Schreck (2017) reported - 93 named storms for 2016, although that number and interannual variability of the tropical atmo sphere. Krish was the recipient of the highest awards decreased to 85 after reanalysis. In terms of accu - given by both the American Meteorological Society mulated cyclone energy (ACE; Bell et al. 2000), the North Atlantic basin had an ACE of about 241% of and the World Meteorological Organization. its 1981–2010 median value and was the only basin Both Charlie and Krish will be greatly missed by all who knew and worked with them, as well as for that featured an above-normal season. For the North Atlantic, this was the fourth most active season since all that they accomplished to advance the science of at least 1950 and the seventh most active season in tropical meteorology. the historical record (since 1854). The western North | S101 AUGUST 2018 STATE OF THE CLIMATE IN 2017

122 November (SON; −0.7°C) and October–December M. L’Heureux, G. Bell, b. ENSO and the tropical Pacific— and M. S. Halpert (OND; −0.9 °C). The El Niño–Southern Oscillation (ENSO) is a ceanic 1) o conditions coupled ocean–atmosphere climate phenomenon Figures 4.2b,h further illustrate that 2017 was over the tropical Pacific Ocean, with opposite phases called El Niño and La Niña. For historical purposes, bookended by below-average SSTs in the east-central equatorial Pacific Ocean. Yet it was only in the last NOAA’s Climate Prediction Center (CPC) classifies and assesses the strength and duration of El Niño and season (SON 2017) that La Niña appeared, as the negative SST anomalies expanded and strengthened La Niña using the Oceanic Niño index (ONI, shown for the last half of 2016 and all of 2017 in Fig. 4.1). from the international dateline to coastal South America (Figs. 4.2g,h). In contrast, the rest of the year The ONI is the 3-month (seasonal) running average was more clearly ENSO-neutral, with near-average of SST anomalies in the Niño-3.4 region (5°N–5°S, SSTs evident across much of the central and eastern 170°–120°W) calculated as the departure from the Pacific Ocean (and above-average SSTs persisting in 1986–2015 base period. ENSO is classified as El Niño (La Niña) when the ONI is at or greater than +0.5°C the western Pacific Ocean; Figs. 4.2c–f). The primary exception to this pattern occurred near coastal South (at or less than −0.5°C) for at least five consecutive, overlapping seasons. America during DJF and March–May (MAM, Fig. The ONI shows 2017 was ENSO-neutral during 4.2d) 2017, when above-average SSTs emerged and became quite intense. This warming is indicative of most of the year, with the onset of La Niña conditions - a so-called “coastal El Niño” (Takahashi and Mar occurring during boreal autumn. ENSO-neutral con - tínez 2017; see Sidebar 7.2). During February–April ditions at the start of 2017 followed the abrupt end of 2017 the SST anomalies exceeded +2.5°C and were a short-lived, weak La Niña in 2016. That event lasted from July–September (JAS) until late December 2016 accompanied by damaging rainfall and f looding in Peru (L’Heureux 2017; Di Liberto 2017). (Bell et al. 2017a). - Consistent with the overall equatorial SST evolu Although officially ENSO-neutral, 2017 started off cooler relative to average, as ref lected by a Decem tion, subsurface temperatures east of the dateline - were generally near average most of the year (Fig. 4.3), ber–February (DJF) 2016/17 ONI value of −0.3°C. with a broad stretch of negative anomalies becoming Also, La Niña’s atmospheric impacts lingered into 2017. However, the equatorial Pacific continued to evident with the onset of La Niña (Fig. 4.3d). West of the dateline, the positive SST anomalies evident warm, and by late boreal spring and early summer the ONI increased to +0.4°C (just shy of El Niño for much of the year were accompanied by higher thresholds) for two consecutive, overlapping seasons subsurface temperatures and a deeper-than-average oceanic thermocline. However, these positive subsur - of April–June (AMJ) and May–July (MJJ). Thereafter, the ONI decreased through the remainder of the year, face anomalies weakened as the year went on, and the thermocline began to shoal in the eastern Pacific in exceeding thresholds for La Niña during September– association with a developing La Niña. subtropics and circulation 2) a tmospheric ropics : t Consistent with the average to below-average SSTs in the east-central equatorial Pacific, atmospheric anomalies during both DJF 2016/17 and SON 2017 were La Niña-like (Figs. 4.4, 4.5). Tropical convec - tion (as measured by outgoing longwave radiation) was enhanced over Indonesia and suppressed over the central Pacific Ocean during these two seasons (Figs. 4.4a,d; 4.5a,d), with some evidence for weak La Niña impacts lingering into MAM 2017 (Figs. . 4.1. Time series of the ONI (°C) during the last half F ig 4.4b, 4.5b). Correspondingly, the low-level (850-hPa) of 2016 and all of 2017. Overlapping, 3-month seasons wind anomalies over the western and central tropical 0.5°C are − are labeled on the x-axis. Values less than Pacific were easterly in both periods, which indicated shaded blue, indicating La Niña conditions. ONI values a strengthening of the trade winds (Figs. 4.4a,b,d). are derived from the ERSSTv5 dataset and are based The associated upper-level (200-hPa) winds over on departures from the 1986–2015 monthly means. (Huang et al. 2017). the central tropical Pacific in both hemispheres indi - | S102 AUGUST 2018

123 F . 4.2. Seasonal SST (left) and anomaly (right) for (a),(b) DJF 2016/17; (c),(d) MAM 2017; (e),(f) JJA 2017; and ig (g),(h) SON 2017. Contour interval for seasonal SST is 1°C. For anomalous SST, contour interval is 0.5°C for anomalies between ±1°C, and 1°C for anomalies > ±1°C. Anomalies are departures from the 1981–2010 seasonal adjusted OI climatology (Reynolds et al. 2002). F ig . 4.3. Equatorial depth–longitude section of Pacific Ocean temperature anomalies (°C) averaged between 5°N and 5°S during (a) DJF 2016/17, (b) MAM 2017, (c) JJA 2017, and (d) SON 2017. The 20°C isotherm (thick solid line) approximates the center of the oceanic thermocline. The data are derived from an analysis system that assimilates oceanic observations into an oceanic general circulation model (Behringer et al. 1998). Anomalies are departures from the 1981–2010 monthly means. | S103 AUGUST 2018 STATE OF THE CLIMATE IN 2017

124 contiguous United States. This teleconnection pat - tern, with three centers of action over the Pacific– North American region, is indicative of La Niña–like forcing from the tropical Pacific. c. Tropical intraseasonal activity— S. Baxter, C. Schreck, and G. D. Bell In the atmosphere, tropical intraseasonal variabil - ity was prominent throughout the year, alternating between constructive and destructive interference with the background low-frequency state. Two aspects of this intraseasonal variability are the Madden–Ju - lian Oscillation (MJO; Madden and Julian 1971, 1972, 1994; Zhang 2005), and convectively coupled equa - torial waves which include equatorial Rossby waves and atmospheric Kelvin waves (Wheeler and Kiladis 1999; Kiladis et al. 2009). There were three distinct periods of MJO activity during 2017 spanning a total of eight months (Fig. 4.6). Between the first two ac - tive MJO periods, intraseasonal variability ref lected F ig . 4.4. Anomalous 850-hPa wind vectors and speed 1 − ) and anomalous OLR (contour interval is 2 m s − 2 (shaded, W m ) during (a) DJF 2016/17, (b) MAM 2017, (c) JJA 2017, and (d) SON 2017. Reference wind vector is below right of color bar. Anomalies are departures from the 1981–2010 monthly means. - cated enhanced mid-Pacific troughs f lanking the re gion of suppressed convection near the dateline (Figs. 4.5a,d). The resulting anomalous cross-equatorial f low near the dateline, f lowing from the Southern Hemisphere tropics into the Northern Hemisphere, was especially prominent during DJF (Fig. 4.5a). However, the upper-level winds were anomalously westerly over the central tropical Pacific only during - SON 2017, which indicates that the broader, overturn ing Pacific Walker circulation was enhanced only late in the year (Fig. 4.5d). In the Northern Hemisphere, a La Niña-like anomalous 500-hPa height anomaly pattern was evident both early (DJF) and late (SON) in the year (see Online Figs. S4.1 and S4.2). In particular, both ig . 4.5. Anomalous 200-hPa wind vectors and speed F periods featured an anomalous ridge over the North − 1 ) and anomalous OLR (contour interval is 4 m s Pacific Ocean in association with a retracted East − 2 (shaded, W m ) during (a) DJF 2016/17, (b) MAM 2017, Asian jet stream. Downstream of the ridge, anoma - (c) JJA 2017, and (d) SON 2017. Reference wind vector lous troughing occurred over western Canada while is below right of color bar. Anomalies are departures an anomalous ridge was apparent over the southern from the 1981–2010 period monthly means. | S104 AUGUST 2018

125 . 4.7. Time–longitude section for 2017 of anomalous ig F − 2 - ) averaged for 10°N–10°S (Lee 2014). Neg OLR (W m - ative anomalies indicate enhanced convection and posi tive anomalies indicate suppressed convection. Con - ig . 4.6. Time–longitude section for 2017 of 5-day F tours identify anomalies filtered for the MJO (black), running anomalous 200-hPa velocity potential 2 6 1 − atmospheric Kelvin waves (red), and Rossby waves (× 10 s m ) averaged for 5°N–5°S, from NCEP–NCAR (blue). Red labels highlight the main MJO episodes. reanlysis (Kalnay et al. 1996). For each day, the period 2 − Contours are drawn at ±12 W m , with the enhanced mean is removed prior to plotting. Green (brown) (suppressed) convective phase of these phenomena shading highlights likely areas of anomalous divergence indicated by solid (dashed) contours. Anomalies are and rising motion (convergence and sinking motion). departures from the 1981–2010 base period. Red lines and labels highlight the periods when the MJO was most active; solid (dashed) lines indicate the 1993; Kousky and Kayano 1994; Kayano and Kousky - MJO enhanced (suppressed) phase. Anomalies are de partures from the 1981–2010 base period daily means. 1999; Cassou 2008; Lin et al. 2009; Riddle et al. 2013; Schreck et al. 2013; Baxter et al. 2014). The MJO is atmospheric Kelvin waves (Fig. 4.7) and tropical often episodic, with periods of moderate or strong activity sometimes followed by little or no activity. cyclone activity. Between the latter two active MJO periods, intraseasonal variability largely ref lected the The MJO tends to be most active during ENSO- evolution to La Niña. neutral and weak ENSO periods. The MJO is often The MJO is a leading intraseasonal climate mode absent during strong El Niño events (Hendon et al. 1999; Zhang and Gottschalck 2002; Zhang 2005), of tropical convective variability. Its convective though the strong El Niño winter of 2015/16 exhibited anomalies often have a similar spatial scale to ENSO unusually strong MJO activity (Baxter et al. 2017). but differ in that they exhibit a distinct eastward - propagation and generally traverse the globe in 30–60 Common metrics for identifying the MJO in days. The MJO affects weather patterns around the clude time–longitude plots of anomalous 200-hPa - globe (Zhang 2013), including monsoons (Krish velocity potential (Fig. 4.6) and anomalous outgoing namurti and Subrahmanyam 1982; Lau and Waliser longwave radiation (OLR; Fig. 4.7), as well as the Wheeler–Hendon (2004) Real-time Multivariate MJO 2012), tropical cyclones (Mo 2000; Frank and Roundy (RMM) index (Fig. 4.8). In the time–longitude plots, 2006; Camargo et al. 2009; Schreck et al. 2012), and the MJO exhibits eastward propagation from upper extratropical circulations (Knutson and Weickmann - left to lower right. In the RMM, the MJO propaga 1987; Kiladis and Weickmann 1992; Mo and Kousky | S105 AUGUST 2018 STATE OF THE CLIMATE IN 2017

126 by westward-moving Rossby - waves (Fig. 4.7, blue con - tours). This period was fol lowed during April and May by a series of fast propagating atmospheric Kelvin waves (Fig. 4.7, red contours). - Impacts from MJO #1 in cluded distinct periods with westerly and easterly zonal wind anomalies over the western Pacific, including a significant westerly wind burst (labeled WWB) and a significant easterly trade wind surge (labeled TWS) event (Fig. 4.9a). These con - ditions produced alternating downwelling and upwell - ing equatorial oceanic Kel - vin waves, the last of which was a downwelling wave whose anomalous warming reached the west coast of South America in early June (Fig. 4.9b). In the extratropics, MJO #1 may have had impacts F ig . 4.8. Wheeler–Hendon (2004) Real-time Multivariate MJO (RMM) index over the North Pacific and during 2017 for (a) Jan–Mar, (b) Apr–Jun, (c) Jul–Sep, and (d) Oct–Dec. Each North America. The 500- point represents the MJO amplitude and location on a given day, and the hPa height anomalies (not connecting lines illustrate its propagation. Amplitude is indicated by distance from the origin, with points inside the circle representing weak or no MJO. - shown) featured an extra The eight phases around the origin identify the region experiencing enhanced - tropical wave train that ter convection, and counter-clockwise movement reflects eastward propagation. minated in an anomalous ridge over the contiguous United States, a pattern tion and intensity are seen as large, counterclockwise circles around the origin. When considered together, associated with the MJO as it traverses the Maritime these diagnostics point to three main MJO episodes Continent (Schreck et al. 2013; Baxter et al. 2014). In - during 2017. MJO #1 was a strong episode that began the second half of February and early March, how ever, there was little evidence of an MJO extratropical in January and continued into March. MJO #2 was a weak but long-lived signal that began in June and response over North America. lasted into early September. MJO #3 featured strong MJO #2 occurred during June–August, and its MJO activity that began in October and continued wave-1 signal circumnavigated the globe about 1.5 through the end of the year. times (Fig. 4.6). The MJO’s periodicity during this episode was about 60 days, which is on the slower MJO #1 featured a zonal wave-1 pattern of strong side of the MJO phase speed envelope. This episode convective anomalies with a periodicity of 30–35 days (Figs. 4.6, 4.8a), which is on the fast end of phase terminated when the anomalous convective pattern speeds for MJO events. The plot of anomalous velocity became more dominated by tropical cyclone activity - and two high-amplitude atmospheric Kelvin waves potential (Fig. 4.6) shows that the MJO circumnavi (Fig. 4.7). The RMM index indicates that MJO #2 was gated the globe almost twice during this period, and quite weak (Fig. 4.8c). Consequently, its impacts were the RMM index (Fig. 4.8a) indicates that the event was strongest in February. The episode ended in March also weak and limited, with no associated equatorial oceanic Kelvin wave activity and only weak linkages when the convective anomalies became dominated | S106 AUGUST 2018

127 played a role in modulating the relative strength and position of anomalous upper-level ridging over the North Pacific and, in turn, supporting cold air outbreaks over east-central North America during late December 2017 and early January 2018 (L’Heureux 2018). - Within the equato rial Pacific Ocean itself, - two key aspects of intra seasonal variability dur - ing 2017 were likely not related to the MJO. The first was the rapid development ig . 4.9. (a) Time–longitude section for 2017 of anomalous 850-hPa zonal wind F during August, and the − 1 (m s ) averaged over 10°N–10°S from CFSR (Saha et al. 2014). Black contours subsequent persistence, of identify anomalies filtered for the MJO. Red labels highlight the main MJO negative upper-ocean heat episodes. Significant westerly wind bursts and trade wind surges (TWS) that content anomalies across - resulted in notable downwelling and upwelling oceanic Kelvin waves are la - the eastern half of the Pa beled. (b) Time–longitude section of the anomalous equatorial Pacific Ocean cific basin. This evolution heat content for 2017, calculated as the mean GODAS temperature anomaly at 0–300-m depth (Behringer et al. 1998). Yellow/red (blue) shading indicates ref lected the developing La above- (below-) average heat content. The relative warming (dashed lines) and Niña. The second was ad - cooling (dotted lines) due to downwelling and upwelling equatorial oceanic ditional strengthening of Kelvin waves are indicated. Anomalies are departures from the 1981–2010 base those negative anomalies period pentad means. from mid-September to to Northern Hemisphere TC activity. This MJO may - mid-November in response to an upwelling equato have played a role in enhancing the eastern North rial oceanic Kelvin wave. This upwelling wave was associated with a trade wind surge event in September Pacific TC activity during July and in suppressing that over the far western Pacific (Fig. 4.9a). basin’s TC activity during August (see Section 4f3). MJO #3 was a period of strong MJO activity that d. Intertropical convergence zones began during October and persisted through the end 1) A. B. Mullan — p acific of the year, making nearly two passes around the globe. The average periodicity was about 45 days, Tropical Pacific rainfall patterns are dominated by but the propagation slowed with time: the first MJO two convergence zones, the intertropical convergence circumnavigation of the globe took about 30 days, zone (ITCZ) and the South Pacific convergence zone while the second took almost twice that (Fig. 4.6). (SPCZ), both of which are strongly inf luenced by This episode had several notable impacts. First, it ENSO and longer time-scale variations (Schneider et al. 2014; Vincent 1994; Folland et al. 2002). Figure was in phases 5–7 during most of October (Fig. 4.8d), 4.10 summarizes the convergence zone behavior for which are generally less favorable phases for Atlantic 2017 using rainfall patterns rather than cloudiness, hurricane activity. Therefore, it may have played a role and it allows comparison of the 2017 seasonal varia - in the October activity being closer to climatology after a record-breaking September, despite La Niña tion against the longer-term 1998–2016 climatology. conditions which typically favor late-season Atlantic Rainfall transects over 20°N to 30°S are presented for each quarter of the year, averaged across successive hurricane activity (Klotzbach et al. 2017). Second, the 30-degree longitude bands, starting in the western MJO’s strong WWB west of the dateline in December (Fig. 4.9a) triggered a downwelling equatorial oceanic Pacific at 150°E–180°. The rainfall is estimated from satellite microwave and infrared data using NOAA’s Kelvin wave that led to warming of the upper ocean by late December (Fig. 4.9b). Third, this MJO likely CPC morphing technique (CMORPH; Joyce et al. 2004) and is available at 0.25° resolution. | S107 AUGUST 2018 STATE OF THE CLIMATE IN 2017

128 Fig. 4.10a, is what is commonly referred to as the double ITCZ - in the far eastern tropical Pa cific. The southern branch of - the ITCZ is seen during Feb ruary through April but only in ENSO-neutral or La Niña years. The southern ITCZ was prominent during these months in 2017 (Fig. 4.11) and was one of the strongest since the beginning of the Tropical Rainfall Measur - ing Mission (TRMM) satel - lite era [equivalent to 2009, not shown; TRMM has since been replaced by the Global Precipitation Measurement (GPM) satellite in 2014 (Hou et al. 2014)]. 1 − ) from CMORPH analysis during 2017 for (a) . 4.10. Rainfall rate (mm day ig F - The tropical Pacific re Jan–Mar, (b) Apr–Jun, (c) Jul–Sep, and (d) Oct–Dec. The separate panels for mained in a neutral ENSO 3-month periods show the 2017 rainfall cross-section between 20°N and 30°S state through April–June and (solid line) and the 1998–2016 climatology (dotted line), separately for four July–September, often with 30° sectors from 150°E–180° to 120°–90°W. weak but conf licting ENSO signals in the atmosphere and ocean. For example, The ITCZ lies between 5° and 10°N and is most active during August–December, when it lies at its conditions at the end of June indicated a negative Southern Oscillation index (i.e., leaning towards El northernmost position and displays more of an east- northeasterly tilt. The SPCZ extends diagonally from - Niño), but large-scale rainfall and convection anoma around the Solomon Islands (10°S, 160°E) to near lies showed more intense-than-normal convection and rainfall over large parts of Indonesia, indicative 30°S, 140°W, and is most active during November– April. As described in Section 4b, 2016 ended with of a La Niña pattern. In general, the ITCZ and SPCZ the demise of a weak La Niña and neutral conditions - were near their climatological positions and intensi ties during this 6-month period, although the SPCZ prevailed for most of 2017, with the eventual emer - was somewhat more active than usual at about 10°S gence of another weak La Niña in October. In the first quarter of 2017 (Fig. 4.10a), both the ITCZ and during April–June (Fig. 4.10b). Associated with this enhanced convection, SSTs were 0.5°–1.0°C above the SPCZ were poleward of their normal positions, - normal throughout the central and western subtropi likely due to persistence of the weak La Niña condi - cal South Pacific (see Fig. 4.2d). tions of the previous year (Mullan 2017). Thus, island In September, the tropical Pacific edged closer to groups close to the equator (e.g., Kiribati and Tokelau) La Niña conditions, with anomalous cooling in the experienced continued dry conditions (www.niwa .co.nz/climate/icu). Conversely, many islands within the Federated States of Micronesia were far enough north of the equator to experience wetter conditions due to the displaced ITCZ (www.weather.gov/peac /update). Please refer to Section 7h2 for more details. Figure 4.10a also shows that, except west of the dateline, the ITCZ was much weaker than normal across the North Pacific. Figure 4.11 gives an alterna - tive view of the much drier-than-normal conditions F ig . 4.11. CMORPH rainfall anomalies over the tropi - that prevailed across most of the tropical Pacific cal Pacific for Jan–Mar 2017, as a percentage of the during January–March. The other unusual feature 1998–2016 average. The white areas indicate anomalies of the convergence zones in this quarter, apparent in within 25% of normal. | S108 AUGUST 2018

129 1996). Equatorial atmospheric Kelvin waves can modulate the ITCZ intraseasonal variability (Guo et al. 2014). ENSO and the southern annular mode (SAM) also inf luence the ITCZ on the interannual time scale (Münnich and Neelin 2005). The SAM is typically positive during La Niña events, and it was generally so in 2017 (from April onwards) when the equatorial Pacific started to anomalously cool from ENSO neutral (Fig. 4.13a) to a La Niña state (Fig. 4.13b). The SAM is the primary pattern of climate − 1 F . 4.12. CMORPH rainfall rate (mm day ) for Oct– ig variability in the Southern Hemisphere (Marshall Dec, for each year 1998 to 2017, averaged over the 2003; Thompson et al. 2011), inf luencing latitudinal longitude sector 180°–150°W. The cross-sections are rainfall distribution and temperatures from the color-coded according to NOAA’s ONI, except for 2017 subtropics to Antarctica. The station-based index of (a La Niña period) shown in black. the SAM is based on the zonal pressure difference eastern and central tropical Pacific, an increase in between the latitudes of 40°S and 65°S. As such, the the Southern Oscillation index, and more enhanced SAM index measures a “see-saw” of atmospheric mass easterly trade winds in the central and western between the middle and high latitudes of the Southern Hemisphere (Marshall 2003). A positive SAM value equatorial Pacific. During the last quarter of 2017, La Niña conditions became more consistent across can be indicative of a number of things; for instance, a positive SAM coupled with La Niña conditions may both atmospheric and oceanic features. The ITCZ and SPCZ remained at their climatological locations lead to increased extratropical cyclone transition west of the dateline; on average, however, both were of tropical cyclones across or toward New Zealand displaced poleward of their normal positions to the (Diamond and Renwick 2015). east of the dateline (Fig. 4.10d). In the central north Pacific (180°–120°W), rainfall was well below normal from the equator to the lati - tude where the ITCZ rainfall peaked (about 8°–9°N as depicted in Fig. 4.10d). In the 180°–150°W sector, - the latitude of peak rainfall matched well with previ ous La Niña events, but the intensity was the lowest since the beginning of the TRMM satellite record. Figure 4.12 shows the south–north rainfall transect of Fig. 4.10d, except that every year from 1998 is shown, color-coded according to NOAA’s Oceanic Niño index. October–December 2017, classified as a La Niña quarter, is highlighted separately in black. Although rainfall north of the equator was unusually weak for a La Niña, conditions along the equator and southwards followed the expected La Niña behavior. Islands near the equator (e.g., Nauru and all the Kiri - bati groups) thus continued the dry conditions they had experienced since the weak La Niña at the end of 2016. In the Southern Hemisphere during October– December, the SPCZ matched well with past La Niña periods with respect to both intensity and latitudinal location (Fig. 4.12). A. B. Pezza and C. A. S. Coelho — 2) a tlantic The Atlantic ITCZ is a well-organized convective band that oscillates approximately between 5°–12°N F ig . 4.13. Spatial distribution global SST anomalies during July–November and 5°N–5°S during January– (°C; Reynolds et al. 2002) for (a) Jan–Apr and (b) Sep–Dec 2017. May (Waliser and Gautier 1993; Nobre and Shukla | S109 AUGUST 2018 STATE OF THE CLIMATE IN 2017

130 While in principle this reversal toward the cool ENSO phase would tend to favor the Atlantic ITCZ moving south, in reality the change occurred too late in the ITCZ’s southern migration season in order to - have a positive effect on the rainy season in north eastern Brazil, even though overall the ITCZ was very active over the ocean. For the most part an enhanced South Atlantic anticyclone, increased trade winds, and relatively warmer waters north of the equator prevailed. This was accompanied by a mostly nega - tive South American sector (SA) index, although not so pronounced as in some previous years (Fig. 4.14a). The SA index, as defined in Fig. 4.14, is given by the . 4.15. Observed 2017 precipitation anomalies (mm ig F − 1 ) for tropical and subtropical South America dur - day ing (a) Jan–May and (b) Jun–Dec. Anomalies calculated based on the 1998–2016 climatology derived from CMORPH (Joyce et al. 2004). SST south of the equator minus the SST north of the equator over key areas of ITCZ inf luence. The ITCZ tends to shift toward the warmer side of this gradient. Indeed, it was generally north of its climatological position for most of 2017, especially in May when it is typically at its southernmost location (Fig. 4.14b). As discussed above, the overall convection was active over the ocean, and although northeastern Brazil remained dry, the eastern Amazon region (Para state) had above-normal precipitation during the wet season (Fig. 4.15a). The ITCZ remained active for the remainder of the year, mostly over the ocean, as La Niña developed (Fig. 4.15b). i g F . 4.14. (a) Monthly OISST (Smith et al. 2008) - anomaly time series averaged over the South Ameri B. Wang e. Global monsoon summary— can sector (SA region, 5°S –5°N, 10°–50°W) minus the The global monsoon (GM) is the dominant mode SST anomaly time series averaged over the North - of annual variations of tropical–subtropical precipita Atlantic sector (NA region, 5°–25°N, 20°–50°W) for tion and circulation (Wang and Ding 2008) and thus 2013–17, forming the Atlantic index. Positive phase of a defining feature of seasonality and a major mode of the index indicates favorable conditions for enhanced - Atlantic ITCZ activity. (b) Atlantic ITCZ position in - variability of Earth’s climate system. Figure 4.16 sum ferred from OLR (Liebmann and Smith 1996) during marizes the monsoon rainfall anomalies for both the May 2017. Colored thin lines indicate the approximate SH summer monsoon (SHSM) from November 2016 position for the six pentads of the month. Black thick to April 2017 and the NH summer monsoon (NHSM) line indicates the Atlantic ITCZ climatological position from May to October 2017. for May. The SST anomalies (°C) for May 2017 based Global land monsoon precipitation is strongly on the 1982–2016 climatology are shaded. The two inf luenced by the status of ENSO, especially the land boxes indicate the areas used for the calculation of the Atlantic index in (a). areas of Asia, Australia, northern Africa, and Central | S110 AUGUST 2018

131 − 1 ) averaged for (a) northern winter season: Nov 2016–Apr 2017 and . 4.16. Precipitation anomalies (mm day ig F (b) northern summer: May–Oct 2017. The red lines outline the global monsoon precipitation domain defined by (a) annual range (local summer minus winter) precipitation exceeding 300 mm and (b) summer mean precipitation >55% of the total annual precipitation amount (Wang and Ding 2008). Here the local summer denotes May–Sep for the NH and Nov–Mar for the SH. Precipitation indices for each regional monsoon are defined by the areal mean precipitation in the corresponding rectangular regions (dashed blue boxes), which are highly correlated with the precipitation averaged over the corresponding real regional monsoon domains (see Table 4.1). Rainfall data were taken from the Global Precipitation Climatology Project (GPCP; Huffman et al. 2009). Note that the threshold of 300 mm excludes a small latitudinal band of the monsoon in the Sahel. America (Wang et al. 2012). As documented in Fig. 4.1 for this year, the equatorial Pacific SSTs evolved from a weak La Niña from NDJ to a neutral state, then toward another weak La Niña during SON. Figure 4.16 indicates that the - monsoon precipitation anom alies are generally in normal states with a few individual regions slightly positive, con - sistent with the near-neutral ENSO SST anomalies. Figure 4.17 shows the time series of the monsoon precipitation and low-level circulation indices for each regional monsoon. Note that the precipitation indices represent the total amount of precipitation over both land and ocean. The definitions of circulation in- dices for each monsoon region are shown in Table 4.1. The precipitation and circulation indices together represent the strength of each regional monsoon system. Fig. 4.17. Normalized summer mean precipitation (green) and circulation (red) During the SH summer indices in each of eight regional monsoon regions (see Table 4.1). Indices are (November 2016–April 2017), normalized by their corresponding std. dev. Numbers shown in the corner of global precipitation exhibited - each panel denote the correlation coefficient between seasonal mean pre a pattern consistent with the cipitation and circulation indices. Dashed lines indicate std. dev. of ±0.5. Here decay of a weak La Niña (aver - the summer denotes May–Oct for the NH and Nov–Apr for the SH. [Source: aged ONI = −0.2): suppressed GPCP for precipitation; Upper air indices as described in Yim et al ( 2014).] | S111 AUGUST 2018 STATE OF THE CLIMATE IN 2017

132 AB 4.1. (Modified from Yim et al. 2013). Definition of the regional summer monsoon circulation le t indices and their correlation coefficients (CCs) with the corresponding regional summer monsoon precipitation indices for the period 1979–2015. All circulation indices are defined by meridional shear of zonal winds at 850 hPa which measures the intensity (relative vorticity) of the monsoon troughs at 850 hPa except the northern African (NAF) and East Asian (EA). The NAF monsoon circulation index is defined by the westerly monsoon strength: U850 (0°–15°N, 60°–10°W) and the EASM circulation index is defined by the meridional wind strength: V850 (20°–40°N, 120°–140°E) which reflects the east–west thermal contrast between the Asian continent and western North Pacific. The precipitation indices are defined by the areal mean precipitation over the blue box regions shown in Fig. 4.18. The correlation coefficients were computed using monthly time series (148 summer months) [Jun–Sep (JJAS) in NH (1979–2015) and Dec–Mar (DJFM) in SH (1979/80–2015/16]. The bolded numbers represent significance at 99% confidence level. Region Circulation Index Definition CC U850 (5°–15°N, 40°–80°E) minus Indian (ISM) 0.71 U850 (25°–35°N, 70°–90°E) U850 (5°–15°N, 100°–130°E) minus Western North Pacific (WNPSM) 0.78 U850 (20°–35°N, 110°–140°E) East Asian (EASM) V850 (20°–40°N, 120°–140°E) 0.73 U850 (5°–15°N, 130°–100°W) minus North American (NASM) 0.84 U850 (20°–30°N, 110°–80°W) U850 (0–15°N, 60°–10°W) Northern African (NAFSM) 0.72 U850 (5°–20°S, 70°–40°W) minus South American (SASM) 0.77 U850 (20°–35°S, 70°–40°W) U850 (0°–15°S, 10°–40°E) minus Southern African (SAFSM) 0.55 U850 (10°–25°S, 40°–70°E) U850 (0°–15°S, 90°–130°E) minus Australian (AUSSM) 0.89 U850 (20°–30°S, 100°–140°E) precipitation over the Pacific ITCZ and southern In f. Tropical Cyclones - dian Ocean convergence zone, and increased precipi — 1) o verview H. J. Diamond and C. J. Schreck - tation over the Maritime Continent and adjacent re - The IBTrACS dataset comprises historical TC gions (Fig. 4.16a). As a result, the Australian summer best-track data from numerous sources around the monsoon region received slightly more precipitation globe, including all of the WMO Regional Specialized than normal, and the strength of the corresponding Meteorological Centers (RSMC; Knapp et al. 2010). circulation was also above normal (Fig. 4.17h). The IBTrACS represents the most complete compilation of southern African summer monsoon precipitation global TC data. From these data, Schreck et al. (2014) and circulation were near normal (Fig. 4.17f), while compiled climatological values of TC activity for each the South American monsoon shows slightly below- basin for 1981–2010 using both the WMO RSMCs and the Joint Typhoon Warning Center (JTWC). These average intensity in both precipitation and circulation values are referenced in each subsection. (Fig. 4.17g). Overall, the SH summer monsoon during - The tallying of the global TC numbers is challeng November 2016 to April 2017 was normal. During the NH summer (May–October) of 2017, ing and involves more than simply adding up basin totals, because some storms cross TC basin bound - ENSO was neutral (average ONI = 0.0) and global aries, some TC basins overlap, and multiple agencies precipitation also tended to be near normal, as did overall NH summer monsoon precipitation (Fig. - are involved in tracking and categorizing TCs. Com 4.16b). On regional scales, the summer precipitation piling the activity using preliminary IBTrACS data over India, East Asia, and western North Pacific were over all seven TC basins (Fig. 4.18), the 2017 season all near normal (Figs. 4.17a–c), while precipitation was (2016/17 in the Southern Hemisphere) had 85 named −1 above normal over the North American monsoon storms (wind speeds ≥ 34 kt or 17 m s ). This number matches the post-analysis 2016 total (Diamond and region and slightly above normal over the northern Schreck 2017) and is slightly above the 1981–2010 African monsoon region (Figs. 4.17d,e). | S112 AUGUST 2018

133 The North Atlantic hurricane season - was above normal in both storm num bers and intensity (Section 4f2). In fact, it was the only basin globally that featured above-normal accumulated cyclone en - ergy (ACE). The central and eastern North Pacific hurricane season was well below normal for number of storms (Section 4f3). The western North Pacific had less than half of its normal annual ACE, and the Southern Hemisphere had one of its quietest TC seasons on record, particularly . 4.18. Global summary of TC tracks overlaid on the associated ig F with respect to ACE (Sections 4f6–8). OISST anomalies (°C) for the 2017 season relative to 1982–2010. Globally, only three storms during the year reached Saffir–Simpson category 5 −1 average of 82 (Schreck et al. 2014). The 2017 season strength (wind speeds ≥ 137 kt or 70.5 m s ), which is also featured 41 hurricanes/typhoons/cyclones (HTC; one less than in 2016 and five fewer than in 2015. The −1 three 2017 storms were Hurricanes Irma and Maria wind speeds ≥ 64 kt or 33 m s ), which is below the 1981–2010 average of 46 HTCs (Schreck et al. 2014). in the North Atlantic and Super Typhoon Noru in the western North Pacific. Sidebars 4.1 and 4.3 detail the Twenty storms reached major HTC status during the −1 records set and devastating local impacts of Irma and ), which 2017 season (wind speeds ≥ 96 kt or 49 m s Maria, respectively. is near the long-term average of 21 and is six fewer than the post-analysis 26 HTCs recorded in 2016 Several other Saffir–Simpson category 3 and 4 (Diamond and Schreck 2017). intensity level systems during 2017 also had major impacts, including: (1) Hurricane Harvey in the In Sections 4f2–4f8, the 2017 seasonal activity is North Atlantic, (2) Typhoons Tembin and Hato in described and compared to the historical record for the western North Pacific, and (3) Tropical Cyclone each of the seven WMO-defined hurricane basins. For simplicity, all counts are broken down by the Debbie in the Australian basin. Also noteworthy United States’ Saffir–Simpson scale. The overall was the development of Tropical Cyclone Donna in picture of global TCs during 2017 is shown in Fig. the southwest Pacific basin in early May 2017, a date which is outside of the formal TC season for that 4.18 and actual counts by category are documented in Table 4.2. basin. Donna became the most intense TC recorded in that basin during the month of May. t AB le 4.2. Global tropical cyclone counts by basin in 2017. ACE Major HTCs TSs SS Cat 5 TDs Basin 4 2 ( 10 HTCs kt ) × 18 17 10 6 2 225 North Atlantic 20 Eastern North Pacific 9 4 0 97 18 Western North Pacific 26 12 4 1 150 35 North Indian 4 4 2 1 0 15 South Indian 5 2 1 0 28 5 Australian Region 8 3 3 0 30 8 Southwest Pacific 7 7 3 1 0 35 3 Totals 85 41 20 98 580 | S113 AUGUST 2018 STATE OF THE CLIMATE IN 2017

134 2) a tlantic basin — G. D. Bell, E. S. Blake, C. W. Landsea, - since at least 1950 and the seventh most active sea S. B. Goldenberg, and R. J. Pasch son in the historical record (since 1854). However, it (i) 2017 Seasonal activity should be noted that reliable basin-wide records for The 2017 Atlantic hurricane season produced 17 exact season-to-season comparisons with ACE began named storms, of which 10 became hurricanes and in the mid-1970s with the advent of the geostationary 6 of those became major hurricanes (Fig. 4.19a). The satellite era (Landsea et al. 2006). HURDAT2 30-year (1981–2010) seasonal averages The occurrence of above-normal and extremely are 11.8 tropical (named) storms, 6.4 hurricanes, and active seasons shows a strong multidecadal signal. 2.7 major hurricanes (Landsea and Franklin 2013). The 2017 season is the 15th above-normal season The 2017 seasonal ACE value (Bell et al. 2000) - and the 9th extremely active season since the cur was about 241% of the 1981–2010 median (92.4 × rent high-activity era for Atlantic hurricanes began 2 4 10 ; Fig. 4.19b). This value is well above NOAA’s in 1995. The previous Atlantic high-activity era kt (1950–70) also featured numerous above-normal thresholds for an above-normal season (120%) and and extremely active seasons. In stark contrast, the an extremely active season (165%), www.cpc.ncep .noaa.gov/products/outlooks/background intervening low-activity era of 1971–94 featured only _information.shtml. two above-normal seasons, and none were extremely active (Goldenberg et al. 2001). This ACE value makes 2017 the most active season since 2005, and the first extremely active season since 2010. It also makes 2017 the fourth most activeseason (ii) Storm formation regions, tracks, and landfalls A main delineator between above-normal and below-normal Atlantic hurricane seasons is the number of hurricanes and major hurricanes that de - velop from storms that are named while in the main development region (MDR, green boxed region in Fig. 4.21a) spanning the tropical Atlantic Ocean and Caribbean Sea between 9.5° and 21.5°N (Goldenberg and Shapiro 1996; Goldenberg et al. 2001; Bell and Chelliah 2006). When activity is high in the MDR, overall seasonal TC activity and ACE are also high. The vast majority of storms which form within the MDR do so during the peak months (August–Octo - ber, ASO) of the season. This peaked climatology is why seasonal hurricane predictions are essentially based on predictions for ASO of the atmospheric and oceanic conditions within the MDR (Goldenberg and Shapiro 1996; Klotzbach et al. 2017). During 2017, seven of the ten Atlantic hurricanes and five of the six major hurricanes first became - named storms during ASO in the MDR. For the sea . 4.19. Seasonal Atlantic hurricane activity during ig F 1950–2017 based on HURDAT2 (Landsea and Franklin son as a whole, MDR-originating storms produced an 2013). (a) Number of named storms (green), hurricanes ACE of 212% of the 1981–2010 median and accounted (red), and major hurricanes (blue), with 1981–2010 for 86% of the total season’s ACE. The strongest and seasonal means shown by solid colored lines. (b) ACE longest-lived MDR storm of the season was Major index expressed as percent of the 1981–2010 median Hurricane Irma, which developed in late August value. ACE is calculated by summing the squares of and by itself produced an ACE value of 77.5% of the the 6-hourly maximum sustained surface wind speed 1981–2010 median. Only one storm in the satellite (knots) for all periods while the storm is at least tropical storm strength. Red, yellow, and blue shad - record since 1966 (Major Hurricane Ivan in 2004) ings correspond to NOAA’s classifications for above-, produced a larger ACE. near-, and below-normal seasons, respectively (www Extremely active seasons have a higher frequency .cpc.ncep.noaa.gov/products/outlooks/background of landfalling tropical storms, hurricanes, and major _information.shtml). The thick red horizontal line hurricanes. During 2017, there were 13 separate storm at 165% ACE value denotes the threshold for an landfalls for the basin as a whole. This count ref lects extremely active season. Vertical brown lines separate ten distinct named storms, of which six formed in high- and low-activity eras. | S114 AUGUST 2018

135 the MDR. These six MDR storms include all three The entire region around the Caribbean Sea also typically sees an increased number of hurricane landfalling major hurricanes and two of the three (excluding Nate which formed in the extratropics) landfalls during extremely active seasons. During landfalling non-major hurricanes. 2017, eight named storms struck the region. These Six named storms struck the United States during included two catastrophic major hurricanes (Irma 2017, including three catastrophic major hurricanes and Maria), two non-major hurricanes (Franklin and (Harvey in Texas, Irma in Florida, and Maria in Katia in eastern Mexico), and four tropical storms - (Bret in Trinidad and Venezuela; Harvey in Barba Puerto Rico and the U.S. Virgin Islands), one non- major hurricane (Nate in Louisiana/ Mississippi), dos and St. Vincent; Nate in Central America; and and two tropical storms (Cindy in Texas and Emily in Philippe in Cuba). Florida). Harvey was the first continental U.S. land - (iii) Atlantic sea surface temperatures falling major hurricane since Wilma struck Florida SSTs were above average during ASO 2017 across in October 2005. the MDR, the Gulf of Mexico, and much of the From a historical perspective, 86% (12 of 14 sea - sons) of extremely active seasons during 1950–2017 extratropical North Atlantic (Fig. 4.21a). The area- averaged SST anomaly within the MDR was +0.54°C featured at least two continental U.S. landfalling (Fig. 4.21b). The area-averaged SST anomaly within hurricanes (Fig. 4.20a). This rate far exceeds the 50% the Caribbean Sea, a subregion of the MDR, was rate (7 of 14 seasons) for above-normal seasons that were not extremely active and is almost triple the rate (30%, 6 of 20 seasons) for near-normal seasons. Only 5% (1 of 20 seasons) of the below-normal seasons since - 1950 produced multiple continental U.S. landfall ing hurricanes. Similarly, 71% (10 of 14 seasons) of extremely active seasons since 1950 featured at least one major hurricane landfall in the continental U.S (Fig. 4.20b). This is more than double the 31% rate (17 of 54 seasons) of landfalling major hurricanes for all other seasons combined. Interestingly, about 20% of below-normal seasons have had a continental U.S. landfalling major hurricane. F ig . 4.20. Continental U.S. landfalling hurricane fre - quencies during 1950–2017 for each of NOAA’s season types. (a) Percent of specified season type with at least two U.S. hurricane landfalls, and (b) percent of . 4.21. (a) ASO 2017 SST anomalies (°C). (b),(c) Time ig F specified season type with one or more U.S. major series of ASO area-averaged SST anomalies (°C) in (b) hurricane landfall. Above-normal seasons include the MDR [green box in (a)] and (c) the Caribbean Sea those labeled “Extremely Active” and “Above Normal [red box in (a) spanning 60°–87.5°W and 10°–21.5°N]. Not Extremely Active.” Season classifications are Red lines in (b) and (c) show a 5-pt. running mean of shown in www.cpc.ncep.noaa.gov/products/outlooks each time series. Data source is ERSSTv4 (Huang et al. /background_information.shtml. Landfall data is based 2015). Anomalies are departures from the 1981–2010 on HURDAT2 (Landsea and Franklin 2013). monthly means. | S115 AUGUST 2018 STATE OF THE CLIMATE IN 2017

136 +0.60°C. This departure for the Caribbean Sea was low activity eras of 1900–20 and 1971–94 were associ - the second highest since 1950 and followed the record ated with the negative (i.e., cool) phase of the AMO. warmth of ASO 2016 (Fig. 4.21c). Another complementary measure of the AMO is the standardized 5-year running mean of the dif - Historically, when assessing links between Atlan - tic SSTs and hurricane season strength, it is important ference between the area-averaged SST departure in to consider their common relationships to larger- the MDR and that of the global tropics (Fig. 4.22b, scale climate patterns. Two key climate patterns are based on Vecchi and Soden 2007). The warm AMO the Atlantic multidecadal oscillation (AMO; Enfield during ASO 2017 featured an anomalously warm and Mestas-Nuñez 1999; Goldenberg et al. 2001; Bell MDR compared to the remainder of the global tropics and Chelliah 2006; Bell et al. 2011, 2012) and ENSO (0.36°C higher), a relationship seen throughout the historical record for active seasons. These observa - (Gray 1984; Tang and Neelin 2004; Bell and Chelliah tions, combined with the seasonal ACE time series 2006). These SST-based phenomena strongly control large-scale atmospheric conditions (such as vertical (Fig. 4.19b), suggest that continuation during 2017 of the current Atlantic high-activity era was associated - wind shear, trade winds, moisture, atmospheric sta bility, etc.) across the MDR, thereby inf luencing the with the ongoing warm phase of the AMO. strength of the hurricane season. Another ocean–atmosphere related factor for the 2017 Atlantic hurricane season was the development The AMO predisposes the ocean–atmosphere of La Niña in October (see Section 4b). La Niña is system to be either more or less conducive to Atlantic hurricane activity for periods of 25–40 years at a time. - conducive to a more active Atlantic hurricane sea One measure of the AMO is the standardized time son because it reduces the vertical wind shear and decreases the atmospheric stability in the western series of the detrended Kaplan AMO index (www MDR (Gray 1984; Tang and Neelin 2004). Cool neu .esrl.noaa.gov/psd/data/correlation/amon.us - - tral ENSO conditions prevailed during the other two .long.data). For ASO 2017, that index was +1.51 stan dard deviation (std. dev.), indicating the positive (i.e., peak months of the season (August and September). warm) phase of the AMO. The standardized 7-year - running mean (using ASO seasons only) of the de (iv) Atmospheric conditions trended Kaplan AMO index for ASO 2017 was +1.75 The atmospheric conditions within the MDR during ASO 2017 ref lected an inter-related set of std. dev. (Fig. 4.22a). Historically, the warm AMO is associated with the Atlantic high activity eras of anomalies which are typical of other extremely ac - tive seasons (Landsea et al. 1998; Bell et al. 1999, 1950–70 and 1995–present. Conversely, the Atlantic 2000, 2004, 2006, 2009, 2011, 2012, 2014, 2015, 2016; Goldenberg et al. 2001; Bell and Chelliah 2006; Kos - sin and Vimont 2007). Historically, the combination - of a warm AMO and La Niña yields the most spa tially extensive set of atmospheric conditions that are conducive for Atlantic hurricane activity, while the combination of El Niño and the cool AMO yields the least conducive conditions (Bell and Chelliah 2006). In the lower atmosphere, the conducive conditions during ASO 2017 included below-average heights/ sea-level pressure (blue shading, Fig. 4.23a) across - the MDR, along with weaker trade winds (i.e., west erly anomalies) extending from the eastern tropical North Pacific across the southern MDR to Africa. These westerly anomalies extended up to 700-hPa, the approximate level of the African easterly jet (AEJ), and were associated with a deep layer of anomalous . 4.22. SST time series for 1900–2017 based on ig F cyclonic relative vorticity across the entire MDR (Fig. ERSSTv4 (Huang et al. 2015). (a) Standardized (std. 4.23b). As noted by Bell et al. (2011), the increased dev.) 7-yr running mean of the detrended Kaplan AMO cyclonic shear along the equatorward f lank of the AEJ index based on the ASO season only. (b) Standardized helps the easterly waves within the MDR to be better (std. dev.) 7-yr running mean of the difference between ASO area-averaged SST anomalies in the MDR and maintained and also provides an inherent cyclonic those for the entire global tropics (20°N–20°S). rotation to their embedded convective cells. | S116 AUGUST 2018

137 . 4.24. ASO 2017: 200–850 hPa vertical wind shear F ig 1 − ) (a) total and (b) anomalies. magnitude (shaded; m s Overlaid in (a) are the total 200-hPa streamfunction 1 2 − 6 field (contours, interval: 5 × 10 m s ) and the 200–850 1 − hPa layer mean wind vectors (m s ) representing the steering current. The upper-level ridge and TUTT discussed in the text are labeled and denoted by thick black lines. Vector scale is below right of color bar. Green box denotes the MDR. Data is from NCEP– NCAR reanalysis (Kalnay et al. 1996). Anomalies are - departures from the 1981–2010 means on total sea F ig . 4.23. ASO 2017: (a) anomalous 1000-hPa heights sonal wind speeds (not shown). In (c) the upper-level (shaded; m); (b) anomalous 700-hPa cyclonic relative ridge and TUTT discussed in the text are labeled and 6 1 − vorticity (shaded; × 10 s - ); (c) 200-hPa total stream denoted by thick black lines. Vector scales are below 6 2 − 1 function (contours, interval: 5 × 10 - m ) and anoma s - right of color bar. Green box denotes the MDR. Anoma lies (shaded), from NCEP–NCAR reanalysis (Kalnay et lies are departures from the 1981–2010 means. al. 1996). The corresponding anomalous wind vectors − 1 in the western MDR along the southern f lank of the (m s ) are shown in each panel. In (b), the thick solid line indicates the axis of the mean African easterly jet anomalous ridge. The resulting vertical wind shear which was hand-drawn based on total seasonal wind (Fig. 4.24a) was also weaker than average across the speeds (not shown). In (c) the upper-level ridge and central and western MDR as well as in the vicinity of TUTT discussed in the text are labeled and denoted the Bahamas (Fig. 4.24b). by thick black lines. Vector scales are below right of −1 ) As a result, weak vertical wind shear (< 10 m s color bar. Green box denotes the MDR. Anomalies are extended across the entire MDR from Africa to Cen - departures from the 1981–2010 means. tral America, as well as northward over the western In the upper atmosphere at 200-hPa, the circu North Atlantic (Fig. 4.24a). Also, the associated - steering current (Fig. 4.24a, vectors) allowed African lation during ASO 2017 featured an extensive and persistent ridge of high pressure across the western easterly waves and named storms to track farther - westward into the region of anomalously weak verti half of the MDR and the western North Atlantic cal wind shear and exceptionally warm SSTs. These (Fig. 4.23c). This pattern was accompanied by an - eastward displacement of the tropical upper tropo conditions greatly increased the number and strength of the TCs within the MDR, as well as the number of spheric trough (TUTT) from the western MDR to the central MDR and central North Atlantic. Consistent landfalling hurricanes. The exceptionally strong and persistent ridge over with this pattern, the upper-level westerly winds were weaker than average (indicated by easterly anomalies) the western Atlantic was a crucial aspect of the 2017 | S117 AUGUST 2018 STATE OF THE CLIMATE IN 2017

138 Atlantic hurricane season. Although La Niña techni - cally developed in October, a La Niña–like pattern of tropical convection was already present in September. The rapid response in the upper-level atmospheric circulation to the developing La Niña likely helped maintain that ridge during October–November [a - period when two hurricanes (including Major Hur ricane Ophelia) and two tropical storms formed] and may have contributed to the September conditions as well. A pronounced ridge such as this was last seen in association with the record strong 2005 Atlantic hurricane season (Bell et al. 2006). Therefore, while the warm AMO and La Niña set the stage for an ex - tremely active 2017 Atlantic hurricane season, these combined climate factors alone do not likely account for the combined magnitude and duration of the western Atlantic ridge, which is seen less frequently. 3) n orth p acific and c entral n orth e astern acific basins — M. C. Kruk and C. J. Schreck p i ) Seasonal activity ( - The eastern North Pacific (ENP) basin is offi cially split into two separate regions for the issuance of warnings and advisories by NOAA’s National Weather Service. NOAA’s National Hurricane Center F ig . 4.25. Seasonal TC statistics for the full ENP/CNP in Miami, Florida, is responsible for issuing warnings basin over the period 1970–2017: (a) number of named in the eastern part of the basin (ENP) that extends storms, hurricanes, and major hurricanes and (b) the from the Pacific Coast of North America to 140°W, 4 2 ACE index (× 10 kt ) with the 2017 seasonal total high - while NOAA’s Central Pacific Hurricane Center in lighted in red. Horizontal lines denote the correspond - Honolulu, Hawaii, is responsible for issuing warnings ing 1981–2010 base period means for each parameter. in the central North Pacific (CNP) region between one storm in 2017: The remnants of Major Hurricane 140°W and the dateline. This section summarizes - the TC activity in both warning areas using com Fernanda moved from the ENP to the CNP as a weak bined statistics, along with information specifically tropical storm before dissipating around 146°W. The addressing the observed activity and impacts in the long-term 1981–2010 IBTrACS mean in the CNP basin is 4.7 storms making the 2017 season much CNP region. below average. The ENP/CNP hurricane season officially spans from 15 May to 30 November. Hurricane and tropi - cal storm activity in the eastern area of the basin (ii) Environmental influences on the 2017 season Figure 4.26 shows the mean environmental con - typically peaks in September, while in the CNP TC ditions that the ENP and CNP TCs experienced in activity normally reaches its seasonal peak in August (Blake et al. 2009). During the 2017 season, a total of 2017. The borderline weak La Niña is indicated by the cool SST anomalies along the equatorial eastern 18 named storms formed in the combined ENP/CNP basin (Fig. 4.25a). This total includes 9 hurricanes, 4 Pacific and warm anomalies to the north (Fig. 4.26a). Much of the TC activity was concentrated along the - of which were major hurricanes. The 1981–2010 IB Mexican coast, which is not unusual during La Niña TrACS seasonal averages for the basin are 16.5 named years (Collins and Mason 2000; Fu et al. 2017). The storms, 8.5 hurricanes, and 4.0 major hurricanes SST anomalies were slightly above normal in that (Schreck et al. 2014). 2 4 kt region. Mixing from the storms themselves may have The 2017 seasonal ACE index was 98.5 × 10 played a role in tempering those anomalies (Hart et (Fig. 4.25b), which is below the 1981–2010 mean of 4 2 al. 2007), but the OLR anomalies (Fig. 4.26b) were (Bell et al. 2000; Bell and Chelliah 132.0 × 10 kt also near-normal and suggest weaker-than-normal 2006; Schreck et al. 2014). The CNP basin only had | S118 AUGUST 2018

139 ig F . 4.26. May–Nov 2017 anomaly maps of (a) SST 2 − (°C; Banzon and Reynolds 2013), (b) OLR (W m ; 1 − Lee 2014), (c) 200–850-hPa vertical wind shear (m s ; vectors) and scalar (shading) anomalies, and (d) 850- − 1 2 − hPa winds (m s ; vectors) and zonal wind (shading) ; F ig . 4.27. Longitude–time Hovmöller of OLR (W m anomalies. Anomalies are relative to the annual cycle Lee 2014) averaged 5°–15°N. Unfiltered anomalies from 1981–2010, except for SST which is relative to from a daily climatology are shaded. Negative anoma - 1982–2010 due to data availability. Hurricane symbols lies (green) indicate enhanced convection. Anomalies with letters denote where each ENP TC attained tropi - filtered for Kelvin waves are contoured in blue at − 10 − 2 cal storm intensity. Wind data obtained from CFSR W m and MJO-filtered anomalies in black at ±10 W − 2 (Saha et al. 2014). m . Hurricane symbols and letters indicate genesis of ENP TCs. convection in the region. Weak easterly vertical A weak MJO event in early July likely contributed shear anomalies, on the other hand, did favor the to an active month that included five named storms, TC activity (Fig. 4.26c). Similarly, a narrow swath of 850-hPa westerly anomalies along 10°–15°N would including Major Hurricanes Eugene and Fernanda. have provided enhanced cyclonic vorticity, wave The subsequent dry phase of the MJO provided the accumulation, and/or barotropic energy conversion longest break (21 days) between named storm forma - tions from June through September. MJO activity (Maloney and Hartmann 2001; Aiyyer and Molinari 2008; Rydbeck and Maloney 2014). appeared to play less of a role in the remainder of the ENP TC activity is strongly inf luenced by the MJO season. However, Kelvin waves probably enhanced conditions for at least three tropical storms: Lidia, (Maloney and Hartmann 2001; Aiyyer and Molinari Ramon, and Selma. 2008; Slade and Maloney 2013), and recent studies have found a greater role for convectively coupled (iii) TC impacts Kelvin waves in modulating tropical cyclogenesis During the 2017 season, five named storms made (Schreck and Molinari 2011; Ventrice et al. 2012a,b; landfall along the western coast of Mexico or Baja Schreck 2015, 2016). Figure 4.27 uses OLR to examine the intraseasonal evolution of convection during the California, while the one storm in the CNP region did 2017 ENP hurricane season. Following Kiladis et al. not make landfall in Hawaii. The long-term annual average number of landfalling storms on the western (2005, 2009), the black contours identify the MJO- coast of Mexico is 1.8 (Raga et al. 2013); thus this year filtered anomalies and the blue contours identify the Kelvin waves. Easterly waves are also apparent in the was exceptional, in part due to the number of storms unfiltered anomalies (shading) as westward moving forming so close to the coast (Fig. 4.26). Tropical Storm Beatriz (31 May–2 June) was the features, such as those leading up to Tropical Storms Jova and Selma. first storm to make landfall in 2017 along the Mexi - can coast, followed closely by Tropical Storm Calvin | S119 AUGUST 2018 STATE OF THE CLIMATE IN 2017

140 1 , the (11–13 June). Both storms brought torrential rainfall activity considered. According to the JTWC - 2017 season had 26 named storms (which is the me and landslides to the Oaxaca and Guerraro areas of dian). These included 12 typhoons (bottom quartile coastal Mexico. Beatriz produced localized rainfall is ≤ 14) two of which reached super-typhoon (130 kt, of up to 380 mm, and five people were killed when −1 ) status (bottom quartile is ≤ 2). In Fig. 4.28a, mudslides washed away their homes and vehicles. 65m s the number of each category per season is shown for Additional rainfall from Calvin, while not as extreme as Beatriz, exacerbated relief efforts and compounded the period 1945–2017. While the number of tropical - - storms matched the climatological median, the num the already saturated soils leading to further land slides and mudslides. ber of typhoons and supertyphoons was below nor - mal. Only 46% of tropical storms became typhoons Tropical Storm Lidia (31 August–03 September) (bottom quartile is ≤ 57%). Further, the percentage tracked northwest along the entire Baja Peninsula. of typhoons reaching supertyphoon intensity (17%) While maximum sustained winds were 55 kt (29 m −1 was below normal (median is 24%). s ), the storm weakened dramatically as it crossed The JMA total for 2017 was 27 named storms over the mountainous terrain of the Baja Peninsula. The biggest impact from Lidia was heavy rainfall, up (above the median of 26; Fig. 4.28b). Guchol was - to 300 mm in San Jose Del Cabo, resulting in numer considered a tropical storm by JMA but only a tropical ous f looded streets and the cancellation of several depression by JTWC. Saola was considered a severe 2 dozen f lights from Mexico City International Airport. . tropical storm by JMA and a typhoon by JTWC In the city of Cuautitlán Izcalli, located in the central - The Philippine Atmospheric, Geophysical, and As tronomical Services Administration named all 22 state of Mexico, roughly 300 people were evacuated after the nearby El Ángulo dam collapsed, and in TCs that entered its area of responsibility, including Ecatepec de Morelos a nearby canal overf lowed, fill - Tropical Depression Bising (February) which was considered a tropical depression by JMA but was not ing many homes with sewage. The wind field from Lidia made a close approach to southern California tracked by JTWC. Only 41% of the storms reached in the United States, where gusty winds were reported typhoon intensity (bottom quartile is ≤ 50%). along area beaches. Hurricane Max (13–15 September) made landfall (ii) Seasonal activity in areas of the Mexican coastline already plagued The season had a slow start, with the first named tropical storm not developing until April (Muifa). No by tropical storms earlier in the season. Max brief ly - TCs formed in May and only one tropical storm (Mer intensified to hurricane strength about 12 hours prior bok) formed in June. In contrast, July was an active to landfall, with maximum sustained winds of 75 kt −1 month with 8 TCs (top quartile is 5) forming: Tropical (39 m s ). The city of Guerraro, f looded by Beatriz Storms Nanmandol, Talas, Sonca, Kulap, Roke, and and Calvin, was also affected by torrential rains from Haitang; Typhoon Nesat; and Supertyphoon Noru. Max. Two people died as rapidly rising rivers swept The two typhoons for July 2017 ranked among the away their residence. Meanwhile over the ocean, bottom quartile. Four TCs were simultaneously active large waves and swell, with peak wave heights of 3–5 m, sunk six boats before they could return to port. in the WNP during 21–23 July, 3 with TS strength on Tropical Storm Selma (27–28 October) was the 1 The TC data used here are from the Joint Typhoon Warning final storm of the 2017 season, and the first storm on Center (JTWC) western North Pacific best-track dataset record to make landfall in El Salvador. When com - for the 1945–2017 period and from the JTWC preliminary bined with a cold front moving through Honduras, operational data for 2017. Climatology is defined using the rainfall was widespread across the region, resulting period 1981–2010, with exception of landfall statistics, where in the overtopping of at least a dozen local rivers. 1951–2010 was used. The best-track data from the RSMC- Tokyo, Japan Meteorological Agency was used in Fig. 4.28b. basin — S. J. Camargo estern 4) w p acific n orth All other figures and statistics were obtained using JTWC (i) Introduction TC data. All statistics are based on the climatological dis - The TC season in the western North Pacific tribution (CLD), unless specifically stated that is based on (WNP) was below normal by most measures of TC the historical record. 2 It is well known that there are systematic differences between the JMA and the JTWC datasets, which have been extensively documented in the literature (e.g., Wu et al. 2006; Nakazawa and Hoshino 2009; Song et al. 2010; Ying et al. 2011; Yu et al. 2012; Knapp et al. 2013; Schreck et al. 2014). | S120 AUGUST 2018

141 F . 4.28. (a) Number of tropical storms (TS), typhoons (TY), and supertyphoons (STY) per year in the western ig North Pacific (WNP) for the period 1945–2017 based on the JTWC best-track dataset. (b) Number of tropical cyclones (TC; all storms which reach TS intensity or higher) from 1951 to 1976; number of TSs, severe tropi - cal storms (STS) and TY from 1977 to 2017 based on the JMA best-track dataset. Panels (c) and (e) show the cumulative number of tropical cyclones with TS intensity or higher (named storms) and number of TYs, per month in the WNP in 2017 (black line), and climatology (1981–2010) as box plots [interquartile range: box; median: red line; mean: blue asterisk; values in the top or bottom quartile: blue crosses; high (low) records in the 1945–2016 period: red diamonds (circles)]. Panels (d) and (f) show the number of NSs and TYs respectively, per month in 2017 (black line) and the climatological mean (blue line), the blue plus signs denote the maximum and minimum monthly historical records and the red error bars show the climatological interquartile range for each month (in the case of no error bars, the upper and/or lower percentiles coincide with the median. [Sources: 1945–2017 JTWC best-track dataset, 2017 JTWC preliminary operational track data for panels (a) and (c)–(f). 1951–2017 RSMCenter-Tokyo, JMA best-track dataset for panel (b).] | S121 AUGUST 2018 STATE OF THE CLIMATE IN 2017

142 22 and 23 July. August was also an active month with value of seasonal ACE in the historical record (Fig. 6 named storms (top quartile is ≥ 6): 3 tropical storms 4.29a). The only months when ACE was not below (Nalgae, Pakhar, and Mawar) and 3 typhoons (Ban the median were April and July; January–March, - yan, Hato, Sanvu), each matching the median for that May, June, September, and November all had ACE month. Only 4 TCs (bottom quartile is ≤ 4) formed in values in their respective bottom quartiles. The bulk of the seasonal ACE occurred in July and August (Fig. September: Tropical Depressions Guchol, 22W, and 4.29b), with those months contributing 25% and 26% Typhoons Talim and Doksuri, with only two storms reaching tropical storm or typhoon intensity, which of the total ACE respectively, followed by October is in the bottom quartile for both distributions (≤ 4 (21%). The ACE values in September and November and ≤ 2.5, respectively). The TC activity increased were the 9th and 11th lowest for those months in the historical record. somewhat in October, with three typhoons (Khanun, Lan, and Saola), in the bottom quartile for named Only 3 typhoons in 2017 were in the top quartile for ACE per storm: Supertyphoons Noru and Lan, storms (≤ 3) but matching the median for typhoons. and Typhoon Talim, contributing 26.6%, 13.4%, and Lan was the second storm of the season to reach 10.8% of the seasonal ACE, respectively. Combined, supertyphoon intensity in the season. In November they accounted for just over half of the seasonal ACE. there were 2 tropical storms (Haiku and Kirogi) and 1 The only storm in the top decile was Supertyphoon typhoon (Damrey), which ranked in the top quartile for named storms (≥ 3) but in the bottom quartile for Noru. It should be noted that Noru contributed to the typhoons (≤ 1). The 2017 typhoon season concluded with two December TCs: Tropical Storm Kai-Tak and Typhoon Tembin, each in the top quartile for their respective categories. The early season (January–June) totals (2 tropical storms and no typhoons) were in the bottom quartile of all storm counts (≤ 3 and ≤ 1, respectively). In con - trast, the peak season (July–October) had 19 named storms (median is 17) and 10 typhoons (median is 12). The late season (November and December) total of 5 named storms and 2 typhoons was in the top quartile for named storms (≥ 4) and equal to the median for typhoons. The overall character of the season was a normal number of TCs, but a low number attaining - typhoon intensity, with the greatest TC activity con centrated from July to August. (iii) Environmental conditions During the peak and latter part of the season, the tropical Pacific SST transitioned from neutral to weak La Niña conditions. The mean genesis location in 2017 was at latitude 15.8°N, longitude 129.9°E, which was a shift northwestward from the climatological mean of latitude 13.2°N, longitude 141.6°E (standard - deviation 1.9° latitude and 5.6° longitude. This north ig F . 4.29. (a) ACE index per year in the western North westward shift is typical during La Niña years (e.g., Pacific for 1945–2017. The solid green line indicates Chia and Ropelewski 2002; Camargo et al. 2007). the median for the climatology (1981–2010), and the dashed lines show the climatological 25th and 75th The mean track position of 19.5°N, 133.7°E was also percentiles. (b) ACE index per month in 2017 (black northwestward relative to the WNP climatological line) and the median during 1981–2010 (blue line), the mean of 17.3°N, 136.6°E (standard deviations of 1.4° red error bars indicate the 25th and 75th percentiles. latitude and 4.7° longitude). Therefore, these shifts - In case of no error bars, the upper and/or lower per were consistent with a La Niña event. centiles coincide with the median. The blue “+” signs Also consistent with a weak La Niña, the total denote the maximum and minimum values during the ACE in 2017 was below normal (Camargo and Sobel 1945–2016. (Source 1945–2016 JTWC best-track data - set, 2017 JTWC preliminary operational track data.) 2005), in the bottom quartile, and the eighth lowest | S122 AUGUST 2018

143 . 4.30. (a) SST anomalies (°C) for Jul–Oct (JASO) 2017. (b) PI anomalies (kt) in JASO 2017. (c) Relative humid - F ig ity 600-hPa relative humidity anomalies (%) in JASO 2017. (d) GPI anomalies in JASO 2017. First positions of storms in JASO 2017 are marked with an asterisk. (e) Zonal winds in JASO 2017 (positive contours are shown in solid lines, negative contours in dash dotted lines and the zero contour in a dotted line) [Source: atmospheric variables: NCEP/NCAR Reanalysis data (Kalnay et al. 1996); SST (Smith et al. 2008).] ACE values for both July and August, as it was active 1948, when the data reliability was much lower). The from 20 July to 9 August. median lifetime of the 2017 season for named storms There were 85 days with named storms. From and typhoons was 4.5 and 5.6 days, respectively, both in the bottom quartile (≤ 6.3 and ≤ 7.8 days). The these active days, 36 had typhoons and 6 had major typhoons (categories 3–5), all in the bottom quar longest living storm was Supertyphoon Noru, which - lasted 19.5 days (20 July–9 August), which places it tiles. The percentage of days during the season with in the 98th percentile for all WNP named storms typhoons and major typhoons were 28.6% and 4.8%, since 1945. Tropical Storm Kai-Tak (10.8 days) was - respectively in the bottom quartile of their distribu - the only other WNP named storm in 2017 in the top tions (≤ 33% and ≤ 10%, respectively). The percent age of major typhoons days is the sixth lowest in the quartile (≥ 10.5 days). All other storms in 2017 had lifetimes at or below the median. The occurrence of historical record (two of those happened in 1945 and | S123 AUGUST 2018 STATE OF THE CLIMATE IN 2017

144 short-lived storms this season is typical of La Niña Tropical Storm Kai-Tak (named Urduja in the Philip - pines) made landfall causing 160 deaths and leaving years (Camargo and Sobel 2015) and related to the 4 northwest shift of TC activity. 163 missing . The costliest typhoon in the season was Including tropical depressions, 26 storms made - Typhoon Hato, with damages totaling almost $7 bil 3 in 2017, ranking in the 95th percentile com lion U.S. dollars, impacting Macau and Hong Kong, - landfall pared with the 1951–2010 climatology. Of these, 11 as well as several provinces along the Pearl River, where storm surge caused major f looding in various made landfall as tropical depressions (second highest in the historical record), 7 as tropical storms (median provinces of mainland China. Hato was the strongest is 6), 8 as typhoons (top quartile is ≥ 7), and none as typhoon to hit Macau and Hong Kong in 50 years. major typhoons (bottom quartile is ≤ 1). Vietnam ndian was hit by 9 storms this season, including Typhoon o 5) n orth i basin cean — M. C. Kruk The North Indian Ocean (NIO) TC season typically Damrey, which was the strongest typhoon to make extends from April to December, with two peaks in ac - landfall in south-central Vietnam in 16 years, and tivity: during May–June and again in November, when Typhoon Doksuri which affected the northern and the monsoon trough is positioned over tropical waters central Vietnam provinces. The median number of in the basin. TCs in the NIO basin normally develop landfalls in Vietnam per year is 4.5; 9 landfalls (at any intensity) is in the 90th percentile of the climatologi over the Arabian Sea and Bay of Bengal between 8° and - cal distribution of landfalls there. 15°N. These systems are usually short-lived, relatively - weak, and often quickly move into the Indian subcon Figure 4.30 shows the environmental conditions tinent (Gray 1968; Schreck et al. 2014). associated with the typhoon activity in 2017. The - main feature is the borderline weak La Niña with According to the JTWC, the 2017 TC season pro duced three tropical storms, one cyclone, and no major - below-normal SST anomalies in the eastern and cen cyclones (Fig. 4.31a). The 1981–2010 IBTrACS seasonal tral Pacific during July–October (JASO; Fig. 4.30a) - averages for the basin are 3.9 tropical storms, 1.4 cy and slightly above normal SST in the WNP. This SST clones, and 0.6 major cyclones (Schreck et al. 2014). The pattern is ref lected in other environmental fields, as 2 4 kt seasonal ACE index was 15.8 × 10 , which is near the can be seen in potential intensity (PI; Emanuel 1988; 2 4 1981–2010 mean of 16.3 × 10 (Fig. 4 . 31b). Ty pic a l ly, kt Fig. 4.30b), 600-hPa relative humidity (Fig. 4.30c), there is enhanced TC activity, especially in the Bay of and genesis potential index (GPI; Emanuel and Nolan 2004; Camargo et al. 2007; Fig. 4.30d) anomalies, Bengal, during the cool phase of ENSO (Singh et al. 2000). While this season was not yet a fully-developed which were positive in the western part of the basin La Niña, two storms developed in the Bay of Bengal and and negative in the eastern part, typical of La Niña only one system, Tropical Storm Four (9 December), years. The GPI anomalies had a maximum near and - developed in the Arabian Sea. east of the Philippines, in the region of high occur The second named storm of the season was Cyclone rence of TC formation. The maximum extent of the Mora (27–30 May), which had maximum sustained monsoon trough, as defined by the zonal wind (Fig. −1 winds of 65 kt (33 m s 4.30e) maximum extension, was confined to the area ) and a minimum central pressure of 978 hPa. The cyclone caused dramatic west of 130°E, consistent with the westward shift of impacts across Sri Lanka, the Andaman Islands, and the genesis location in 2017. Bangladesh due to widespread f looding rains and (iv) TC impacts significant storm surge. At landfall, the storm surge Many storms had significant social and economic - was a stunning 3 m above astronomical high tide, re sulting in an inland penetration of saltwater nearly 20 impacts in 2017. Typhoon Tembin, known as Vinta km. The government of Bangladesh estimated 52 000 in the Philippines, struck the Philippine province of Mindanao in late December, causing 200 deaths with homes were destroyed by the storm which displaced an 172 missing, making it the deadliest WNP TC of 2017. estimated 260 000 people. In Sri Lanka, Cyclone Mora Tembin hit the Philippines less than one week after exacerbated ongoing f looding from an active period of the southwest monsoon, resulting in numerous 3 Landfall is defined when the storm track is over land and f loods and landslides, killing more than 200 people the previous location was over ocean. In order not to miss and displacing 630 000 more. landfall over small islands, first the tracks were interpolated 4 Casualty statistics are from the Relief Web site; for Tembin see from 6-hourly to 15 minute intervals before determining https://reliefweb.int/disaster/tc-2017-000182-phl and for Kai- if the storm track was over land or ocean using a high- Tak, see https://reliefweb.int/disaster/tc-2017-000180-phl. resolution land mask. | S124 AUGUST 2018

145 M. C. Kruk and 6) s outh i ndian o cean basin C. Schreck — The South Indian Ocean (SIO) basin extends south of the equator from the African coastline to 90°E, with most cyclones developing south of 10°S. The SIO TC season extends from July to June encompassing equal portions of two calendar years (the 2017 season includes storms from July to December 2016 and from January to June 2017). Peak activity typically occurs during December–April when the ITCZ is located in the Southern Hemisphere and migrating toward the equator. Historically, the vast majority of landfalling cyclones in the SIO affect Madagascar, Mozambique, and the Mascarene Islands, including Mauritius and Réunion Island. The Regional Specialized Meteoro - logical Centre (RSMC) on La Réunion serves as the official monitoring agency for TC activity within the basin. The 2016/17 SIO storm season was below aver - age with five named storms, of which two were cy - clones and one was a major cyclone (Fig. 4.32a). The F i g . 4.31. Annual TC statistics for the NIO for 1970–2017: (a) number of tropical storms, cyclones, and major cyclones and (b) estimated annual ACE 2 4 index (in kt × 10 ) for all TCs at least tropical storm strength or greater intensity (Bell et al. 2000). The 1981–2010 means (horizontal lines) are included in both (a) and (b). The most intense storm in the basin was Cyclone Ockhi late in the season, from 29 November to 6 December, with maximum sustained winds of 90 kt −1 (45 m s ) and a minimum central pressure of 976 hPa. The storm originated over Sri Lanka and moved west-northwest into the Arabian Sea and then turned northeast where it was affected by a cold continental airmass which led to its quick demise west of the Gujarat coastline. However, the storm again plagued areas of Sri Lanka with additional rainfall and gale- force winds. The strong winds forced the diversion of f lights to Mattala and closed schools. Farther west across the Maldives, two cargo boats were capsized by the cyclone, with more than a dozen other boating ig . 4.32. Annual TC statistics for the SIO for 1980– F incidents reported during the height of the storm. As 2017: (a) number of tropical storms, cyclones, and the storm turned northeast back toward India, it gener - major cyclones and (b) estimated annual ACE index (in ated large ocean swells which led to substantial erosion 4 2 kt × 10 ) for all TCs at least tropical storm strength along the west-facing Mumbai beaches. In addition, or greater intensity (Bell et al. 2000). The 1981–2010 the cyclone-generated waves deposited over 80 00 kg 0 means (horizontal lines) are included in both (a) and of trash and debris on the Mumbai beaches following (b). Note that ACE is estimated due to lack of consis - 125 mm of rainfall. tent 6-h sustained winds for each storm. | S125 AUGUST 2018 STATE OF THE CLIMATE IN 2017

146 suggested a broad area of unfavorable subsidence (Fig. 1981–2010 IBTrACS seasonal median averages are 4.33b). The western half of the basin, on the other eight tropical storms, four cyclones, and one major hand, experienced westerly vertical shear anomalies cyclone (Schreck et al. 2014). The 2016/17 seasonal 2 −1 4 , which would have precluded in excess of 4.5 m s , which is about one- kt ACE index was 30.8 × 10 4 2 kt third of the 1981–2010 average of 91.5 × 10 (Fig. significant activity there. 4.32b), and the lowest since the 2010/11 season. SSTs During the 2016/17 season, the strongest storm was Cyclone Enawo (3–10 March), which reached and 850-hPa winds were both near normal in 2016/17 category 4 equivalent with peak sustained winds of (Figs. 4.33a,d). The quiet season likely relates more to −1 changes in the upper-level circulation. Positive OLR 125 kt (64 m s ) and an estimated minimum central pressure of 932 hPa. The storm was the strongest anomalies across the eastern portion of the basin to strike Madagascar since Gafilo in 2004. Enawo initially developed near the center of the basin out of the monsoon trough and gradually strengthened as it headed southwest towards Madagascar. The intense cyclone attained its maximum intensity just prior to landfall on 7 March before impacting the towns of Sambava and Antalaha. According to advisories from the RSMC La Réunion, storm surge was estimated to be 3–4 m across these areas, which ultimately led to swamped rice fields, displaced residents, and an estimated 81 fatalities due to the storm. ustralian basin — B. C. Trewin 7) a (i) Seasonal activity The 2016/17 TC season was near normal in the - broader Australian basin (areas south of the equa 5 tor and between 90° and 160°E , which includes the Australian, Papua New Guinean, and Indonesian areas of responsibility), despite a late start, with only one cyclone before mid-February. The season pro - duced nine TCs (Fig. 4.34), near the 1983/84–2010/11 6 average of 10.8, and consistent with neutral to cool ENSO conditions. The 1981–2010 IBTrACS seasonal averages for the basin are 9.9 named storms, 7.5 TCs, and 4.0 major TCs. All references to TC category in this section use the Australian Bureau of Meteorology TC intensity scale. 7 There were six TCs in the western sector of the - Australian region during 2016/17, three in the north 8 - ern sector, and one in the eastern sector . Three sys tems made landfall in Australia as tropical cyclones, two in Western Australia and one in Queensland, 5 - The Australian Bureau of Meteorology’s warning area over laps both the southern Indian Ocean and southwest Pacific. . 4.33. Jul 2016–Jun 2017 anomaly maps of (a) SST ig F 6 Averages are taken from 1983/84 onwards as that is the start 2 − ; Lee (°C; Banzon and Reynolds 2013), (b) OLR (W m of consistent satellite coverage of the region. − 1 2014), (c) 200 850-hPa zonal wind shear (m s − ; vectors) 7 The western sector covers areas between 90° and 125°E. The and scalar (shading) anomalies, and (d) 850-hPa winds eastern sector covers areas east of the eastern Australian − 1 (m s ; vectors) and zonal wind (shading) anomalies. - coast to 160°E, as well as the eastern half of the Gulf of Car Anomalies are relative to the annual cycle from 1981– pentaria. The northern sector covers areas from 125°E east 2010, except for SST which is relative to 1982–2010 due to the western half of the Gulf of Carpentaria. to data availability. Letter symbols denote where each 8 Frances passed through both the western and northern SIO TC first attained tropical storm intensity. Wind data obtained from CFSR (Saha et al. 2014). sectors. | S126 AUGUST 2018

147 bie then moved southwest into inland Queensland, weakening below cyclone intensity by 1600 UTC. The remnant low then took a south to southeast track through Queensland, passing back out to sea near Brisbane late on 30 March. A wind gust of 142 kt −1 (73 m s ), the strongest measured gust on record in Queensland, was observed at the elevated Hamilton −1 Island Airport site on 28 March, and 89 kt (46 m s ) at Proserpine. There was extremely heavy rainfall near landfall, as well as from the remnant low; totals near landfall included 635 mm in 24 hours at Mount Jukes and 986 mm in 48 hours at Clarke Range on 28–29 March, while near the Queensland–New South Wales border, 24-hour totals on 31 March included 602 mm at Upper Springbrook, 507 mm at Chillingham, and 478 mm at Boat Harbour. A 2.6-m storm surge (0.9 m above highest astronomical tide) was observed at Laguna Quays, north of Mackay. Debbie caused extensive wind damage in the Whitsunday region on the mainland and on offshore islands, including Airlie Beach, Proserpine, Bowen, Hamilton, and Daydream Islands, and inland to Col - linsville. There was also severe f looding both in the region near landfall, including the fifth highest height on record for the Fitzroy River at Rockhampton and in the Logan, Albert, and Tweed catchments near . 4.34. Annual TC statistics for the Australian basin ig F the Queensland–New South Wales border. Moisture for 1980–2017: (a) number of tropical storms, cyclones, and major cyclones and (b) estimated annual ACE from the remnant low also contributed to major 4 2 index (in kt × 10 ) for all TCs at least tropical storm f looding in parts of the North Island of New Zealand strength or greater intensity (Bell et al. 2000). The on 4–5 April, including the inundation of large parts 1981–2010 means (horizontal lines) are included in of the town of Edgecumbe where a stopbank of the both (a) and (b). Note that ACE is estimated due to Rangitaiki River was breached on 6 April. In total, lack of consistent 6-h sustained winds for each storm. three direct deaths and several indirect deaths were attributed to Debbie, while insured losses for Debbie while two others made landfall after weakening below tropical cyclone intensity. The first landfall in Australia, according to the Insurance Council of of the season did not occur until 6 March, the latest Australia, were assessed at $1.565 billion AUS dol - first landfall since comprehensive satellite records lars ($1.207 billion U.S. dollars), the second-largest began in 1970. (inf lation-adjusted) insurance loss on record for an Australian tropical cyclone (after Cyclone Tracy in (ii) Landfalling and other significant TCs 1974). An additional $91.5 million NZ dollars ($66.7 The most significant cyclone of the season was million U.S. dollars) of insured damages happened in New Zealand from Debbie’s extratropical remnants. Debbie, which affected eastern Australia in late March. Debbie formed as a tropical disturbance south Blanche formed as a tropical low within a trough over the Arafura Sea on 2 March. It began to of Papua New Guinea and initially moved south, strengthen on the 3rd and moved southwest on the reaching cyclone intensity on 24 March near 17°S, 4th while strengthening, crossing over the Tiwi 152°E, before turning southwest and intensifying. Islands (northwest of Darwin) early on 5 March. It reached its peak intensity of category 4 while just Continuing to move southwest over the Timor Sea, off the Queensland coast at 0000 UTC on 28 March, it reached tropical cyclone intensity at 1200 UTC on with maximum 10-minute sustained winds of 95 kt −1 ). It later made landfall at 0240 UTC (1240 the 5th, when approximately 200 km west of Darwin. (49 m s It strengthened further to category 2 while moving local time) just north of Airlie Beach, by which time southwest, with peak 10-minute sustained wind it had weakened slightly to a category 3 storm with −1 −1 - speeds of 55 kt (28 m s maximum sustained winds of 80 kt (41 m s ), before making landfall ). Deb | S127 AUGUST 2018 STATE OF THE CLIMATE IN 2017

148 −1 at that intensity at 0300 UTC on 6 March, on the ) at sity (maximum sustained winds 115 kt (62 m s northeast Kimberley coast of Western Australia - 1200 UTC on 7 April near 16°S, 111°E, before turn between Kalumburu and Wyndham. Point Fawcett, ing west-southwest and weakening, dropping below on the Tiwi Islands, received 384 mm of rain in the - tropical cyclone intensity on 10 April. The other ma jor cyclone of the season was Frances, which reached 24 hours prior to 0900 local time on 5 March, its wettest day on record, while in the Kimberley, the category 3 intensity on 28–29 April, with maximum −1 highest recorded 48-hour rainfall was 207 mm at Me ), as it tracked west- sustained winds of 70 kt (36 m s No Savvy, between Fitzroy Crossing and Halls Creek. southwest through the Timor Sea between Timor and the Australian mainland. Neither Ernie nor Frances Tropical cyclone warnings were issued for Darwin but approached any land areas, although heavy rain as - no major impacts occurred there. sociated with Frances did affect the Tiwi Islands. The third landfall of the season occurred on 23 March, at 0500 UTC just west of Port Hedland. The outhwest 8) s original low formed on 19 March north of the Kim P. R. Pearce, A. M. Lorrey, — basin - acific p and H. J. Diamond berley coast, before moving west and then southwest (i) Seasonal activity and intensifying shortly before landfall. The cyclone The 2016/17 season in the southwest Pacific of - was not named operationally but was analyzed as - a category 2 system based on post-analysis [maxi ficially began in November 2016, but the first named −1 mum sustained winds 50 kt (26 m s )] on the basis storm did not occur until February 2017, despite numerous tropical depressions during the early part of observed surface winds, including a gust of 61 kt −1 (32 m s of the season. Storm track data for November 2016– ) at a beacon offshore from Port Hedland. April 2017 was gathered from the Fiji Meteorological There was minor wind damage in the Port Hedland Service, Australian Bureau of Meteorology, and New area and significant river rises in the Pilbara coastal Zealand MetService, Ltd. The southwest Pacific basin rivers, De Grey River, and Fortescue River. Minor to as defined by Diamond et al. (2012) (135°E–120°W) major f looding occurred at some locations in the De - had six tropical cyclones, including three major tropi Grey catchment. Port Hedland received 268 mm of cal cyclones (based on the Australian TC intensity rain during 22–24 March. scale). As noted in Section 4f1, Fig. 4.35 shows the Yvette, in late December, and Alfred, in mid- standardized TC distribution based on the basin February, were both cyclones that weakened below - spanning the area from 160°E–120°W to avoid over cyclone intensity before making landfall near Broome laps with the Australian basin that could result in in Western Australia and the Northern Territory/ - Queensland border, respectively. Alfred peaked off double counting of storms. However, it is important shore as a category 2 and Yvette as a category 1. to use the definition of the southwest Pacific basin of Moisture from Yvette combined with a separate Diamond et al. (2012) as that is how annual TC out - looks are produced and disseminated. All references tropical low to bring heavy rains through a large area of central and southern Australia in the final to TC category in this section use the Australian TC intensity scale. days of December. Walungurru, near the Northern - The 1981–2010 Southwest Pacific Enhanced Ar Territory/Western Australia border, received 287 mm of rain during 25–26 December, while Adelaide (61.2 chive of Tropical Cyclones (SPEArTC) indicates a seasonal average of 10.4 named tropical cyclones and mm on 28th) had its third wettest December day on 4.3 major tropical cyclones. Therefore, the 2016/17 TC record. There was significant f lash f looding in parts season had less-than-normal activity. The first storm of metropolitan Melbourne. Record high dewpoints (Tropical Cyclone Alfred) developed as a tropical dis - and precipitable water levels were observed at nu - turbance in the Gulf of Carpentaria in mid-February. merous sites in South Australia and Victoria. Alfred The season concluded in mid-May with Tropical brought some f looding and minor wind damage, and Cyclone Ella affecting Wallis and Futuna and Samoa. 862 mm of rain was recorded from 18 to 22 February The ratio of major TCs relative to the total number at Sweers Island. The most intense Australian tropical cyclone of named TCs in 2016/17 was 50%, down from 63% during the previous season. Tropical Cyclone Donna, of the season was Ernie. This storm formed as a - tropical low on 4 April near 10°S, 115°E, well north which caused significant damage in northern Vanu of Western Australia. Ernie reached tropical cyclone atu and the Solomon Islands in May, was the strongest intensity late on 6 April near 14°S, 111°E, and then TC to form outside the official southwest Pacific TC intensified exceptionally rapidly, reaching category season (which ended on 30 April 2017) on record per - 5 intensity within 24 hours. It reached its peak inten the SPEArTC dataset (Diamond et al 2012). | S128 AUGUST 2018

149 Port Vila, Vanuatu. Cook brought heavy rain and destructive winds to parts of New Caledonia, where one fatality was reported. Cook also caused wind damage to trees and infrastructure in parts of New Zealand’s North Island, one week after ex-Tropical Cyclone Debbie caused major f looding in the same area. Cook achieved category 3 status with 10-minute −1 sustained winds of 84 kt (43 m s ) and a minimum central pressure of 961 hPa. Tropical Cyclone Donna formed to the north of Vanuatu on 1 May, which is just past the traditional end of the season (30 April). It achieved named storm status on 3 May, and late on 4 May it began to show a clear eye and was upgraded to a category 3 tropical cyclone. On 6 May, Donna was upgraded to category 4 status. It weakened to a category 3 storm later on 6 May but then strengthened again to category 4 status the next day before being upgraded to category 5 status on 8 May. Donna’s peak 10-minute sustained −1 wind speed reached 111 kt (57 m s ) and its lowest minimum central pressure was 935 hPa. As a result, Donna became the strongest out-of-season TC on - record for May in the southwest Pacific. Donna de graded quickly to tropical low strength on 10 May. The storm caused significant damage in Vanuatu. Entire villages across the Torres Islands in Torba . 4.35. Annual TC statistics for the southwest Pacific ig F Province were forced to seek shelter from the storm for 1980–2017: (a) number of tropical storms, cyclones, in caves. Throughout the province, many buildings - and major cyclones and (b) estimated annual ACE in 4 2 were destroyed or severely damaged. On the island of dex (in kt ; Bell et al. 2000). The 1981–2010 means × 10 (horizontal lines) are included in both (a) and (b). Note Efate, heavy rainfall led to f looding of low-lying areas. that ACE is estimated due to lack of consistent 6-h Structures collapsed in Port Vila because they were sustained winds for each storm. undermined during f lash f loods. Across the northern half of Vanuatu, crops sustained significant damage (ii) Landfalling and other significant TCs and communications were severed with the rest of Tropical Cyclone Alfred developed as a tropical the country. In the Temotu Province of the Solomon low on 16 February in the southern Gulf of Carpen - Islands, Donna caused two fatalities. In New Zealand, taria. The low gradually intensified into a category Donna’s remnants produced heavy rain over much 1 TC on 20 February and remained at TC strength of the North Island and the west coast of the South before weakening approximately 24 hours later. Al - Island on 11–12 May. fred was the first tropical cyclone to make landfall in The season concluded with Tropical Cyclone Ella, Australia’s Northern Territory since 2015. Alfred’s which formed southwest of American Samoa on 9 −1 peak 10-minute wind speed was 46 kt (24 m s May. Just three hours later, the system intensified ) and into a category 1 TC, and it reached category 2 status its lowest central pressure was 994 hPa. on 10 May. Its peak 10-minute sustained wind speed Tropical Cyclone Bart was a short-lived cyclone −1 was 59 kt (31 m s which lasted from 19 to 22 February, forming south ) with a minimum central pressure of 977 hPa. of Samoa and traveling southeast to the south of the southern Cook Islands. Bart reached category 1 g. Tropical cyclone heat potential— status, where peak 10-minute sustained wind speeds G. J. Goni, J. A. Knaff, −1 were 40 kt (21 m s ) and minimum central pressures I.-I. Lin, and R. Domingues This section summarizes the changes in upper reached 994 hPa. ocean thermal conditions within the seven tropical Tropical Cyclone Cook was named on 8 April cyclone (TC) basins (see Table 4.1), using tropical cy - after forming northeast of Vanuatu. Some trees clone heat potential (TCHP; Goni and Trinanes 2003) were felled and power was cut to some residents in | S129 AUGUST 2018 STATE OF THE CLIMATE IN 2017

150 dynamics (Lin and Chan 2015) than to upper ocean as the main parameter. The assessment presented conditions. here focuses on the vertically-integrated upper ocean In the Gulf of Mexico, TCHP anomalies ranged temperature conditions during the TC season of each −2 between −10 and 20 kJ cm - with the spatial distribu ocean basin with respect to the long-term mean and tion largely determined by the mesoscale field, such as to values observed during the previous year. TCHP the extension of the Loop Current, and cold cyclonic is defined as the excess heat content contained in the water column between the sea surface and the depth features. In the eastern Gulf of Mexico, prominent of the 26°C isotherm. This parameter has been linked intrusion of the Loop Current caused TCHP values to TC intensity changes (Shay et al. 2000; Mainelli in 2017 to be 50% larger than the mean; a noticeable et al 2008; Lin et al. 2014) with TCHP values above change with respect to conditions in 2016, which −2 providing the necessary ocean conditions was characterized by a small intrusion of the Loop 50 kJ cm Current. The TCHP in the western Gulf of Mexico for Atlantic hurricane intensification when favorable once again exhibited positive anomalies, with values atmospheric conditions are present. The magnitude approximately 30% larger than the long-term mean. of the TCHP has been identified as modulating the Compared to 2016, TC activity increased in the Gulf effective SST under a TC during air–sea coupling of Mexico in 2017 with a total of five TCs including due to latent and sensible heat f luxes (Mainelli et al. the rapidly intensifying category 4 Hurricane Harvey. 2008; Lin et al. 2013). In addition, improved temporal and spatial sampling of the ocean has been shown to In the eastern North Pacific (ENP) basin, TCHP −2 values were 10–20 kJ cm lead to the correct representation of the upper ocean above the long-term mean - associated with a continued positive phase of the Pa density field (Domingues et al. 2015), which in turn led to reducing the error in hurricane intensifica cific decadal oscillation (Zhang et al. 1997). Anoma - - lies observed in 2017, however, were not so large as tion forecasts within operational numerical models (Dong et al. 2017). Fields of TCHP show high spatial the values observed in 2016. This change is largely due to the ENSO conditions described in Section 4b. and temporal variability associated mainly with As a consequence, average TC activity was observed oceanic mesoscale features, year-to-year variability in the ENP, with nine hurricanes in 2017 (Fig. 4.36). (e.g., ENSO), or long-term decadal variability. The assessment of this variability on various timescales The TCHP in the WNP basin is also closely modu - lated by ENSO variability (Lin et al. 2014; Zheng et can be accomplished using a combination of satellite altimetry and in situ observations (Goni et al. 1996; Lin et al. 2008; Goni and Knaff 2009; Pun et al. 2013). To assess year-to-year variations in TCHP, two fields are presented. First, Fig. 4.36 presents TCHP anomalies (departures from the 1993–2016 mean values) for the primary months of TC activity in each hemisphere: June–November in the Northern Hemisphere, and November 2016–April 2017 in the Southern Hemisphere. TCHP anomalies gener - ally show large variability within and among the TC basins. Figure 4.37 shows the differences of TCHP between this season (2017) and last year (2016). Most basins exhibited positive TCHP anomalies 2 − F - . 4.36. Global anomalies of TCHP (kJ cm ) corre ig in 2017 (Fig. 4.36), except for a small region just east sponding to 2017 computed as described in the text. of 60°E in the southwest Indian basin. Above-average Boxes indicate the seven regions where TCs occur: - TCHP in most basins provided anomalously favor from left to right, southwest Indian, north Indian, west able ocean conditions for the intensification of TCs. North Pacific, southeast Indian, South Pacific, East Pacific, and North Atlantic (shown as Gulf of Mexico In the tropical Atlantic basin, TCHP values observed and tropical Atlantic separately). Green lines indicate in 2017 were approximately 10% larger than the long- the trajectories of all tropical cyclones reaching at least term mean, consistent with the above-normal activity Saffir–Simpson category 1 during Nov 2016–Apr 2017 there. Meanwhile, the western North Pacific (WNP) in the SH and Jun–Nov 2017 in the NH. The numbers basin had below-normal activity despite TCHP values above each box correspond to the number of category being over 30% larger than the mean conditions. This 1 and above cyclones that traveled within each box. is explained because the number of TCs in the WNP Gulf of Mexico conditions during Jun–Nov 2017 are shown in the inset in the lower right corner. during a season is more closely related to atmospheric | S130 AUGUST 2018

151 al. 2017b). Hurricanes Irma and Maria, for example, −1 ) had sustained winds that reached 160 kt (67 m s −1 ), respectively. Both storms were and 150 kt (72 m s well observed by reconnaissance aircraft equipped with stepped frequency microwave radiometers that provide accurate estimates of surface wind speeds (Uhlhorn and Black 2003). An increase in TCHP values with respect to the previous season was recorded in the North Indian - Ocean (Arabian Sea), southeast Indian Ocean, south west Pacific, and WNP ocean basins. The largest changes with respect to the previous season were 2 − F ) between 2017 ig . 4.37. TCHP differences (kJ cm observed in the south Indian Ocean basin, and in and 2016. the WNP north of 10°N, with differences above −20 −2 and 20 kJ cm respectively. Super Typhoon Noru was the fifth named storm to develop during the season al. 2015). For example, from the 1990s to 2013 the WNP experienced a long-term decadal surface and and experienced rapid intensification from tropical storm into a category 5 TC as it moved from an area subsurface warming associated with more prevalent −2 of low TCHP (~40 kJ cm La Niña-like conditions (Pun et al. 2013; England ) into an area with TCHP −2 . et al. 2014; Lin and Chan 2015). With the ENSO values of ~80 kJ cm - conditions during 2014/15, however, this warming Ocean conditions of four of the six major hurri canes (Harvey, Irma, Jose, and Maria) of the Atlantic trend stopped, but it recovered again in 2016. In 2017, further warming of the WNP basin and TCHP basin are described here. Data from the ocean observ - −2 anomalies as large as 40 kJ cm ing system, including observations from underwater were observed, which is approximately 30% larger than the long-term mean gliders that were deployed to collect data in support for the region. However, the overall TC activity over of operational hurricane intensity forecasts, are the WNP basin was not so active as in 2016 due to presented here (Fig. 4.38). These observations were - less favorable atmospheric dynamic conditions (Lin collected because a better representation of the up per ocean temperature and salinity conditions has and Chan 2015; Section 4f4). - been shown to reduce the error in Atlantic hurricane For each basin, the differences in the TCHP val ues between this season and 2016 (Fig. 4.37) indicate intensity forecasts within the NOAA experimental HYCOM-HWRF operational model (Dong et al. that three of the seven active TC basins exhibited a decrease in TCHP values, namely the: (1) South 2017). Ocean conditions before, during, and after Indian Ocean, (2) eastern North Pacific Ocean, and the passage of these hurricanes were continuously (3) North Atlantic Ocean basins. It is likely that lower monitored by some of these gliders. Hurricane Harvey traveled through the Caribbean TCHP values in the south Indian Ocean played a role in suppressing TC activity in 2016/17, which observed Sea south of Puerto Rico on 20 August, where the up - per ocean exhibited TCHP values higher than 80 kJ only one major TC during the season. However, −2 despite showing a moderate decrease in TCHP with cm . In this area, underwater glider data showed that respect to 2016, above-normal TC activity in terms of a relatively shallow mixed layer favored cooling of the upper ocean, which together with the moderate wind category 4 and 5 storms was observed in the tropical Atlantic and Gulf of Mexico, with the development shear contributed to its lack of that intensification in region. Once it reached the Gulf of Mexico, Hurricane of six major Atlantic hurricanes. Intense hurricane Harvey intensified from a tropical depression into a activity in the Atlantic during the last season likely −1 benefited from above-normal TCHP in the tropical category 4 hurricane with 115 kt (51 m s ) winds in a period of less than 48 hours as it traveled over positive Atlantic and Gulf of Mexico (Fig. 4.36) combined with favorable atmospheric conditions associated TCHP anomalies in the western Gulf of Mexico. Har - vey produced the largest amount of rain on record in with a cool neutral ENSO state, which is known for the continental United States, which caused extensive decreasing vertical wind shear and trade wind inten - f looding in the Houston, Texas, metropolitan area sity, supporting TC development and intensification (see Sidebar 4.3 for detailed information about the (Gray 1984). In addition, atmospheric conditions in precipitation associated with Harvey). the tropical Atlantic, as described in Section 4f2, fa - vored the development of intense TC activity (Bell et | S131 AUGUST 2018 STATE OF THE CLIMATE IN 2017

152 F ig . 4.39. (a), (b) Temperature (°C) and (c) salinity pro - files sampled by underwater gliders before and after the passage of three major North Atlantic hurricanes (Irma, Jose, and Maria) in 2017. weakening from a category 4 hurricane to category 3 during this time. - Hurricane Maria traveled through the eastern Ca ribbean Sea and later through the same approximate area as Irma transited the tropical North Atlantic, F ig . 4.38. Tracks of major Atlantic hurricanes that On 20 September, after entering the Caribbean Sea traveled over the Caribbean Sea and tropical North following a landfall in Dominica, Maria peaked in Atlantic Ocean during the 2017 hurricane season. Blue intensity with maximum sustained winds of 150 kt lines indicate the location of some of the underwater −1 (77 m s ) and a minimum pressure of 908 hPa, mak - gliders, which were parked in fixed locations (green ing Maria the tenth-most intense Atlantic hurricane stars) during the passage of the major hurricanes. Background colors show values of TCHP averaged on record. When Maria’s path was close to the glid - − 2 for Aug 2017, with thin contours every 10 kJ cm , ers in the Caribbean Sea, these ocean observations − 2 − 2 and thick contours indicate 50 kJ cm , and 80 kJ cm revealed the existence of a very stable barrier layer respectively. of approximately 30-m depth (Fig. 4.39c) providing Hurricane Irma, the strongest TC globally in 2017, ocean conditions conducive for intensification. Maria −1 made landfall in Puerto Rico on 20 September as an ) reached its maximum intensity of 160 kt (82 m s intense category 4 hurricane. Interaction with land on 6 September while traveling over waters north - further weakened the hurricane, though it regained of Puerto Rico and Hispaniola that had TCHP val −2 . Underwater glider data ues higher than 70 kJ cm some strength as it traveled over waters with TCHP −2 north of Hispaniola (Fig. 4.38). values of ~70 kJ cm showed that the upper ocean conditions exhibited low As it traveled farther to the north it encountered lower salinity values at the surface, partially suppressing TCHP which helped to contribute to Maria’s weaken - upper ocean mixing with colder underlying waters, similar to what happened with Hurricane Gonzalo in ing to a tropical storm on 28 September. 2014 (Domingues et al. 2015; Dong et al. 2017), but In summary, 2017 was characterized by higher- opposite to the conditions experienced during Hur than-normal values of TCHP by 10%–30% over - ricane Harvey. Glider observations also revealed that most TC basins. Overall, TCHP anomalies observed the upper 50 m of the ocean cooled by approximately in 2017 were not so large as anomalies observed in 2016, which likely contributed to both fewer overall 1°C (Fig. 4.39a) as a result of storm-induced mixing. TCs as well as fewer category 5 TCs globally. Ocean Hurricane Jose was the third strongest Atlantic observations during 2017 indicated that upper ocean hurricane in 2017 and was the seventh longest-lived conditions may have favored the intensification of Atlantic named storm in the satellite era (since 1966). major TCs, but atmospheric conditions (especially While Jose was off Puerto Rico, 2°–3° latitude to the north of where Irma traveled, its trajectory coincided - in the western North Pacific) were likely not as con ducive for strong TCs. at a time with the cold wake left behind by the passage of Hurricane Irma. Therefore, Jose experienced a rela - h. Indian Ocean dipole— tively cooler and well mixed upper ocean as observed J.-J. Luo The Indian Ocean dipole (IOD), referring to the by underwater glider data (Fig. 4.39b). These cooler ocean conditions may have partly contributed to its anomalous SST gradient between the western and eastern equatorial Indian Ocean, is a major internal | S132 AUGUST 2018

153 climate mode in the tropical Indian Ocean (IO). It event was quite weak and uncoupled. The positive often starts to grow during boreal summer, peaks in west-minus-east zonal SST gradient did not bring anomalous easterlies along the equatorial IO during September–November, and ends rapidly in December April–August (Fig. 4.40b). Moreover, while the cold in association with the reversal of monsoonal winds along the west coast of Sumatra (Saji et al. 1999). The SST anomalies in the eastern IO and warm anomalies in the western IO formed a positive dipole SST pattern IOD displays a strong asymmetry with the magnitude during April–August, local rainfall anomalies did not of the positive IOD being much larger than that of the follow the SST anomalies. Instead, positive rainfall negative IOD (e.g., Hong et al. 2008). Correspond - ingly, air–sea coupling strength and predictability anomalies occurred in the eastern IO, while drier - of the positive IOD are usually strong and high, re conditions occurred in the western IO (Fig. 4.40a). spectively, compared to those of the negative IOD Following the strong El Niño event of 2015/16, back-to-back La Niña events occurred in late 2016 (Luo et al. 2007). Following a negative IOD event in 2016 (Luo 2017), - and late 2017 (Figs. 4.1, 4.40c). In addition, a nega tive IOD started in May 2016 and persisted until a positive IOD event developed during April–August - 2017, despite the occurrence of neutral ENSO condi January 2017 (Fig. 4.40b). Correspondingly, during December 2016–February 2017, basin-wide cold SST tions during this time (Fig. 4.40). This positive IOD F ig . 4.40. (a) Monthly anomalies of SST (°C; solid lines) 1 − and precipitation (mm day ; dashed lines) in the east - ern pole (IODE; 10°S–0°, 90°–110°E; blue lines) and the western pole (IODW; 10°S –10°N, 50°–70°E; red lines) - of the IOD. (b) As in (a), but for the IOD index (mea sured by the SST difference between IODW and IODE, − 1 green line) and surface zonal wind anomaly (m s ) in the central equatorial IO (Ucio; 5°S –5°N, 70°–90°E; ig F . 4.41. SST (°C; colors) and precipitation (contoured 1 − black line). (c) As in (a), but for the SST anomalies in ). Solid/ at: 0, ±0.5, ±1, ±2, ±3, ±4, and ±5 mm day o-3.4 region (5°S –5°N, 190°–240°E; black line) the Ni ñ dashed lines denote positive/negative values, and thick and the tropical IO (IOB; 20°S –10°N, 40°–120°E; red solid lines indicate the zero contour) anomalies during line). Anomalies are relative to the 1982–2017 base (a) Dec 2016–Feb 2017, (b) Mar–May 2017, (c) Jun–Aug period. [Sources: NOAA OISST (Reynolds et al. 2002); 2017, and (d) Sep–Nov 2017. Anomalies were calcu - monthly GPCP precipitation analysis (available at lated relative to 1982–2017. [Sources: NOAA OISST http://precip.gsfc.nasa.gov/); and JRA-55 atmospheric (Reynolds et al. 2002) and monthly GPCP precipitation reanalysis (Ebita et al. 2011).] analysis (available at http://precip.gsfc.nasa.gov/).] | S133 AUGUST 2018 STATE OF THE CLIMATE IN 2017

154 anomalies appeared in the tropical IO, and warm cold Rossby waves and anomalous southerlies in the southeastern IO favor the occurrence of cold SST SST anomalies were observed around Indonesia (Fig. anomalies there (Figs. 4.41a,b). 4.41a). Consistently dry conditions occurred in the western–central IO with wet conditions in the eastern During January–July, La Niña dissipated rapidly IO to the Maritime Continent. Westerly anomalies - and warm anomalies appeared in the central–east ern equatorial Pacific (Fig. 4.40c). However, the wet were present in the central–eastern equatorial IO condition continued in the eastern IO–Maritime that helped deepen the thermocline in the east and Continent while the western IO remained dry (Fig. generate warm upper ocean temperature in that re - gion (Figs. 4.41a, 4.42a). Cyclonic wind anomalies in 4.41b). The persistent wet condition around Indonesia is consistent with a strong increasing SST trend there. the southeastern IO, which often happen following La Niña and/or a negative IOD (Behera et al. 2006; Moreover, the corresponding westerly anomalies in Luo et al. 2010), tend to upwell the local thermo - the central–eastern equatorial IO did not generate cline and drive westward-propagating cold Rossby warm upper ocean temperature anomalies in the east, probably owing to the arrival of eastward-propagat waves. During December 2016–February 2017, cold - upper–300-m mean temperature anomalies occurred ing equatorial cold Kelvin waves. The cold subsurface along 10°S and in the western IO, reminiscent of the anomalies helped generate cold SST anomalies along Rossby wave activities (Fig. 4.42a). Meanwhile, the the west coast of Sumatra (Fig. 4.41b), which may have prevented the development of a negative IOD. Meanwhile, SSTs in the western IO increased during F ig . 4.41. SST (°C; colors) and precipitation (contoured 1 − ). Solid/ at: 0, ±0.5, ±1, ±2, ±3, ±4, and ±5 mm day dashed lines denote positive/negative values, and thick . 4.42. Upper 300-m mean ocean temperature ig F 1 − solid lines indicate the zero contour) anomalies during (°C; colored scale) and surface wind (m s ) anomalies (a) Dec 2016–Feb 2017, (b) Mar–May 2017, (c) Jun–Aug during (a) Dec 2016–Feb 2017, (b) Mar–May 2017, (c) - 2017, and (d) Sep–Nov 2017. Anomalies were calcu Jun–Aug 2017, and (d) Sep–Nov 2017. [Sources: NCEP lated relative to 1982–2017. [Sources: NOAA OISST ocean reanalysis (available at www.cpc.ncep.noaa.gov (Reynolds et al. 2002) and monthly GPCP precipitation /products/GODAS/) and JRA-55 atmospheric reanaly - analysis (available at http://precip.gsfc.nasa.gov/).] sis (Ebita et al. 2011).] | S134 AUGUST 2018

155 March–August (Fig. 4.41b,c), partly due to less cloud driven by the corresponding weak La Niña (Lim and cover (i.e., dry condition) and strong increasing SST Hendon 2017), the positive dipole SST pattern in 2017 may be largely caused by the internal mechanisms in trend in that region. Thus, a positive dipole SST pat - the IO that are responsible for the biennial character tern formed. However, the persistent wet condition - of the IOD (Behera et al. 2006). Cold subsurface tem around Indonesia tends to induce westerly anomalies perature anomalies in the southern IO, which were in the eastern IO, which prevents the occurrence of induced by the 2016 negative IOD, may have provided a positive air–sea feedback to intensify the positive an important precursor for the occurrence of the cold dipole SST pattern. During September–November, in association with the development of the second SST anomalies in the eastern IO in 2017. However, the La Niña event, a negative IOD signal with westerly annually persistent anomalous westerlies in the IO, anomalies in the central IO became apparent. associated with the persistent wet condition around In summary, the positive IOD event in 2017 was Indonesia, suppressed the positive air–sea interaction - weak and uncoupled. It did not appear to exert sig during the positive IOD event in 2017 and may have nificant impacts on the climate in surrounding areas. led to the occurrence of a weak uncoupled positive Since the negative IOD in 2016 does not appear to be dipole SST event in April–August 2017. | S135 AUGUST 2018 STATE OF THE CLIMATE IN 2017

156 SIDEBAR 4.1: HURRICANE IRMA: REWRITING THE — RECORD BOOKS P. J. KLOTZBACH Hurricane Irma generated the highest ACE values (Bell which provides six-hourly estimates of historical Atlantic et al. 2000) of any Atlantic hurricane during the extremely tropical cyclone wind speeds, pressures, and locations active 2017 season. Irma developed from a tropical wave since 1851 (Landsea and Franklin 2013). in the eastern Atlantic, reaching tropical storm status on Irma began to set records as it approached the north - 30 August. Over the next several days, Irma intensified ern Leeward Islands. It intensified into a 155-kt (80-m −1 ) category 5 hurricane late on 5 September, making it into a major hurricane in an environment of anomalously s the strongest Atlantic hurricane outside of the Gulf of weak vertical wind shear and anomalously high SSTs. Mexico and Caribbean on record. Irma also shattered the On 5 September, Irma reached category 5 intensity - old record for strongest hurricane to impact the north as it bore down on the northern Leeward Islands. Over ern Leeward Islands (defined as 15°–19°N, 65°–60°W), the next several days, Irma devastated many islands in −1 ) set by the breaking the old record of 140 kt (72 m s the eastern and central Caribbean, then went on to make Lake Okeechobee Hurricane of 1928 and Hurricane David landfall in Cuba before making two landfalls in Florida. It (1979). Irma brought devastation to Barbuda (Fig. SB4.1), - finally weakened to a tropical depression early on 12 Sep Anguilla, and portions of the U.S. and British Virgin Islands tember near the Georgia/Alabama border. In this sidebar, and then passed north of Puerto Rico. During its track several of Hurricane Irma’s most notable meteorological across the Caribbean, Irma made four category 5 land - records are highlighted. All statistics for Irma listed in this falls at: Barbuda, St. Martin, Virgin Gorda (British Virgin sidebar are from the formal National Hurricane Center Islands), and Little Inagua (Bahamas). - report on Hurricane Irma (Cangialosi et al. 2018). Histori cal statistics are calculated from the HURDAT2 database, ig . SB4.1. GOES-16 infrared satellite image of Hurricane Irma as it made landfall over Barbuda at F 0600 UTC on 6 Sep 2017. | S136 AUGUST 2018

157 Despite weakening slightly as it tracked across the ing landfall near Cudjoe Key, Florida (Fig. SB4.2). Irma’s Caribbean, Irma maintained its category 5 intensity for landfall pressure in the Florida Keys of 931 hPa tied with 2.75 consecutive days—the longest contiguous period that Hurricane Carla (1961) for the tenth lowest on record for an Atlantic hurricane has spent at category 5 intensity in a continental U.S. landfalling hurricane. This also marked the satellite era (since 1966). It became the first category the first time on record that two category 4 hurricanes - 5 hurricane to make landfall in the Bahamas since Hur (Harvey and Irma) made landfall in the continental U.S. ricane Andrew in 1992. Irma briefly weakened to category in the same calendar year. Irma made a second landfall 4 strength but then re-intensified to category 5 before near Marco Island as a category 3 hurricane. At the time making landfall in Cuba on 9 September (Fig. SB4.2). The of its second landfall, Irma had maximum winds of 100 kt −1 ) and a central pressure of 936 hPa—the exact - last category 5 hurricane to hit Cuba was the Cuba Hur (51 m s same maximum sustained winds and 4 hPa lower central ricane of 1924. pressure than Hurricane Wilma had when it made landfall Land interaction with Cuba caused Irma to weaken to a in virtually the exact same location in 2005. category 3 hurricane, but it then re-intensified to category 4 over the warm waters of the Florida Straits before mak - ig . SB4.2. GOES-16 infrared satellite image of Hurricane Irma making landfall near Cudjoe Key on F 1315 UTC on 10 Sep 2017. | S137 AUGUST 2018 STATE OF THE CLIMATE IN 2017

158 THE NEW GOES-R SERIES: MUCH IMPROVED SIDEBAR 4.2: “GLASSES ” TO VIEW THE TROPICS — C. S. VELDEN NOAA’s Geostationary Operational Environmental What are the implications of improved hurricane intensity analyses and forecasts? - Satellites (GOES) have historically been one of the prin The primary mission at NHC/CPHC is to save lives, ciple tools utilized by tropical analysis and forecast centers mitigate property loss, and improve economic recovery to monitor hurricane activity. NOAA’s National Hurricane efficiency by issuing the best possible watches and warn - Center (NHC), Central Pacific Hurricane Center (CPHC), - ings of approaching hazardous tropical weather condi - and Satellite Analysis Branch (SAB), as well as the Depart tions. The 2017 Atlantic hurricane season was historic, ment of Defense Joint Typhoon Warning Center (JTWC), with notable landfalling Hurricanes Harvey, Irma, and employ GOES data and derived products for critical Maria. These storms were powerful examples of devas - analysis of storm intensity and motion. Over the years, tating disasters that could have been even worse if not algorithms have been developed to estimate hurricane for the accurate and timely track forecasts and warnings intensity from GOES imagery. The new GOES-R series issued by the NHC. While hurricane track forecasts (-R/S/T/U which become -16/17/18/19 when operational) have generally improved, less progress has been made - includes an advanced imager with improved spatiotem with intensity forecasts, which has prompted the NHC poral and spectral resolution that will enable better as - to elevate this issue to its top priority for the tropical sessment of hurricane structure and intensity. The first of meteorology research community. While gains clearly this series, GOES-16, was operated in experimental mode have been made, the losses due to the hurricanes in 2017 °W. It was declared operational for much of 2017 near 90 show that work remains to be done to fully address the °W to by NOAA in December 2017 and positioned at 75 goals set by the NHC. cover the Atlantic hurricane belt. . SB4.3. Multispectral GOES-16 imagery: (a) infrared window (10.3 μm) (b) and (c) water vapor ig F (6.19 μm, upper right, 7.34 μm, respectively,) and (d) visible (0.64 μm), at 2130 UTC on 5 Sep 2017 during Hurricane Irma. | S138 AUGUST 2018

159 In addition to operational aspects of hurricane inten - Both the DT and ADT will benefit from the improved sity estimation, climate analyses depend heavily on the attributes of the GOES-R series imager. The superior fidelity of the estimates. Trends in hurricane intensity sensor performance and higher spatiotemporal resolu - (along with frequency, duration, and landfalls) may be tion provide an improved ability to characterize storm linked to climate change, and these records are intrinsi - cloud patterns and detect features such as emerging eyes. cally dependent on satellite analyses. The GOES, along For example, Fig. SB4.3 shows the sharp contrast of the with counterparts around the world (e.g., Meteosat and - warm eye and cold eyewall in Hurricane Irma. This infor Himawari), have been the backbone of the satellite-based mation translates into more confident DT/ADT intensity observing system since the late 1970s. estimate analyses, which can be used in conjunction with data from the GOES-R series Geostationary Lightning How will the GOES-R series address hurricane intensity? Mapper (GLM) instrument. The GLM on GOES-16 is the The most common use of satellite imagery to estimate first operational lightning mapper flown in geostationary tropical cyclone intensity is via the Dvorak technique orbit and maps total lightning (in-cloud and cloud-to- (DT; Dvorak 1984), which employs recognizable patterns ground) continuously over the Americas and adjacent in enhanced infrared and/or visible satellite imagery to ocean regions. Data from GLM will inform forecasters quantitatively estimate the intensity of a tropical system. about changes in lightning activity in the eyewall and rain Indications of continued development and/or weakening bands of hurricanes, which can be used as an indicator can also be found in the cloud features. Trained satellite of intensity changes, especially rapid intensification (De - analysts identify the cloud pattern types, and along with Maria 2012; Xu and Wang 2018; Stevenson et al. 2018). a series of standardized technique rules, a fairly accurate Improved hurricane intensity analyses from the GOES-R intensity analysis can be made. An objective offshoot series should result in better intensity forecasts and also of the DT is the advanced Dvorak technique, or ADT benefit the fidelity of the climate record. (Olander and Velden 2007). The ADT follows some of the same procedures and rules as the DT, but it is com - pletely computer-based and includes many enhancements to the DT. | S139 AUGUST 2018 STATE OF THE CLIMATE IN 2017

160 SIDEBAR 4.3: HURRICANE HARVEY: THE HALLMARK STORM OF A BUSY AND WET 2017 HURRICANE SEASON FOR THE — D. M. ROTH AND J. W. NIELSEN-GAMMON UNITED STATES The 2017 tropical cyclone season was busy for the United States, with nine Atlantic, Caribbean, and Gulf of Mexico sys - tems affecting the nation. Harvey, originally a tropical storm over the western tropical Atlantic and eastern Caribbean Sea, traversed the Yucatán Peninsula, then redeveloped in the Bay of Campeche. It made landfall on the evening of 25 August five miles east of Rockport, as the first category 4 or stronger storm to make landfall in Texas since Carla in 1961. Its 3-m 0 00 homes destroyed and another storm surge resulted in 15 0 25 00 damaged. Remarkably, there were no deaths caused by storm surge or wind damage during landfall (Blake and Zelinsky 2018), perhaps attributable to NHC issuance of storm surge watches and warnings made operational in early 2017. Harvey’s impact, and memory of the storm, however, will be associated with its historic inland rainfall and associated F ig . SB4.4. GOES-16 ABI Band 1 (0.47 μm) and color- flooding. It is the wettest known tropical cyclone to impact the coded GLM parallax-corrected observations of light - ning groups in the 5 minutes prior to the nominal time United States, on a number of time and spatial scales. of the ABI image (red: oldest; yellow: latest), 1247 UTC After landfall, positioned near a col in the steering flow, on 25 Aug 2017, just prior to the rapid intensification Harvey’s forward motion slowed to a virtual halt about 100 km of Hurricane Harvey. inland. Harvey quickly weakened to tropical storm strength but maintained this status over land. It eventually moved southeast, average of 838 mm falling across Harris County—roughly moving out over the Gulf of Mexico during the morning of 28 two-thirds its typical annual rainfall—represents over a trillion August. Still a tropical storm, Harvey curved northeastward gallons of water. and made another landfall in southwest Louisiana early on Several factors contributed to Harvey’s extensive rainfall 30 August. The storm then accelerated northeastward and footprint and extreme volume across southeast Texas. It spent weakened as it neared the Ohio River Valley. nearly 60 hours inland at tropical storm or greater intensity, the At the station scale, daily rainfall totals exceeding 254 mm longest such duration over Texas. The cyclone moved slowly, occurred on five successive days as the storm wandered across with a continuous fetch of warm, humid Gulf air. It was large in the area. The highest Harvey storm total precipitation pres - size, based on its radius of tropical storm force winds and radius ently recognized by the National Weather Service is 1538.7 of outermost closed isobar (ROCI). During its overland time in mm at an automated gauge one mile southwest of Nederland, Texas. This far exceeds the previous known tropical cyclone record of 1320.8 mm. For the same gauge, the three-day total of 1338.1 mm appears to exceed any previously measured U.S. value for any type of event. Rainfall at Jack Brooks Regional Airport near Nederland shattered records for wettest day (661 mm vs. 324 mm), August (1390 mm vs. 438 mm), month (1390 mm vs. 578 mm), and summer (1814 mm vs. 804 mm). Houston Intercon - - tinental Airport recorded its wettest 1–6 days, Au gust, month, and year on record. Houston’s monthly total doubled the previous record associated with Tropical Storm Allison in June 2001. Area-averaged totals appear to far exceed any F ig . SB4.5. Observed rainfall totals in association with Harvey previously measured in the United States. The and its remnants. (Source: Weather Prediction Center, NOAA.) | S140 AUGUST 2018

161 Texas, it interacted with a weak frontal boundary that provided some additional focus for convection (Blake and Zelinsky 2018). Additionally, the area of heaviest rain pivoted from the storm’s northeast to - northwest quadrants prior to its landfall in Loui siana, generally keeping the storm’s heaviest rains over southeast Texas during that time. The magnitude of its rainfall was captured well by numerical weather prediction guidance. NOAA’s Weather Prediction Center forecasts indicated 600+ mm areal average amounts by 24 August, and areal average amounts of 1000+ mm by 25 August. The annual exceedance, or recurrence interval, for rainfall of this magnitude in southeast Texas was less than 0.1% in any given year (per the current NOAA Atlas 14), or less frequent than once in 1000 years. Harvey’s rainfall totals have been included in the preliminary version of NOAA Atlas 14 Version 11 for Texas ( 2018, manuscript under review ). Several studies have already examined Har - F ig . SB4.6. Tropical Storm Allison (2001) versus Hurricane vey’s rainfall in the context of climate change. For Harvey (2017) rainfall (mm) in southeast Texas (images use the example, van Oldenborgh et al. (2017) found that same color scale). trends in three-day rainfall totals along the north - too much water within the reservoirs would also increase the ern Gulf Coast accounted for an increased chance of Harvey- risk of uncontrolled releases or even dam failure (Brust 2017). like rainfall occurring within the region in any given year, from - The unprecedented flooding presented numerous chal 0 roughly 1 in 27 00 to 1 in 9000, with similar trends found in - lenges for disaster response. At Houston, familiar with lo forced climate simulations. calized flooding, the simultaneous inundation of watersheds For portions of southeast Texas, Harvey became the flood throughout the metropolitan area exhausted the capacity of of record. Lake Conroe exceeded its previous record maxi - first responders to conduct water rescues. Public officials called mum lake level, set in October 1994, by 18.3 cm. Major and on the public to help with evacuation. Hundreds responded record flooding occurred in the bayous of Houston and along with boats, jet skis, and even monster trucks (Sullivan 2017; rivers from the Colorado to the Sabine. The entire town of Collier 2017). Rising floodwaters caused primary and backup Port Arthur was submerged. water supply systems in Beaumont to fail. Several months later, Prior to Harvey, water crossing between the basins of the some residents of southeast Texas were still required to boil Sabine/Calcasieu Rivers, the Neches/Sabine Rivers, and, more water for drinking (Gstalter 2017). Aid for Beaumont, virtually unusually, between those of the San Bernard and Colorado inaccessible by land from Texas, arrived from Louisiana. Rivers had been observed. However, the magnitude and dura - Of the 68 fatalities directly caused by Harvey, 65 were due tion of basin crossovers during Harvey is unique in hydrologic - to freshwater flooding. About 35 additional deaths are indi records. 00 structures rectly attributable to Harvey. An estimated 300 0 West of Houston, two flood control reservoirs, normally were flooded, nearly half of those in Harris County. Up to dry, rapidly filled. Many residents discovered their homes 500 00 vehicles were also flooded. Total direct damages from 0 were built within a reservoir footprint, though beyond the Harvey are estimated by NOAA at approximately $125 billion reservoir’s 100-year floodplain. Reservoir operators faced a U.S. dollars, making Harvey the second-costliest United States difficult challenge: flooding was unavoidable, but the release tropical cyclone in inflation-adjusted dollars, behind Hurricane rates from the reservoirs would determine how much flooding Katrina (Blake and Zelinsky 2018). would occur within, versus below, the reservoirs. Retaining | S141 AUGUST 2018 STATE OF THE CLIMATE IN 2017

162 | S142 AUGUST 2018

163 5. THE ARCTIC — J. Richter-Menge, M. O. Jeffries, and Arctic since 2005. In 2017, snow cover extent was E. Osborne, Eds. again below the 1981–2010 average across the North E. Osborne, J. Richter-Menge, and M. O. Jeffries a. Introduction— American Arctic, driven by earlier snow melt across the Canadian Arctic. Annual average Arctic air temperatures (above 60°N) in 2017 continued to increase at twice the Terrestrial permafrost, a critical component of - the Arctic landscape, supports much of the built in rate of the rest of the world, with the annual average frastructure in the region (e.g., buildings, highways, surface air temperature second highest (2016 ranked - airstrips, pipelines) and continues to experience no first) since the year 1900. Extreme warm conditions table change. Climate variables, such as atmospheric were particularly prevalent in Alaska at the end of - temperature, rain events, and snow depths, are driv 2017 when the atmospheric circulation drove warm southern air masses into the Pacific Arctic region. ing higher permafrost temperature and increasing - The same wind pattern, along with high sea sur active layer thickness (surface soil layer that thaws face temperatures, slowed the southward advance and refreezes seasonally). In 2017, five of six per - of the sea ice edge, leading to a month-long delay mafrost observatories on the North Slope of Alaska in autumn freeze up in the Chukchi Sea and Bering reported record warm permafrost temperatures. In Strait regions of the Pacific Arctic, setting another the same region, tundra greening, or an increase in new record for the satellite era (1978–present). Ear - above-ground vegetation, has been linked to changes in the permafrost active layer thickness, the warming lier in the year, on 7 March, the Arctic sea ice winter Arctic climate, the extended growing season, and maximum extent measured by satellite was the lowest even reductions in sea ice cover. Over the 35-year on record (since 1979), covering 8% less area than the - observational time series, tundra greenness has in 1981–2010 mean. The 2017 sea ice minimum on 13 creased throughout the majority of the circumpolar September was the eighth lowest on record and cov - ered 25% less area than the long-term mean. Ten of Arctic, most notably on the North Slope of Alaska, Canadian low Arctic tundra, and eastern Siberia. the lowest September sea ice minimum extents have been recorded in the last eleven years. Continued loss Another phenomenon, tundra browning, is emerging in the relatively sparse regions of western Alaska, the of thick multiyear ice (evidenced by <1% multiyear Canadian Archipelago high Arctic, and northwestern ice present in March 2017 relative to 16% in 1985) Siberia and may be attributed to winter warming also contributes a positive feedback to ice loss, as the majority of today’s sea ice is thin first-year ice prone events and perhaps even insect outbreaks. The Arctic tundra is also impacted by wildland fires, which are to breakup and melt. increasing as a result of warming climate conditions. As summer sea ice extents continue to shrink While 2017 was an average wildfire season in Alaska back, seasonal buildup of upper ocean heat in ice-free - regions is increasing. In August 2017, sea surface tem 04 acres burned), significantly warmer and 9 (652 perature (SST) records were broken for the Chukchi - drier conditions in the Upper Yukon zone of north Sea, with some regions as warm as +11°C, or 3° to east Alaska resulted in high fire danger for much of - 4°C warmer than the long-term mean (1982–pres the season and accounted for more than half of the ent). Most other boundary regions and marginal seas, acres burned in the United States. which are typically ice free during summer months, High above the Arctic, atmospheric ozone con - also had anomalously warm SSTs in 2017. As in winter centrations in winter 2016/17 were unremarkable 2016/17, the delayed freeze up in the Pacific Arctic in and well above previous record minima in 2010/11 late 2017 extended the exposure of the upper ocean and 2015/16. UV radiation, which depends on at - in the Chukchi Sea to the sun’s heat. Mean SSTs from mospheric ozone concentrations and other factors, 1982–present show statistically significant warming varied in time and space across the Arctic. trends over much of the Arctic. While observational time series are central to After a rapid start to the Greenland ice sheet melt monitoring Arctic change, paleoclimate reconstruc - season in early April, moderate to below-average melt tions based on fossil records can help scientists place persisted for much of the remainder of the season. the rates and magnitudes of modern change into a As a result, summertime area-averaged albedo for long-term, geological context. Arctic paleoceano - the entire Greenland ice sheet was the third high - graphic records indicate that the magnitude and est value since 2000. Glaciers and ice caps outside sustained rate of declining sea ice trend observed of Greenland continue to show declining trends in the modern era is unprecedented in any existing in cumulative mass balance. Long-term terrestrial high resolution paleoclimate sea ice reconstruction snow cover estimates show dramatic declines in the for at least the last 1450 years. Similarly, according | S143 AUGUST 2018 STATE OF THE CLIMATE IN 2017

164 to paleoclimate studies, today’s abnormally warm Arctic air and sea surface temperatures have not been observed in the last 2000 years. Indigenous knowl - edge gathered by Arctic Peoples over many millennia is another means to holistically understand Arctic change beyond instrumental records. Coproduction of knowledge can bring together knowledge systems - of scientists and indigenous knowledge–holders to de velop suitable sustainability and adaptation practices to address issues arising from the changing Arctic system (see Sidebar 5.2). . 5.1. Arctic (land stations north of 60°N) and ig F global mean annual land surface air temperature (SAT) anomalies (°C, 1981–2010 base period) for 1900–2017. b. Surface air temperature— . Overland, E. Hanna, J - Note that there were few stations in the Arctic, par I. Hanssen-Bauer, S.-J. Kim, J. E. Walsh, M. Wang, U. S. Bhatt, ticularly in northern Canada, before 1940. (Source: and R. L. Thoman CRUTEM4 dataset.) Arctic surface air temperature is an indicator of both regional and global climate change. Although - Arctic amplification. Mechanisms for Arctic am plification include: reduced summer albedo due to - natural variability contributes to year-to-year and re losses of sea ice and snow cover; the increase of total gional differences in air temperature, the magnitude - of the long-term temperature trend across the entire water vapor content in the Arctic atmosphere; a sum mer decrease and winter increase in total cloudiness Arctic is an indicator of global climate change and the (Makshtas et al. 2011); the additional heat generated impact of increasing greenhouse gas concentrations (Overland 2009; Notz and Stroeve 2016). by newly sea ice–free ocean areas that are maintained After a warm Arctic-wide autumn 2016, early later into the autumn (Serreze and Barry 2011); and 2017 had notable short-term, regional temperature the lower rate of heat loss to space in the Arctic, anomalies in response to a highly variable jet stream. - Spring and summer 2017 had near-average air tem peratures relative to the 1981–2010 climatology. The spring and summer conditions were reminiscent of those occurring before the long-term, above-average temperature increases began in the 1990s. Rather than higher sea level pressure extending over much of the Arctic, as observed in many recent years, weak low pressures were seen in 2017—a return to a wind forcing typical from a decade ago. The atmospheric forcing in spring and summer 2017 is consistent with a year when some Arctic indicators ran counter to the recent long-term trend over the previous decade. For - example, Eurasian spring snow extent was above aver age for the first time in over a decade (see Section 5i). At +1.6°C, the mean annual 2017 surface air temperature (SAT) anomaly for land stations north of 60°N is the second highest value (after 2016) in the record starting in 1900 (Fig. 5.1). Despite near- average temperatures during spring and summer months, extreme heat during autumn and winter, particularly over the Chukchi Sea and extending northward to the pole, contributed to near-record ig F . 5.2. Seasonal anomaly patterns for near-surface air breaking warm conditions in 2017 (Fig. 5.2). Cur - temperatures (°C, 1981–2010 base period) for 2017 in rently, the Arctic is warming at more than twice the (a) JFM, (b) AMJ, (c) JAS, and (d) OND. Temperatures rate of lower latitudes. are from slightly above the surface layer (925 mb) to The greater rate of Arctic temperature increase, emphasize large spatial patterns rather than local features. (Source: NOAA/ESRL.) compared to the global increase, is referred to as | S144 AUGUST 2018

165 of a warm feature observed in March. This regional relative to the subtropics, due to lower mean surface temperatures in the Arctic (Pithan and Mauritsen warming supported early sea ice loss in the Chukchi 2014). Recent reductions in air pollution in Europe Sea (see Section 5d). May saw anomalous high pres - sure extend between Greenland and Norway, with are reducing the relative rate of Arctic warming due relatively warm but unexceptional temperatures over to decreased downward longwave radiation, coun - tering other mechanisms that contribute to Arctic Greenland. Similar to summer 2016, neutral temperature amplification (Acosta Navarro et al. 2016). Seasonal air temperature variations in 2017 are anomalies occurred across the central Arctic in divided into winter (January–March, JFM), spring summer 2017 (Fig. 5.2c), in contrast to the warm conditions observed during much of the previous (April–June, AMJ), summer (July–September, JAS), decade. The summer 2017 conditions did not support and autumn (October–December, OND; Fig. 5.2). continued overall extreme summer sea ice loss (see These seasonal SAT divisions are chosen to coincide Section 5d). Mean coastal Greenland temperatures with the seasonal cycles of key Arctic variables. For were near climatological levels, in contrast to some example, the summer sea ice minimum occurs in summers in the recent decade. September, and autumn cooling continues through Alaska/northwestern Canada was the only region December. with above-average summer surface air temperatures. On a seasonal basis, winter was unremarkable Several locations in the interior of Alaska had the in terms of major features (Fig. 5.2a). However, there were notable short-term, regional temperature warmest calendar month of record in July. On a more anomalies in response to highly variable jet stream local and short-term basis, many stations in the north shapes. For instance, Iceland experienced a record and east of Iceland reported record high temperatures high maximum temperature of 19.1°C in February, for September. exceeding the previous February (1998) record of Summer sea level pressure was characterized by 18.1°C by a full degree (Trausti Jonsson, Icelandic Met negative anomalies in the central Arctic (Fig. 5.4). Office, 2017, personal communication). March 2017 This pattern prevented extra heat in the midlatitudes had major warmth across Siberia (Fig. 5.3) including from penetrating into the central Arctic. Such added eastern Asia. heat from outside the Arctic is associated with low Spring showed some positive temperature anoma - sea ice summers (Parkinson and Comiso 2013). This lies in the East Siberian Sea (Fig. 5.2b), a continuation sea level pressure pattern was accompanied by cloud . 5.3. Arctic Mar 2017 air temperature anomalies F ig . 5.4. Arctic mean sea level pressure field (hPa) for ig F (°C). summer 2017. | S145 AUGUST 2018 STATE OF THE CLIMATE IN 2017

166 - cover that limited the solar heating of the lower at warm temperatures over Alaska can help maintain the persistence of this North American weather pat - mosphere in the central Arctic. - tern. Contributing to the relatively warm tempera A broad swath of extreme warm temperature - tures in Alaska in autumn was the delayed freeze-up anomalies (> +4°C) stretched across the central Arc of sea ice in Alaskan waters. Freeze-up lasted well tic in autumn (Fig. 5.2d). The warmest temperature extremes, north of the Bering Strait and north of into December and set a new record for the satellite Svalbard, were due to heat stored in the upper Arctic era beginning in 1978 (see Section 5d). Ocean (see Section 5c) and to advection of warm c. Sea surface temperature— M.-L. Timmermans, C. Ladd, air from the south (generated from the Pacific and Atlantic Oceans). and K. Wood Summer sea surface temperatures (SST) in the December 2017 had extreme warm temperatures Arctic Ocean are determined mainly by absorption in Alaska and cold temperatures in the central and eastern U.S., with incidences of snow as far south of solar radiation into the surface layer. In the Barents - and Chukchi Seas, there is an additional contribu as Mississippi (Fig. 5.5a). This temperature pattern is associated with large north–south meanders of tion from advection of warm water from the North the tropospheric jet stream (Fig. 5.5b). Because the Atlantic and North Pacific Oceans, respectively. Solar extratropical mid-troposphere wind direction ap - warming of the ocean surface layer is inf luenced by - proximately follows the contour direction of geopo the distribution of sea ice (with more solar warming tential heights, Fig. 5.5b shows warm winds from the in ice-free regions), cloud cover, water color, and southwest extending into Alaska and cold air moving upper-ocean stratification. River inf luxes inf luence the latter two, as well as provide an additional source southeast from Canada in December. Warm air is less dense and supports rising geopotential heights. Thus, of warm water. SSTs are an essential indicator of the role of the ice–albedo feedback mechanism in any given melt season; as the area of sea ice cover de - creases, more incoming solar radiation is absorbed by the ocean and the warmer ocean in turn melts more sea ice. SST data presented here are from the NOAA Opti - mum Interpolation (OI) SST Version 2 product (OIS - - STv2), which is a blend of in situ and satellite measure ments (Reynolds et al. 2002, 2007). Compared to in situ temperature measurements, the OISSTv2 product showed average correlations of about 80%, with an overall cold SST bias of −0.02°C (Stroh et al. 2015). August SSTs provide the most appropriate repre - sentation of Arctic Ocean summer SSTs because they are not affected by the cooling and subsequent sea ice growth that typically takes place in the latter half of September. Mean SSTs in August 2017 in ice-free regions ranged from ~0°C in some regions to as high as 11°C in the Chukchi and Barents Seas (Fig. 5.6a). Compared to the 1982–2010 August mean (note the monthly SST record begins in December 1981), most boundary regions and marginal seas had anomalously high SSTs in August 2017 (Fig. 5.6b). Particularly high anomalies (around 3°–4°C above the 1982–2010 . 5.5. Dec 2017 fields show the cause of warm ig F average) were observed in the Beaufort, Chukchi, temperatures in Alaska and simultaneous cold and southern Barents Seas. SSTs in the boundary temperatures in the central and southern U.S. (a) regions and marginal seas, which are mostly ice free 925-hPa air temperature anomalies (°C) and (b) in August, are linked to the timing of local sea ice corresponding 500-hPa geopotential height field (m), retreat, which facilitates the direct solar heating of showing the strong wave tropospheric jet stream the exposed surface waters. pattern extending north into Alaska and south into eastern North America. | S146 AUGUST 2018

167 of the Arctic Ocean (Fig. 5.6d); the cooling trends in the Laptev and northern Barents Seas are notable exceptions. Warming trends coincide with declining trends in summer sea ice extent (including late-season - freeze-up and early melt, e.g., Parkinson 2014; see sec tion 5d), increased solar absorption (e.g., Pinker et al. 2014), release of stored ocean heat (e.g., Timmermans 2015), and milder air temperatures (see Section 5b). Mean August SSTs for the entire Chukchi Sea region exhibit a statistically significant warming trend of −1 about +0.7°C decade , based on a linear fit. d. Sea ice cover— D. Perovich, W. Meier, M. Tschudi, S. Farrell, S. Hendricks, S. Gerland, C. Haas, T. Krumpen, C. Polashenski, R. Ricker, and M. Webste r 1) s ea ice extent Arctic sea ice (1) acts as a barrier between the underlying ocean and the atmosphere, (2) limits the amount of absorbed solar energy due to its high albedo, (3) provides a habitat for biological activity, and (4) limits human access to the Arctic Ocean and adjacent seas. The extent of the Arctic sea ice cover - . 5.6. (a) Mean SST (°C) in Aug 2017. White shad ig F varies substantially over the course of a year, with the ing is the Aug 2017 mean sea ice extent (shown in end-of-winter ice cover generally two to three times as each panel) and gray contours indicate the 10°C SST large as at the end of summer. The months of March isotherm. (b) SST anomalies (°C) in Aug 2017 relative - to the Aug 1982–2010 mean (dotted black contour in and September are of particular interest because they dicates zero anomaly). Black line indicates the median are the months when the sea ice typically reaches its ice edge for Aug 1982–2010. (c) SST anomalies (°C) in maximum and minimum extents, respectively. Figure Aug 2017 relative to Aug 2012. Black line indicates the 5.7 shows the monthly average Arctic sea ice extents median ice edge for Aug 2012. (d) Linear SST trend in March 2017 and September 2017, measured by − 1 (°C yr ) for Aug of each year from 1982–2017. Trend satellite-based passive microwave instruments. is only shown for values that are significant at the 95% Sea ice extent is the total area covered by at least confidence interval; the region is gray otherwise. Black 15% concentration of sea ice. Based on data from the line indicates the median ice edge for Aug 1982–2010. (Sources: SST data are from the NOAA OISSTv2; sea ice extent and ice-edge data are from NSIDC Sea Ice Index, Version 3, Fetterer et al. 2017.) In August, regions off the west and east coasts of Greenland and in the southern Barents Sea were mark - edly cooler (by up to 3°C) than in August 2016 (see Timmermans 2017). It is notable also that compared to August 2012 (the summer of lowest minimum sea - ice extent in the satellite record, 1979–present), Au gust 2017 SSTs in the Chukchi Sea region were up to 3°C higher (Fig. 5.6c). This illustrates the significant interannual and spatial variability in summer SSTs. Cooler SSTs in August 2012 (compared to August 2017) in the Chukchi Sea were related to the persistence of . 5.7. Average monthly sea ice extent in (a) Mar ig F sea ice in that particular region (even while the main (left) and (b) Sep (right) 2017 illustrate the respective ice pack retreated) and a strong cyclonic storm in the winter maximum and summer minimum extents. The region that brought cool conditions late in the summer magenta line indicates the median ice extents in Mar season (see Timmermans et al. 2013). and Sep, respectively, for the period 1981–2010. Maps Mean August SSTs from 1982 to 2017 show statis - are from NSIDC at https://nsidc.org/data/seaice_index tically significant linear warming trends over much (Fetterer et al. 2017). | S147 AUGUST 2018 STATE OF THE CLIMATE IN 2017

168 National Snow and Ice Data Center (NSIDC) sea ice region (see Section 5c). Anomalous southerly winds index (Fetterer et al. 2017), the sea ice cover reached during October–December also played a significant 2 a maximum extent of 14.42 million km on 7 March, role by advecting warm air and ocean waters into the region through the Bering Strait (see Section 5b) and which was 8% below the 1981–2010 average. This is the lowest maximum value ever observed in the preventing southward advancement of the ice edge. satellite record. of Ge 2) a On 13 September, the sea ice extent reached a sum - ice the 2 mer minimum value of 4.64 million km . This is the The age of sea ice is another key descriptor of eighth lowest extent in the satellite record. While the the state of the sea ice cover. Compared to younger ice, older ice tends to be thicker, stronger, and more 2017 minimum extent represents a modest increase - resilient to changes in atmospheric and oceanic forc from the 2016 minimum, it was 25% less than the 1981–2010 average minimum ice extent. The 10 lowest ing (i.e., changes in atmospheric circulation patterns and ocean heat). The age of the ice is measured us September extents have occurred in the last 11 years - (Parkinson and DiGirolamo 2016). ing satellite observations and drifting buoy records In 2017, sea ice extent showed decreasing trends to track ice parcels over several years (Tschudi et al. in all months and virtually all regions, except for 2010; Maslanik et al. 2011). This method has been used to provide a record of the age of the ice since the Bering Sea during winter (Meier et al. 2014). 1985 (Tschudi et al. 2015, 2016). The September (typical Arctic sea ice minimum) - Very old ice (>4 years old) continues to be a dimin monthly average trend for the entire Arctic Ocean −1 relative to the 1981–2010 is now −13.2% decade ishingly small fraction of the Arctic ice pack in March (Fig. 5.9). The extent of the oldest ice has declined average (Fig. 5.8). Trends are smaller during March 2 −1 from 2.54 million km in March 1985 (representing , (typical Arctic sea ice maximum), at −2.7% decade 2 but the decrease is statistically significant. Both the in March 16% of the total ice pack) to 0.13 million km September and March trends are significant at the 2017 (0.9% of the total ice pack). The distribution of ice age in March 2017 was similar to that of March 99% confidence level. Freeze-up in the Chukchi Sea was extremely slow, 2016, although there was a decrease in the fractional coverage of the oldest ice, from 1.2% in March 2016 and the sea ice extent in the region at the beginning to 0.9% in March 2017. First-year ice dominates the of December 2017 was the lowest in the satellite record. It was not until the end of December that winter sea ice cover, comprising ~79% of the ice cover the region was substantially frozen over, a month later than normal. Upper ocean heat accumulated during the summer, through the absorption of solar radiation, likely slowed ice growth in the Chukchi F . 5.8. Time series of sea ice extent anomalies (%) in ig Mar (the month of maximum ice extent) and Sep (the month of minimum ice extent). Anomaly value for each - . 5.9. (a) Arctic sea ice age coverage by year, ex ig F year is the percent difference in ice extent relative to pressed as the fraction of the total ice area, 1985–2017. the 1981–2010 mean. The black and red dashed lines Sea ice age coverage maps for (b) Mar 1985 and (c) are least squares linear regression lines. Mar 2017. | S148 AUGUST 2018

169 in March 2017, compared to ~55% in the 1980s. The thinner, younger ice is more mobile and susceptible to mechanical wind forcing, and it is vulnerable to complete melting in the summer and contributes to the observed decrease in summer sea ice extents by enabling more heat to be absorbed by the upper ocean. depth ice thickness and ea 3) s snow - Satellite remote sensing and regular airborne sur vey programs continued to record changes in Arctic sea ice thickness and volume. These survey programs derive ice thickness and volume by observing the free - board of the ice cover, which is the distance between the surface of the ocean and the top of the ice. During CryoSat-2 this past year the ESA - radar altimeter mis - sion completed its seventh year of operation, provid ing sea ice thickness estimates between October and April (Laxon et al. 2013). The CryoSat-2 freeboard measurements expand the data record of satellite and submarine-based observations that document the decline in sea ice thickness since 1958 (Kwok and Rothrock 2009; Lindsay and Schweiger 2015). CryoSat-2 products from the Al - In spring 2017, fred Wegener Institute indicated a spatially variable pattern of ice thickness (Fig. 5.10a), which is typical. The April 2017 thickness anomaly, compared to the period 2011–16 (Fig. 5.10b), shows below-average thicknesses in the multiyear ice region north of the Queen Elizabeth Islands of the Canadian Arctic Ar - chipelago, the Chukchi Sea, and the shelf regions of the East Siberian Sea. Above-average thicknesses were observed in the Beaufort Sea and the eastern part of the central Arctic Ocean. Sea ice volume estimates were generated from Cryosat-2 observations for 2011–17 for the months of October through April. Results for the central Arctic Ocean show a decline from 2011 to 2013, an increase in 2014, followed by a steady decline from 2014 to 2017. The April 2017 sea ice volume (13.19 ± 1.15 × 3 3 km ) ranks as the third lowest spring volume after 10 3 3 April 2012 (13.14 ± 1.27 × 10 km ) and 2013 (12.56 ± 3 3 km ) in the CryoSat-2 data record (2011–17). 1.21 × 10 The difference between the three lowest volume es - timates lies within the observational uncertainties of the instrument. For more information regarding instrument uncertainty see Ricker et al. (2014). F ig . 5.10. Apr 2017 (a) sea ice thickness (m) derived from CryoSat-2 radar altimeter data and (b) sea ice thickness anomaly (m; base period 2011–16). (c) Snow depth (m) on Arctic sea ice at the end of winter, prior to melt onset; recent in situ measurements (stars), made in 2015 and 2017, and airborne observa - tions (multiple airborne survey lines), made in Mar and May in 2009–12 and 2014–15, are overlaid on the long term mean snow depth for the months of Mar and Apr (adapted from Warren et al. 1999). Black line and arrows in (c) designate the western Arctic. | S149 AUGUST 2018 STATE OF THE CLIMATE IN 2017

170 SIDEBAR 5.1: PALEOCLIMATE RECORDS: PROVIDING CONTEXT AND UNDERSTANDING OF CURRENT ARCTIC CHANGE — E. OSBORNE, T. CRONIN, AND J. FARMER rafted debris and sea ice-dependent diatoms in Arctic marine At present, the Arctic Ocean is experiencing changes - sediments indicate that the first Arctic sea ice formed approxi in ocean surface temperature and sea ice extent that are mately 47 million years ago (St. John 2008; Stickley et al. 2009; unprecedented in the era of satellite observations, which Fig. SB5.1), coincident with an interval of declining atmospheric extend from the 1980s to the present (see Sections 5c,d). ) concentration, global climate cooling, carbon dioxide (CO To provide context for current changes, scientists turn 2 and expansion of Earth’s cryosphere during the middle Eocene. - to paleoclimate records to document and study anthro The development of year-round (i.e., perennial) sea ice in the pogenic influence and natural decadal and multidecadal central Arctic Ocean, similar to conditions that exist today, is climate variability in the Arctic system. Paleoceanographic evident in sediment records as early as 14–18 million years ago records extend limited Arctic instrumental measurements (Darby 2008). These records suggest that transitions in sea ice back in time and are central to improving our understand - cover occur over many millennia and often vary in concert with ing of climate dynamics and the predictive capability of the waxing and waning of circum-Arctic land ice sheets, ice climate models. By comparing paleoceanographic records shelves, and long-term fluctuations in ocean and atmosphere with modern observations, scientists can place the rates concentrations (Stein et temperatures and atmospheric CO and magnitudes of modern Arctic change in the context 2 al. 2012; Jakobsson et al. 2014). of those inferred from the geological record. Over shorter time scales, shallow sediment records from Over geological time, paleoceanographic reconstruc - Arctic Ocean continental shelves allow more detailed, higher- tions using, for instance, marine sediment cores indicate resolution (hundreds of years resolution) reconstructions that the Arctic has experienced huge sea ice fluctuations. 7 00 of sea ice history extending through the Holocene (11 These fluctuations range from nearly completely ice-free years ago to present), the most recent interglacial period. to totally ice-covered conditions. The appearance of ice- . SB5.1. The oldest known paleoclimate evidence of sea ice in the Arctic are (a) fossilized remains of sea ice ig F dwelling diatoms ( spp.) and (b) ice rafted debris that date back to 47 million years ago (Stickley et Synedropsis al. 2009). (c) Global compilation of paleoclimate records indicates that cooling ocean temperatures (°C) and declining atmospheric CO (ppm) coincide with major NH sea ice development (data: Beerling and Royer 2011; 2 Zhang et al. 2013; Anagnostou et al. 2016). Global ocean temperature anomalies are determined by millions of stable oxygen isotopic measurements of fossilized calcite benthic foraminifera shells. Arrows indicate cooling temperature and declining CO concentrations through the greenhouse to icehouse transition. Red and orange 2 “+” on the right y-axis indicate the CMIP5 multimodel mean projected temperature and atmospheric CO , 2 respectively, in the year 2050 and 2100. | S150 AUGUST 2018

171 A notable feature of these records is an early Holocene sea ice minimum, corresponding to a thermal maximum (warm) period from 11 0 00 to 5000 years ago, when the Arctic may have been warmer and had less summertime sea ice than today (Kaufman et al. 2004). However, it is not clear that the Arctic was ice-free at any point during the Holocene (Polyak et al. 2010). High-resolution paleo–sea ice records from the western Arctic in the Chukchi and East Siberian Seas indicate that sea ice concentrations increased through the Holocene in concert with decreasing summer solar insolation (sunlight). Sea ice extent in this region also varied in response to the volume of Pacific water delivered via the Bering Strait into the Arctic Basin (Stein et al. 2017; Polyak et al. 2016). Records from the Fram Strait (Müller et al. 2012), Laptev Sea (Hörner et al. 2016), and Canadian Arctic Archipelago (Vare et al. 2009) also indicate a similar long-term expansion of sea ice and suggest sea ice extent in these regions is modulated by the varying influx of warm Atlantic water into the Arctic Basin (e.g., Werner et al. 2013). Taken together, available records support a circum- Arctic sea ice expansion during the late Holocene. A notably high-resolution summer sea ice history (<5-year resolution) has been established for the last 1450 years using a network of terrestrial records (tree ring , lake sediment, and ice core records) located around the margins of the Arctic . SB5.2. (a) Atmospheric CO F concentrations (ppm), ig 2 Ocean (Kinnard et al. 2011). Results summarized in Fig. SB5.2 (b) paleoclimate reconstructions of summer Arctic sea - indicate a pronounced decline in summer sea ice extent be 2 ice extent (km ; Kinnard et al. 2011), and (c) annual ginning in the 20th century, with exceptionally low ice extent atmospheric temperature anomalies (°C; McKay and recorded since the mid-1990s, consistent with the satellite Kaufman 2014) and sea surface temperature anoma - lies (°C; Spielhagen et al. 2011) spanning the last 1500 record (see Section 5d). While several episodes of reduced years. Atmospheric (red solid line: 5-yr mean and light and expanded sea ice extent occur in association with climate gray: annual mean) and upper-ocean (dark gray with anomalies such as the Medieval Climate Warm Period (AD circles indicating individual data points) temperature - 800–1300) and the Little Ice Age (AD 1450–1850), the magni anomalies are plotted together to show respective tude and pace of the modern decline in sea ice is outside of the temperature trends. Vertical dashed line indicates the range of natural variability and unprecedented in the 1450-year onset of the Industrial Revolution. Atmospheric CO 2 - reconstruction (Kinnard et al. 2011). A radiocarbon-dated drift concentrations [shown in (a)] are from the Law Dome - ice core record (Etheridge et al. 1996, 1998) and mod wood record of the Ellesmere ice shelf in the Canadian High ern observations from the Mauna Loa observatory [Dr. Arctic, the oldest landfast ice in the Northern Hemisphere, Pieter Tans, NOAA/ESRL (www.esrl.noaa.gov/gmd also demonstrates a substantial reduction in ice extents over - /ccgg/trends/), and Dr. Ralph Keeling, Scripps Institu the 20th century (England et al. 2017). A supporting sediment tion of Oceanography (www.scrippsco2.ucsd.edu/)]. record indicates that inflowing Atlantic water in Fram Strait has warmed by 2°C since 1900, driving break up and melt of et al. 2012), further driving sea ice melt and warming seas. sea ice (Spielhagen et al. 2011). Complementary mooring and Similar high-resolution proxy records from Arctic regions satellite observations show the “Atlantification” of the eastern also indicate that the modern rate of increasing annual surface Arctic due to enhanced inflow of warm saline water through air temperatures has not been observed over at least the last Fram Strait (Nilsen et al. 2016) and nutrient-rich Pacific water 2000 years (McKay and Kaufman 2014). Scientists conclude via the Bering has increased by more than 50% (Woodgate that broad-scale sea ice variations recorded in the paleo | S151 AUGUST 2018 STATE OF THE CLIMATE IN 2017

172 PALEOCLIMATE RECORDS: PROVIDING CONTEXT CONT. SIDEBAR 5.1: AND UNDERSTANDING OF CURRENT ARCTIC CHANGE — E. OSBORNE, T. CRONIN, AND J. FARMER record were predominantly driven by changes in basin- studies and observational time series attribute the decline scale atmospheric circulation patterns, fluctuations in air in sea ice extent and thickness over the last decade to temperature and strength of incoming solar radiation, both enhanced greenhouse warming and natural climate and changes in the inflow of warm water via Pacific and variability. While understanding the interplay of these fac - Atlantic inflows (Polyak et al. 2010). tors is critical for future projections of Arctic sea ice and - There is general consensus that ice-free Arctic sum ecosystems, most observational time series records cover mers are likely before the end of the 21st century (e.g., only a few decades. This highlights the need for additional Stroeve et al. 2007; Massonnet et al. 2012), while some paleoceanographic reconstructions across multiple spatial climate model projections suggest ice-free Arctic summers and temporal domains to better understand the drivers and as early as 2030 (Wang and Overland 2009). Paleoclimate implications of present and future Arctic Ocean change. cycles at regional and global scales. The high albedo Snow plays several critical roles in the growth and of the ice sheet contributes to a modulation of the melt of Arctic sea ice. These roles include insulating amount of solar energy absorbed by Earth, and the the ocean from the atmosphere, dampening heat location and topography of the ice sheet affects atmo f luxes, reducing sea ice growth, ref lecting more than - 80% of the incoming sunlight, delaying ice melt, and - spheric circulation. The GrIS is also a major contribu tor to current and projected sea level rise, through contributing to melt pond formation (Granskog et surface runoff and iceberg calving. The summer of al. 2017). Prior to the 1990s, observations of snow on 2017 over the Greenland ice sheet was characterized Arctic sea ice were limited to in situ measurements. - Warren et al. (1999) compiled many of these obser by below-average (1981–2010) melt extent and above- average surface albedo, with the net ablation being vations into a long-term record. New approaches to measure snow depth have since emerged, including below the 2008–17 average at many test sites but still improved instruments for in situ and autonomous above the average for the 1961–90 reference period observations and remote sensing. Field observations when the ice sheet was in steady equilibrium. Overall, total mass loss in 2017 was close to the average of the from recent years underscore significant regional years 2003–16. and interannual variability in snow on Arctic sea ice. Figure 5.10c shows the historical snow depth record, - G 1) s meltin plus a compilation of airborne snow depth measure urface ments collected between March and May in 2009–12 Estimates of melt extent across the GrIS are and 2014–15, and in situ measurements made in 2015 obtained from brightness temperatures measured and 2017. The recent mean snow depths range from by the Special Sensor Microwave Imager/Sounder 0.05 to 0.55 m. Compared to the record published by (SSMIS) passive microwave radiometer (e.g., Mote Warren et al. (1999) there has been an overall decrease 2007; Tedesco et al. 2013). These estimates point to in snow depths of 37% ± 29% in most of the western a rapid start of the melting season in 2017, similar to 2016, with melt extent in early April reaching an area Arctic (aka North American Arctic) at the end of once typical of early June (Fig. 5.11a). From mid-June winter (Fig. 5.10c). The trend in the mean anomalies is −1 through mid-July 2017, however, melt extent was −0.27 cm yr with 99% significance. This decrease is potentially associated with later sea ice formation and persistently below the 1981–2010 average. The spatial thus later onset of snow accumulation in the previous extent of melt for summer 2017 (June–August, JJA) autumn (Webster et al. 2014; Kurtz and Farrell 2011; was above average on 15 of 93 days (16%) and reached Blanchard-Wrigglesworth et al. 2015). its maximum extent of 32.9% of the ice sheet area on 26 July. The maximum extent of surface melt in M. Tedesco, J. E. Box, J. Cappelen, 2017 was lower than the average maximum extent e. Greenland ice sheet— of 39.8% for the period 1981–2010 and was the low - R. S. Fausto, X. Fettweis, K. Hansen, M. S. Khan, S. Luthcke, T. Mote, I. Sasgen, C. J. P. P. Smeets, D. van As, est maximum extent since 1996. There was regional variability in the characteristics of the summer melt. R. S. W. van de Wal, and I. Velicogna The Greenland ice sheet (GrIS) plays a crucial role Most of the western and northeast ice sheet margins in the climatological, hydrological, and ecological had more days than average with melt (relative to | S152 AUGUST 2018

173 balance 2) s urface mass 1981–2010), while the southeast margin had fewer days than average. The magnitude and evolution of - Consistent with the low-to-moderate surface melt ing described above, the August 2016–August 2017 surface melt in 2017 were consistent with the state of the dominant atmospheric circulation pattern, as surface mass balance (SMB) year along the K-transect at 67°N in west Greenland (Fig. 5.11b; van de Wal et defined by the Arctic Oscillation and North Atlantic Oscillation, both of which were strongly positive al. 2012) was characterized by moderate mass loss (Tedesco et al. 2017). over the ablation region (Tedesco et al. 2017). The - SMB along the transect line, which has been continu ously monitored for 28 years, was approximately one standard deviation below the 1990–2017 mean. The equilibrium line altitude (defined as the elevation at which mass losses balance mass gain, i.e., SMB = 0) in 2017 was around 1490 meters, which is 40 m below the 28-year mean. The mass balance gradient was 3.4 −1 −1 yr mm w.e. (water equivalent) m , which is about 6% lower than the average (Tedesco et al. 2017). Due to the relatively low summer temperatures, net ice ablation averaged over the PROMICE sites (Fig. 5.11b), distributed around Greenland in the - ablation zone, was about 20% (or 0.6 standard devia tions) lower in 2017 than compared to the 2008–17 average. The largest ablation anomaly values, more than one standard deviation below average, occurred at the southwest and northwest margins. The largest absolute ablation of 5.5 m of ice was measured at the southern tip of the ice sheet. More details can be found in Tedesco et al. (2017). While the surface mass balance observations indicate that surface melt was relatively moderate in 2017, compared to that observed in the last decade, it was still higher than observed prior to 1990. When referencing the values to the 1961–90 climatological standard period (Van As et al. 2016), all eight low-elevation PROMICE station sites experienced above-average ablation anomalies in 2017 (Fig. 5.11b). However, only three stations were beyond the estimated uncertainty: KPC_L (+96% ± 49%), SCO_L (+15% ± 14%) and KAN_L (+48% ± 35%). 3) a lbedo The area-averaged albedo (the fraction of incident ig F . 5.11. (a) Spatial extent of melt, derived from the solar radiation ref lected by a surface) for the entire satellite product, as a percentage of the ice sheet area Greenland ice sheet for summer 2017 was 80.9%, as during 2017 (red line) and the 1981–2010 mean spatial - determined using data from the Moderate Resolu extent of melt (dashed blue line). Light and dark gray tion Imaging Spectroradiometer (MODIS; after areas represent the interdecile and interquartile Box et al. 2017; Fig. 5.11c). This is the third highest ranges, respectively. (b) 2017 ablation anomalies (% of average, relative to 1961–90) at lower PROMICE summer albedo value, after those of 2000 and 2013, (Programme for monitoring of the Greenland ice sheet during the 2000–17 MODIS period. Positive albedo weather station sites in the Greenland ice sheet) abla - anomalies are consistent with reduced melting in 2017 tion area, using historical coastal temperature records. and snowfall events during the summer. The highest (c) Distribution of albedo anomalies (%, 2000–09 refer - 2017 summer albedo anomalies occurred along the ence period) for summer 2017, derived from MODIS . western margins of the ice sheet (Tedesco et al. 2017). Area within the rectangle in (c) indicates the location of the K-transect. | S153 AUGUST 2018 STATE OF THE CLIMATE IN 2017

174 4) t balance otal mass - GRACE satellite gravity estimates obtained fol lowing Velicogna et al. (2014), Sasgen et al. (2012), and Luthcke et al. (2013) and available since 2002, indicate that between April 2016 and April 2017 (the most recent 12-month period of reliable data) there was a net ice mass loss of 276 ± 47 Gt (gigatonnes; Fig. 5.12). This is 144% greater than the April 2015–April 2016 mass loss (191 ± 28 Gt) and close to the average April-to-April mass loss (255 ± 7 Gt) for 2003–17 (Sas - gen et al. 2012). The updated trends of total ice mass −1 loss for the 15-year GRACE period are 264 Gt yr − 1 (Velicogna et al. 2014) and 270 Gt yr (Sasgen et al. 2012); the different values ref lect the slightly different 2 . 5.13. Glacier area change (km ig ) from LANDSAT F computational approaches adopted in the two stud - and ASTER imagery available since 1999 for 45 of the ies. The GRACE mission came to an expected end in widest and fastest-flowing marine-terminating glaciers October 2017. No further data will be available from (after Box and Hansen 2015). this important source. It is anticipated that gravity Among the surveyed glaciers, 22 retreated, 10 were measurements from space will resume and ice mass estimates will be possible again when the GRACE stable, and 13 advanced. Overall, the annual net area follow-on mission is launched. At the time of writing, change of the 45 glaciers at the end of the 2017 melt the expected launch window is in spring 2018. season, which started in June and ended in September, 2 . This is ~80% of the 18-year survey was −102.8 km 2 −1 year terminatin period average of −126.6 km arine - . The largest area 5) m G Glaciers Marine-terminating glaciers are the outlets by losses were in eastern Greenland, where the Helheim and Kangerdlugssauq glaciers lost, respectively, 11.6 which the Greenland ice sheet discharges ice mass 2 2 to the ocean. Glacier area measurements from in area. The largest advance was and 9.9 km km LANDSAT and ASTER, available since 1999 (Box observed at Petermann glacier, northwest Greenland, 2 . where the area increased by 11.5 km and Hansen 2015) for 45 of the widest and fastest- f lowing marine-terminating glaciers, reveal a pattern urface air 6) s temperatures of continued relative stability since 2012/13 (Fig. 5.13). Measurements at 20 weather stations of the Danish Meteorological Institute (Cappelen et al. 2018; Table 5.1) indicate widespread above or near-average air temperatures for 2017, relative to the period 1981– 2010. The exception was during spring 2017 (March– May, MAM) in coastal northeast Greenland and the start of July in western Greenland, when many sites experienced relatively cool temperatures. Looking in more detail, during winter 2016/17 (December– February, DJF) a new seasonal record high was set in Aputiteeq, located in eastern Greenland. February in Aputiteeq was particularly warm, with a new monthly record set. At Kap Morris Jesup, along the northern coast, the winter season was the second warmest (only exceeded in 2011), with December 2016 . 5.12. Change in the total mass (Gt) of the Green - ig F matching the record warmth of December 2009. April land ice sheet between Apr 2002 and Jun 2017, es - 2017 was generally colder than average at many sites, timated from GRACE measurements. (Due to the compared to April 2016 when record breaking high decommissioning of the GRACE satellite, no data temperatures were recorded. In autumn (September– are available after Jun 2017.) Data are based on an November, SON) some individual months were unweighted average of JPL RL05, GFZ RL05, and CSR record setting at Danmarkshavn, Daneborg, and RL05 solutions, which reduce noise in the GRACE data for 2017 (Sasgen et al. 2012). Ittoqqortoormiit. At Danmarkshavn, Daneborg, and | S154 AUGUST 2018

175 t le AB 5.1. Seasonal and annual surface air temperature anomalies (°C) relative to the 1981–2010 average at 20 Danish Meteorological Institute weather stations in Greenland, where observations have been made for a minimum of 30 years. Seasons are winter (DJF 2016/17); spring (MAM 2017); summer (JJA 2017); autumn (SON 2017). Highlighted cell indicates a new seasonal record. The year that observations began is given, together with the station name, geographic coordinates, and elevation. Jan–Dec DJF SON MAM JJA Station Name, Start Year; 2017 2017 2017 2017 2016/17 Latitude, Longitude, Elevation Pituffik/Thule AFB 1.1 0.5 0.2 0.2 1.4 Anomaly (°C) 1948; 1957 M a x Ye a r 2010 1986 1953 2010 76.5°N, 68.8°W, M in Ye a r 1992 1949 1992 1996 1964 77 m a.s.l. Upernavik 0.0 Anomaly (°C) 1.2 0.7 1.4 0.7 1873; 1932 M a x Ye a r 2010 1947 2010 2012 72 .8°N, 56.1°W, M in Ye a r 1887 1983 1896 1873 1917 126 m a.s.l. Aasiaat 0.8 0.3 0.6 Anomaly (°C) 0.9 0.8 1958; 2010 2010 2016 2012 2010 M a x Ye a r 68.7°N, 52.8°W, 1984 1993 1972 1986 1983 M in Ye a r 43 m a.s.l. Ilulissat 0.4 Anomaly (°C) 0.1 0.1 − 0.5 0.5 1807; 1929 M a x Ye a r 2010 2010 1847 1960 69.2°N, 51.1°W, M in Ye a r 1863 1863 1813 1863 1837 29 m a.s.l. Kangerlussuaq 0.3 0.6 − 0.7 − 0.4 Anomaly (°C) 0.7 1949; M a x Ye a r 2010 1986 2016 1960 2010 67°N, 50.7°W, M in Ye a r 1984 1983 1993 1983 1982 50 m a.s.l. Sisimiut 0.6 1.2 Anomaly (°C) 1.2 1.2 0.4 1961; 2010 M a x Ye a r 2010 2010 2010 2012 70°N, 53.7°W, M in Ye a r 1984 1983 1972 1982 1984 10 m a.s.l. Nuuk Anomaly (°C) 0.6 0.6 0.1 0.2 0.6 178 4; 1932 M a x Ye a r 2010 2010 2010 2012 6 4.2°N , 51.7 ° W, M in Ye a r 1818 1818 1802 1819 1811 80 m a.s.l. Paamiut 0.9 0.0 0.2 Anomaly (°C) 1.0 1.3 − 1958; M a x Ye a r 2010 2010 2005 2010 2010 62°N, 49.7°W, M in Ye a r 1984 1984 1993 1969 1982 36 m a.s.l. Ivittuut/Narsarsuaq Anomaly (°C) 1.4 1.4 1.3 0.2 0.9 1873; 2010 2016 M a x Ye a r 2010 2010 2010 61.2°N, 45.4°W, 1884 1984 1989 1873 1874 M in Ye a r 27 m a.s.l. Qaqortoq 0.1 0.7 − Anomaly (°C) 0.7 1.0 0.3 1807; M a x Ye a r 2010 2010 1932 1929 2010 60.7°N, 46.1°W, 1811 M in Ye a r 1884 1863 1811 1874 32 m a.s.l. Kap Morris Jesup Anomaly (°C) 1.5 0.8 0.4 − 0.4 5.2 1980; 2 011 2016 1995 M a x Ye a r 2014 2 011 83.7°N, 33.4°W, 1985 M in Ye a r 1985 198 8 1997 1990 4 m a.s.l. | S155 AUGUST 2018 STATE OF THE CLIMATE IN 2017

176 t AB le 5.1. ( cont .) Station Name, Start Year; SON JJA MAM Jan–Dec DJF 2017 Latitude, Longitude, Elevation 2016/17 2017 2017 2017 Station Nord 0.4 1.8 2.7 1.0 Anomaly (°C) − 2.2 1961; 2006 2016 2003 M a x Ye a r 2016 2 011 81.6°N, 16.7°W, 1967 1989 1970 M in Ye a r 1968 1961 36 m a.s.l. Danmarkshavn 1.1 Anomaly (°C) 0.6 4.4 1.0 2.1 − 1949; 2005 2016 M a x Ye a r 2016 2016 1976 76.8°N, 18.7°W, 1971 1983 M in Ye a r 1967 1966 1955 1 m a.s.l. Daneborg 3.1 − 0.3 − 4.8 0.5 Anomaly (°C) 0.1 1958; 2005 1996 2016 2016 2016 M a x Ye a r 74.3°N, 20.°W 2, 1975 1985 M in Ye a r 1968 1971 1961 44 m a.s.l. . Ittoqqortoormiit 0.2 Anomaly (°C) 1.0 2.5 − 0.9 3.6 1949; 2016 2016 1996 2014 2016 M a x Ye a r 70. 5°N, 22°W, 1951 1956 M in Ye a r 1966 1955 1951 70 m a.s.l. Aputiteeq 4.4 2.2 Anomaly (°C) 1.6 1.4 − 0.2 1958; 2016 M a x Ye a r 1974 2017 2016 2016 67. 8 ° N , 32 . 3° W, M in Ye a r 1973 1969 1969 1967 1973 13 m a.s.l. Ta s i i l a q 1.6 0.2 1.3 2.3 1.2 Anomaly (°C) 1895; 1941 M a x Ye a r 2016 1929 1929 2016 65.6°N, 37.6°W, 1983 1899 1918 1899 M in Ye a r 1917 53 m a.s.l. Ikermiuarsuk 1.1 0.1 − — — — Anomaly (°C) 1958; 2010 M a x Ye a r 2003 2 011 1999 1961 61.9°N, 42°W, 1983 1976 1967 1983 1969 M in Ye a r 39 m a.s.l. Prins Chr. Sund 1.3 0.2 − 0.2 0.6 0.5 Anomaly (°C) 1958; 2010 M a x Ye a r 2010 2010 2010 2005 60.1°N, 42.2°W, 1970 M in Ye a r 1993 1993 1989 1982 88 m a.s.l. Summit −0.6 Anomaly (°C) 0.6 1.4 0.6 2.7 1991; 2012 2010 2016 2002 2010 M a x Ye a r 72 .6°N, 38. 5°W, 2009 M in Ye a r 1992 1993 1992 1992 3202 m a.s.l. Aputiteeq the autumn season was second warmest, f. Glaciers and ice caps outside Greenland— M. Sharp, B. Wouters, G. Wolken, L. M. Andreassen, D. Burgess, L. Copland, exceeded only by 2016. J. Kohler, S. O’Neel, M. S. Pelto, L. Thomson, and T. Thorsteinsson At Summit, the highest elevation of the GrIS, - winter 2016/17 was the fourth warmest, with Febru The Arctic is the world’s third most heavily glaci - ated region, after Antarctica and Greenland. Though ary 2017 second warmest after February 2005. May the total mass of glaciers and ice caps in the region was the second warmest since 1991, after May 2010. A is significantly less than that of the Antarctic and new July record-breaking low temperature of −33.0°C Greenland ice sheets, ice loss from Arctic glaciers was measured at Summit on 4 July. On 28 July, a new record high July temperature of 1.9°C was measured and ice caps has become a significant contributor to - current global sea level rise as a result of recent sum at Summit. mer warming (Gardner et al. 2011, 2013; Jacob et al. 2012; Millan et al. 2017). | S156 AUGUST 2018

177 The state of glaciers, ice caps, and ice sheets is often described by their mass balance. The annual ) is defined as climatic mass balance of a glacier (B clim the difference between the annual snow accumulation on the glacier and the annual mass loss by surface melting and runoff. For the purposes of calculation, a “mass balance year” is usually taken as the period between the ends of successive summer melt seasons. Variations in the mass of most monitored Arctic gla - ciers and ice caps are controlled largely by changes in their climatic mass balance. However, those glaciers that terminate in the ocean [e.g., Devon Ice Cap NW (Arctic Canada), and Hansbreen and Kongsvegen (Svalbard); Table 5.2; Fig. 5.14] or in lakes can also lose mass by melting below the waterline. However, this mass balance term is rarely routinely measured. Here, B measurements made in 2015–16 and clim 2016–17 at individual glaciers monitored across the . 5.14. Locations of the 27 sites on 25 Arctic glaciers F ig 1 ). All Arctic region are reported (Table 5.2; Fig. 5.14 - and ice caps (black circles) that have long-term obser data are from the World Glacier Monitoring B clim vations of annual climatic mass balance (B ). Areas clim outlined in yellow are the Randolph Glacier Inventory Service (WGMS 2018). Positive (negative) annual B clim (RGI) regions of the Arctic (Pfeffer et al. 2014). Some values indicate that a glacier gained (lost) mass over individual glaciers are too close for identification and the course of the mass balance year that includes a can be identified by the numbers shown at the edge of winter accumulation season, when snow deposition the RGI region. They can also be referenced in Table typically exceeds meltwater runoff (positive mass 5.2. Red shading indicates glaciers and ice caps, includ - balance), followed by a summer ablation season, ing ice caps in Greenland outside the ice sheet. when the opposite is the case (negative mass bal - ance). The timing and duration of the accumulation with the tendency for predominantly negative mass and ablation seasons vary from region to region and balance anomalies over the past decade. However, since - the long-term tendency of the cumulative B from year to year, but in most cases, net accumula clim the mid-1990s continues to be toward more negative tion occurs from late autumn to late spring, and net cumulative balances in all five regions (Fig. 5.15), ablation from late spring to late autumn. At the time of writing, estimates for the 2016–17 mass balance indicating continuing mass loss. With the exception of Svalbard (where there has been no obvious recent year were available for only 16 glaciers [two in Alaska, nine in Iceland (nine measurement locations at seven acceleration of mass loss rates; Fig. 5.15), rapid mass glaciers), three in Svalbard, and two in Norway] of loss across the five regions typically began during the 27 that are regularly monitored (Fig. 5.14). So the 1990s. New data on the length of the summer melt season that a complete cycle of results can be reported, B clim measurements for the 2015–16 mass balance year are at Wolverine and Gulkana glaciers in Alaska (Fig. also reported. 5.16) show that since measurements began in 1966 Relative to the long-term (1985–2015) mean B the summer melt season has increased by about 18 clim days (14%) at Wolverine Glacier, located in a maritime values, 20 of the 24 values reported for 2015–16 were climate, and 24 days (24%) at Gulkana Glacier, located more negative than the mean, and four were more positive. Five of the 18 annual net balances reported in a more continental climate. These data suggest that for 2016–17 were more negative than the 1985–2015 increases in summer melt played a significant role in generating more negative annual mass balances - mean, and 13 were more positive. The mix of posi tive and negative anomalies in 2016–17 contrasts in this region. B measurements for the 2015–16 mass balance clim 1 Table 5.2 lists 25 glaciers and ice caps by name while Fig. year are from 24 glaciers: three in Alaska, four in is measured. 5.14 shows the location of 27 sites where B clim Arctic Canada, nine in Iceland, four in Svalbard, two The difference in numbers is accounted for by Hofsjökull, in northern Norway, and two in northern Sweden is measured at three different sites on a Iceland, where B clim (Table 5.2). All these glaciers had a negative annual single ice cap (no. 9 in Table 5.2). B in 2015–16. At Austre Broggerbreen and Midtre clim | S157 AUGUST 2018 STATE OF THE CLIMATE IN 2017

178 mass balance of 25 glaciers in Alaska (3), Arctic Canada (4), Iceland (7), t 5.2. Measured B AB climatic le clim Svalbard (4), and Northern Scandinavia (7) for 2015/16 and 2016/17, together with the 1985–2015 mean and standard deviation for each glacier [(Hofsjökull (Iceland) is treated as a single glacier, although measure - ments are made in three different sectors of this ice cap)]. (* Indicates one or more years of data missing from the record). Negative (positive) values for B indicate mass loss (gain). Data are from the World clim Glacier Monitoring Service (WGMS 2018), with updates for Alaska from S. O’Neel and M. Pelto, White Glacier from L. Thomson, Svalbard from J. Kohler, and mainland Norway (Engabreen and Langfjordjokulen) from L. M. Andreassen. Numbers in column 1 refer to the glaciers located in Fig. 5.14. Results for 2016/17 may be based on measurements made before the end of the melt season and may be subject to revision. B B Std. dev. Mean B B clim clim clim clim Glacier – 2 – 1 1 – 2 – 2 – 2 – 1 – – 1 Region (kg m (kg m ) ) yr yr (kg m yr (kg m ) ) yr (record length, years) 2016 –17 2015 –16 (1985–2015) 1985–2015 Alaska 400 – 1 Wolverine (52) – 603 1016 – 116 0 – 3 Lemon Creek (65) – 640 798 – 120 0 1480 140 0 — 2 Gulkana (52) – 778 721 – Arctic Canada Devon Ice Cap (NW) – – 483 — 205 204 7 (56) — 775 397 26 – Meighen Ice Cap (55) 5 – Melville South Ice Cap – 418 477 – 792 — 4 (52) White (54) – 308 316 – 268 — 6 Iceland 8 Langjökull S. Dome (19) – 128 8* 855 – 1677 — 9 Hofsjökull E (25) – 545* 871 – 112 0 – 650 490 9 Hofsjökull N (26) – 565* 754 – 830 – 80 450 – 1017 802* – Hofsjökull SW (25) 9 10 – 475* 738 – 642 — Köldukvislarjökull (22) 11 Tungnaarjökull (24) – 112 8 * 830 196 — 806 12 Dyngjujökull (18) 146* – M — 13 – 258* 683 – 342 — Brúarjökull (23) 14 Eyjabakkajökull (24) – 709* 839 – 930 — Svalbard – 420 379 330 – 120 0 – 17 Midre Lovenbreen (49) 530 – 1450 – 363 486 – Austre Broggerbreen (50) 16 – Kongsvegen (31) 114 * 15 360 – 320 40 1078 Hansbreen (28) – — 397* 474 – 18 was the most negative in northern Scandinavia) had a more positive annual Lovenbreen in Svalbard, B clim B ever recorded. This is attributed to relatively low snow in 2016–17 than in the previous year, while two clim (both in Alaska) had a more negative annual B accumulation in winter 2015–16 and high summer clim than in the previous year. In Svalbard, the positive melt in 2016, especially in the record warm and rainy mass balance on Kongsvegen in 2016–17 is linked month of July. Of the 18 glaciers with measurements for both 2015–16 and 2016–17, 16 (two in Arctic to above-average winter snowfall, which delayed the Canada, all nine in Iceland, three in Svalbard, and two onset of ice melt in summer 2017. | S158 AUGUST 2018

179 .) t AB le cont 5.2. ( B Std. dev. Mean B B B clim clim clim clim Glacier – – – 2 – 1 2 1 2 – 1 – – 2 – 1 (kg m (kg m yr yr ) Region ) (kg m (kg m yr ) yr ) (record length, years) 2016 –17 2015 –16 1985–2015 (1985–2015) Northern Scandinavia 127 1024 Engabreen (48) 19 230 1250 – – – – 270 948* Langfjordjokulen (27) 737 – – 1660 20 550 — 370 460* 21 Marmaglaciaren (24) – – 465* — — – Rabots Glaciar (31) 22 659 592* 785 — — 23 Riukojetna (26) – Storglaciaren (71) – 760 – 240 — 24 153 – 25 198* 1118 — — Tarfalaglaciaren (19) et al. 2015). This approach provides regional mass Although some of the 2016–17 mass balance mea - surements are provisional, 12 of the reporting glaciers change estimates for Iceland, Svalbard, the Russian (two in Alaska, one in Arctic Canada, six in Iceland, Arctic, and the Canadian Arctic (Fig. 5.17). Cryo - Sat-2 estimates for the period 2011–17 identify the two in Svalbard, and one in northern Scandinavia) had negative annual balances, and six (Meighen Canadian Arctic as the most important of these four Ice Cap, Canada; Hofsjokull SW, Brúarjökull, and regional sources of glacier mass loss (7-year mean: −1 −1 ), followed by Svalbard (−18.95 Gt yr ), Dyngjujökull, Iceland; Kongsvegen, Svalbard; and −60.19 Gt yr −1 the Russian Arctic (−13.46 Gt yr Engabreen, Norway) had positive balances (Table 5.2). ), and Iceland (−2.36 −1 Gt yr ). Estimates for Alaska and northern Scandi - Estimates of regional scale ice mass changes since radar altimetry, 2011 can be derived from navia are not available. CryoSat-2 which measures glacier surface elevation (Wouters in . 5.15. Cumulative climatic mass balance (B ig F clim − 2 kg m ) for glaciers and ice caps in five regions of the Arctic, and for all monitored glaciers and ice caps (Pan- ig F . 5.16. Length (days) of the annual ablation season Arctic). Average annual climatic balances for each at Gulkana (red) and Wolverine (blue) glaciers, Alaska, region are calculated for each year using the measured − 1 showing the mean rate of change (days yr ) over the annual mass balances for all monitored glaciers in the 1966–2017 observation period at each site. Coefficients region which are then summed over the period of 2 of determination (r ) determined by least squares record to produce the cumulative B . Note that the clim linear regression are 0.133 for Wolverine Glacier monitoring periods vary between regions and that the = 0.008) and 0.08 for Gulkana Glaciers ( p = 0.04). ( p number and identity of glaciers monitored in a region (Source: S. O’Neel, USGS.) may vary between years. | S159 AUGUST 2018 STATE OF THE CLIMATE IN 2017

180 SIDEBAR 5.2: INDIGENOUS KNOWLEDGE AND THE COPRODUCTION OF KNOWLEDGE PROCESS: CREATING A HOLISTIC UNDERSTANDING OF ARCTIC CHANGE — C. BEHE AND R. DANIEL Rapid changes occurring within the Arctic heighten methodologies in data collection and analyses to be agreed the need to understand the many causes of the changes upon by all knowledge holders. and their cumulative impacts. Most importantly, to better Successful coproduction of knowledge fosters an understand Arctic change a holistic view is needed that can - environment of trust and respect, works toward empow only be achieved by bringing together multiple knowledge erment and capacity building, and recognizes indigenous systems and scientific disciplines. This includes Arctic knowledge sovereignty; it is important to recognize the Indigenous Peoples and their knowledge. sovereign rights of indigenous peoples, including those Arctic Indigenous Peoples have been an integral part related to their own knowledge. This includes indigenous of the Arctic ecosystem from time immemorial and have peoples fully understanding the risks and opportunities of acquired and built upon a unique knowledge system—an participating in a research project, having authority over indigenous knowledge—shaped by that environment. It how data and information are shared, and the right to not is a systematic way of thinking, which is applied to phe - participate in a research project. The principles of free, nomena across biological, physical, cultural, and spiritual - prior, and informed consent are critical to the coproduc systems. It includes insights based on evidence acquired tion of knowledge process (UN General Assembly 2007). through direct and long-term experiences and extensive Successful coproduction of knowledge is built upon and multigenerational observations, lessons, and skills. - long-term partnerships. A good first step is an understand Indigenous knowledge has developed over millennia ing of the lay of the land in indigenous homelands. Just as - and is still developing in a living process, including knowl scientists understand the importance of networks in their edge acquired today and in the future, and it is passed research, so indigenous peoples also live in complex social on from generation to generation (Inuit Circumpolar and governance systems, allowing the opportunity to Council-Alaska 2016). Indigenous knowledge stresses the leverage existing indigenous networks, institutions, and importance of understanding interconnecting systems, organizations. It is important to understand partnership that is, ecological, physical, cultural, and social systems, building from an indigenous perspective and to know that the relationship between those components, and the partnership and participation are not synonymous. Clear, - need to understand cumulative impacts (Inuit Circum transparent, culturally appropriate terms of reference are polar Council-Alaska 2015). This world view and way of recommended to ensure there are no misunderstandings understanding will aid in gaining a holistic understanding and to help with relationship building. of the Arctic and the changes that are occurring there. Indigenous knowledge and modern science have - To gain a truly holistic understanding of the chang different approaches, methodologies, analyses, and ing Arctic, it is necessary to bring together indigenous validation processes. The coproduction of knowledge knowledge and science through a coproduction of process requires respect for each knowledge system and knowledge process. Such a process offers opportunities avoiding translation of one knowledge system into the to bring together different knowledge systems to develop other, that is, trusting an indigenous knowledge holder’s adaptation policies and practices for sustainability, and to ability to analyze their own information and respect that address biodiversity conservation and ecosystem-based each person at the table comes with the credentials management in a changing Arctic. needed to be there. While some credentials are built from The coproduction of knowledge process brings to - academic degrees and publications, others come from gether indigenous knowledge holders and scientists to holding and demonstrating a body of knowledge through work in partnership from the inception of a project, for a lifetime of hunting, fishing, gathering, or being an elder. - example, identification of research questions and hypoth Many Arctic science projects have aimed to build eses, through analysis and output. Equity is a cornerstone partnerships with indigenous communities, but few of the process, ensuring fairness and the opportunity to have used a true coproduction of knowledge process engage in all aspects of a project. All participants have a that brings together indigenous knowledge holders and fair and equal chance of succeeding. The coproduction scientists equitably from the inception of the project. An of knowledge process requires culturally appropriate exemplar that demonstrates the process is the Ikaa ġ vik | S160 AUGUST 2018

181 g. Terrestrial permafrost— V. E. Romanovsky, S. L. Smith, K. Isaksen, N. I. Shiklomanov, D. A. Streletskiy, A. L. Kholodov, H. H. Christiansen, D. S. Drozdov, G. V. Malkova, and S. S. Marchenko Permafrost is an important component of the Arctic landscape, inf luencing hydrological systems and ecosystems, and presenting challenges for built - infrastructure, for example, buildings, roads, rail Sikukuun (Ice Bridges) project in Kotzebue in northwest ways, airports, and pipelines. Permafrost temperature Alaska (Mahoney et al. 2017). This four-year (2017–20) and active layer thickness (ALT) are key indicators project, which aims to understand fundamental processes of changes in permafrost conditions. Permafrost is underlying the mechanisms and impacts of changing coastal defined as earth materials (e.g., soil, rock) that exist at sea ice, first brought together indigenous knowledge holders or below 0°C continuously for at least two consecutive with scientists from different disciplines to develop the research focus and questions, decide on a methodology, and then agree on a plan for implementing the project together. Indigenous knowledge will also inform the synthesis and dissemination of the results of the project. The success of a coproduction of knowledge process must be defined by both the indigenous knowledge holders and scientists involved in a project. Experts from both knowledge systems must agree that a coproduction of knowledge occurred and it will hold all of the basic elements presented here. These include recognizing and working toward equity through building capacity, empowering indigenous partners, fostering an environment for trust and respect, building a collaborative process that involves multiple steps and continuous evaluation and which is defined by all those involved in a project, and respecting indigenous knowledge sovereignty. F ig . 5.18. Location of the permafrost temperature monitoring sites shown in Fig. 5.19 superimposed on average surface air temperature anomalies (°C) dur - ing 2000–16 (with respect to the 1981–2010 mean) from the NCEP-reanalysis (Kalnay et al. 1996). Data provided by the NOAA/ESRL Physical Sciences Divi - sion (www.esrl.noaa.gov/psd). Sites shown in Fig. 5.19 for (a) Barrow (Ba), West Dock (WD), KC-07 (KC), Duvany Yar (DY), Deadhorse (De), Franklin Bluffs (FB), Galbraith Lake (GL), Happy Valley (HV), Norris Ck (No); (b) College Peat (CP), Old Man (OM), Chandalar Shelf (CS), Birch Lake (BL), Coldfoot (Co), Norman Wells (NW), Wrigley 1 and 2 (Wr), Healy (He), Gul - . 5.17. Cumulative regional glacier mass anomalies ig F kana (Gu); (c) Eureka EUK4 (Eu), Alert BH1, BH2, and (in, Gt) for Iceland, Arctic Canada, Arctic Russia, and BH5 (Al), Resolute (Re), Arctic Bay (AB), Pond Inlet CryoSat2 - radar al Svalbard, derived using data from (PI), Pangnirtung (Pa); (d) Janssonhaugen (Ja), Bayelva timetry (2011–17) (B. Wouters, Utrecht University). (Ba), Kapp Linne 1 (KL), Urengoy #15-06 and #15-10 Cumulative mass anomalies in each region are defined (Ur), Juvvasshøe (Ju), Tarfalaryggen (Ta), Polar Ural relative to the glacier mass measured in the region at (ZS), Bolvansky #56, #59, and #65 (Bo), Iskoras Is-B-2 the start of the measurement period. Trend lines and (Is). Information about these sites is available at http: 1 − average annual rates of mass change (Gt yr ) in each //gtnpdatabase.org/, http://perma-frost.gi.alaska.edu region are shown. Annual cycles in the accumulation /sites_map, and www2.gwu.edu/~calm/data/data-links and removal of mass are evident in each region. .html. | S161 AUGUST 2018 STATE OF THE CLIMATE IN 2017

182 years. The active layer is the seasonally thawed layer 2017 was substantial and comparable to the highest above the permafrost. Permafrost temperatures, at a - rate of warming observed in this region, which oc - curred during 1995–2000 (Fig. 5.19a). Since 2000, depth where seasonal temperature variations are neg permafrost temperature increase at 20-m depth in ligible, are powerful indicators of long-term change. −1 this region has ranged from 0.21° to 0.66°C decade On the other hand, the active layer responds to shorter term f luctuations in climate and is especially (Fig. 5.19a; Table 5.3). In Interior Alaska, following the slight cooling of sensitive to changes in summer air temperature and 2007–13, permafrost temperatures increased and were precipitation. Warming and thawing of permafrost higher in 2017 than in 2016 at all sites (Coldfoot, Old in the Arctic are reported here. Man, College Peat, Birch Lake, Gulkana, and Healy Changes up to 2017 (most recent data available) in Fig. 5.19b). The largest changes, at Birch Lake and in mean annual permafrost temperatures and ALT Old Man, were associated with new record highs in are summarized for a number of sites throughout 2017 for the entire 33-year measurement period (Fig. the Arctic (Fig. 5.18). Recent long-term changes in permafrost temperature are driven mostly by air 5.19b; Table 5.3). temperature trends (Romanovsky et al. 2015). Other important inf luences on permafrost temperature, such as snow depth and density, vegetation charac - teristics, and soil moisture, can affect the observed permafrost temperature trends at the local scale. In general, the increase in permafrost temperatures observed since the 1980s is more significant in the higher latitudes where the largest increase in air tem - perature is observed (Fig. 5.18). - 1) p ermafrost tempera tures Temperatures in the cold continuous perma- frost of northern Alaska, Northwest Territories - (Canada), and northeast ern East Siberia continue to rise (Fig. 5.19a). In 2017 on the North Slope of ig . 5.19. Time series of mean annual ground temperature (°C) at depths of 9 F - Alaska, record high tem to 26 m below the surface at selected measurement sites that fall roughly into peratures at 20-m depth priority regions of the Adaptation Actions for a Changing Arctic Project (AMAP occurred at all perma - 2015): (a) cold continuous permafrost of northern Alaska, Northwest Territo - ries (Canada), and NE East Siberia; (b) discontinuous permafrost in Interior - frost observatories (Bar Alaska and northwestern Canada; (c) cold continuous permafrost of eastern row, West Dock, Franklin and High Arctic Canada (Baffin Davis Strait); and (d) continuous to discontinu - Bluffs, Happy Valley, and ous permafrost in Scandinavia, Svalbard, and Russia/Siberia (Barents region). Deadhorse; Fig. 5.19a) with Temperatures are measured at or near the depth of zero annual amplitude the exception of Galbraith where the seasonal variations of ground temperature are negligible. Note that Lake. The permafrost tem - the temperature scales are different in each graph. Data are updated from perature increase (+0.1° to Christiansen et al. 2010; Romanovksy et al. 2017; Smith et al. 2015, 2017; Ednie and Smith 2015; Boike et al. 2018. +0.2°C) between 2016 and | S162 AUGUST 2018

183 − 1 AB ) for sites shown in Fig. 5.19. For le 5.3. Change in mean annual ground temperature (°C decade t sites where measurements began prior to 2000, the rate for the entire available record is provided as well as the rate for the period after 2000. The names of the stations with record high temperatures in 2017 are shown in red. Note that some records only began after 2007, as shown in Fig. 5.19. Region Sites Entire Record Since 2000 West Dock (WD), Deadhorse (De), +0.36 to +0.8 +0.44 to +0.65 Alaskan Arctic plain Franklin Bluffs (FB), Barrow (Ba) Northern foothills of the +0.34 to +0.47 Happy Valley (HV), Galbraith Lake (GL) +0.3 to +0.42 Brooks Range, Alaska Southern foothills of the Coldfoot (Co), Chandalar Shelf (CS), +0.08 to +0.35 +0.14 to +0.25 Brooks Range, Alaska Old Man (OM) College Peat (CP), Birch Lake (BL), Interior Alaska +0.07 to +0.22 +0.03 to +0.1 Gulkana (Gu), Healy (He) Up to +0.1 <+0.1 to +0.2 Norman Wells (NW), Wrigley (Wr) Central Mackenzie Valley +0.5 to +0.9 Norris Ck (No), KC-07(KC) Northern Mackenzie Valley — Pangnirtung (Pa), Pond Inlet (PI), — +0.5 to +0.7 Baffin Island Arctic Bay (AB) +0.4 to +0.7 — Resolute (Re), Eureka (Eu) High Canadian Arctic +1.2 +0.5 Alert (Al) at 15 m High Canadian Arctic +0.3 to +0.4 24m +0.7 to +0.9 North of East Siberia — +0.3 Duvany Yar (DY) Urengoy 15-06 and +0.31 to +0.47 +0.1 to +0.19 North of West Siberia 15-10 (Ur) Bolvansky 56, 59 , and 65 (Bo), +0.18 to +0.46 +0.1 to +0.83 Russian European North Polar Ural (ZS-124) Janssonhaugen (Ja), Bayelva (Ba), +0.7 +0.6 to +0.7 Svalbard Kapp Linne 1 (KL) Northern Scandinavia Iskoras Is-B-2 (Is) — +0.1 to +0.4 Tar falarg gen ( Ta), +0.2 +0.2 Southern Norway Juvvasshøe (Ju) - In northwestern Canada, the temperature of In northeastern Canada, the 2016/17 mean perma permafrost in the central Mackenzie Valley (Nor frost temperatures in the upper 25 m of the ground - at Alert, northernmost Ellesmere Island in the high man Wells and Wrigley in Fig. 5.19b; Table 5.3) has generally increased since the mid-1980s (Smith et Arctic, were among the highest recorded since 1978 - (Fig. 5.19c). Although permafrost at Alert has gener al. 2017). Although less warming has been observed since 2000, permafrost temperatures in 2017 at these ally warmed since 1978, permafrost temperatures have increased at a higher rate since 2000 (Table 5.3; sites were the highest recorded. Greater recent warm - Smith et al. 2015), consistent with air temperature ing has been observed in the colder permafrost of the trends (Fig. 5.18). There has been little change at Alert northern Mackenzie region (Norris Ck, KC-07 in Fig. 5.19a and Table 5.3; Smith et al. 2017), with the since 2010 (Fig. 5.19c), which coincides with a period of lower mean annual air temperatures. At other sites highest temperatures during the observation period occurring in 2016/17. in the Queen Elizabeth Islands (Resolute and Eureka) and on Baffin Island (Pangnirtung, Pond Inlet, and | S163 AUGUST 2018 STATE OF THE CLIMATE IN 2017

184 Arctic Bay), permafrost temperature measurements are important. Greater warming occurred in colder since 2008, at 10- to 15-m depth, indicate an overall permafrost on Svalbard and in northern Scandinavia - (Table 5.3). In the discontinuous permafrost zone of warming (Fig. 5.19c; Table 5.3). A decrease in perma frost temperature since 2012 appears to be associated southern Norway, permafrost warmed between 2015 and 2017, following a period of cooling between 2011 with lower mean annual air temperatures over the last few years in the region. and 2014 (Fig. 5.19d). Increases in permafrost temperature over the last 2) a 30–35 years in northern Russia have been similar thickness layer ctive In 2017, standardized, mechanical probing of to those in northern Alaska and the Canadian high Arctic (Drozdov et al. 2015). In the Russian European ALT was conducted at 76 Circumpolar Active-Layer North and western Siberian Arctic, temperatures at Monitoring (CALM) program sites in Alaska and −1 10-m depth have increased by ~0.4° to 0.6°C decade Russia. Each site consists of a spatial grid varying 2 from 1 ha to 1 km in size and is representative of since the late 1980s at colder permafrost sites (Fig. the regional landscape (Shiklomanov et al. 2012). 5.19d, sites Bolvansky #59, Urengoy #15-06 and #15- 10) and increased less in warmer permafrost sites Additional active-layer observations, derived from (Table 5.3; Fig. 5.19d, sites Bolvansky #56 and Urengoy thaw tubes (Duchesne et al. 2015), are available from #15-06; Drozdov et al. 2015). In these regions, there 25 Canadian sites located in the Mackenzie Valley, are differences in permafrost temperature (cold vs. northwestern Canada. The average ALT in 2017 for 20 North Slope of warm) because surface conditions such as vegetation, surface wetness, and soil moisture vary according to Alaska sites was 0.52 m, which is 0.06 m (or +12%) landscape types, while climatic conditions are largely higher than the 1996–2005 mean and is one of the independent of surface condition and landscape. highest in the 22-year data record. Previous maxima In the Nordic region, where the temperature at occurred in 1998, 2013, and 2016 (Fig. 5.20). The interior of Alaska is characterized by a pronounced 20-m depth has increased between 0.1° and 0.7°C −1 ALT increase over the last 22 years (Fig. 5.20). How - (Fig. 5.19d; Table 5.3) since 2000, warm - decade ing and thawing of permafrost have been observed ever, after reaching the 22-year maximum of 0.77 m recently (Christiansen et al. 2010; Isaksen et al. 2011; in 2016, the 2017 ALT decreased by 10% to 0.69 m. Farbrot et al. 2013). Lower rates of warming occur Records from the sites with thaw tubes in the where permafrost temperatures are close to 0°C Mackenzie Valley indicate that there has been a gen - and latent heat effects related to melting ground ice eral increase in ALT in this region since 2008 (Fig. F ig . 5.20. Long-term annual active-layer thickness change (m) in six different Arctic regions for 2017 as observed by the CALM program relative to the 2003–12 mean. Positive (negative) anomaly values indicate the active layer was thicker (thinner) than average. Thaw depth observations from the end of the thawing season were used. Only sites with at least 20 years of continuous thaw depth observations are shown. | S164 AUGUST 2018

185 5.20; Duchesne et al. 2015; Smith et al. 2017). ALT in 2017; Macias-Fauria et al. 2017; see Section 5d). Other 2016 (most recently available data) was on average controls on tundra greening include increases in - snow water equivalent (see Section 5i) and soil mois 0.06 m greater than the 2003–12 mean, similar to the ture, increases in active layer depth (see Section 5g), previous peak value in 2012. A decrease in ALT from 2016 to 2017 was reported changes in the patterns of herbivore activity, and even a reduction in the human use of the land (Fauchald for all Russian regions. In West Siberia, the average 2017 ALT was 1.25 m, which is 0.2 m (or 14%) smaller et al. 2017; Horstkotte et al. 2017; Martin et al. 2017; Westergaard-Nielsen et al. 2017). than the 20-year maximum observed in 2016. In the Using Earth-observing satellites with subdaily Russian European North, the 2017 ALT was 1.08 m return intervals, Arctic tundra vegetation has been compared to 1.24 m in 2016. A 2017 ALT of 0.69 m continuously monitored since 1982. Here, data are was reported for East Siberia, which is 0.1 m smaller reported from the Global Inventory Modeling and than the regional average 2016 ALT value. The small - est decrease was reported in the Russian Far East Mapping Studies (GIMMS) 3g V1 dataset, based (Chukotka), where the ALT in 2017 was 0.03 m (or largely on the AVHRR sensors aboard NOAA satel - 5%) less than that reported in 2016. lites (Pinzon and Tucker 2014). At the time of writing, the GIMMS3g V1 dataset was only available through In the Nordic region, active layer records (1999– 2016. The GIMMS product (at 1/12° resolution for 2017) indicate a general ALT increase of 0.10 to 0.30 this report) is a biweekly, maximum-value compos m since 1999. The particularly warm summer of 2014 - ited dataset of the normalized difference vegetation in the Nordic region contributed to the thickest active index (NDVI). NDVI is highly correlated with above- layer measured so far at some places. ground vegetation (e.g., Raynolds et al. 2012), or “greenness,” of the Arctic tundra. Two metrics based h. Tundra greenness— H. Epstein, U. Bhatt, M. Raynolds, on the NDVI are used: MaxNDVI and TI-NDVI. D. Walker, J. Pinzon, C. J. Tucker, B. C. Forbes, T. Horstkotte, MaxNDVI is the peak NDVI for the year (growing M. Macias-Fauria, A. Martin, G. Phoenix, J. Bjerke, H. Tømmervik, season) and is related to yearly maximum above- P. Fauchald, H. Vickers, R. Myneni, T. Park, and C. Dickerson ground vegetation biomass. TI (time-integrated) Vegetation in the Arctic tundra has been respond - NDVI is the sum of the biweekly NDVI values for ing to environmental changes over the course of the last several decades, with the tendency being an the growing season and is correlated with the total increase in the amount of above-ground vegetation, above-ground vegetation productivity. Examining the overall trend in tundra green - that is, “greening” (Bhatt et al. 2010). These vegetation ness for the now 35-year record (1982–2016), it is changes vary spatially throughout the circumpolar apparent that the MaxNDVI and the TI-NDVI have Arctic in both direction and magnitude, and they increased throughout most of the circumpolar Arctic are not always consistent over time. This suggests tundra (Fig. 5.21). Regions with some of the greatest complex interactions among the atmosphere, ground increases in tundra greenness are the North Slope of (soils and permafrost), vegetation, and animals of the Arctic system. Changes in tundra vegetation can Alaska, the low Arctic (southern tundra subzones) of the Canadian tundra, and eastern Siberia. However, have important effects on permafrost, hydrology, carbon and nutrient cycling, and the surface energy tundra greenness has declined (i.e., the tundra has been “browning”) on the Yukon–Kuskokwim Delta balance (e.g., Frost et al. 2017; Kępski et al. 2017), as of western Alaska, in the high Arctic of the Canadian well as the diversity, abundance, and distribution of both wild and domesticated herbivores (e.g., Fauchald Archipelago, and in northwestern Siberia. Regions of et al. 2017; Horstkotte et al. 2017). We continue to greening and browning, measured by NDVI increases evaluate the state of the circumpolar Arctic vegeta - and decreases, respectively, tend to be consistent between MaxNDVI and TI-NDVI. tion, to improve our understanding of these complex interactions and their impacts on the Arctic system Following 2–3 years of successive declines prior to and including 2014, the NDVI or greenness of Arctic and beyond. tundra increased in 2015 and 2016 for both indices The reported controls on tundra greening are - (MaxNDVI and TI-NDVI) and both continents numerous and varied. They include increases in sum (North America and Eurasia), exhibiting substantial mer, spring, and winter temperatures and increases in growing season length (Bhatt et al. 2017; Fauchald recovery from the previous years of “browning.” et al. 2017; Horstkotte et al. 2017; Myers-Smith et (Fig. 5.22). One exception was the TI-NDVI for al. 2018; Vickers et al. 2016), in part controlled by North America, which continued to decrease in reductions in Arctic Ocean sea ice cover (Bhatt et al. 2015. MaxNDVI and TI-NDVI for the entire Arctic | S165 AUGUST 2018 STATE OF THE CLIMATE IN 2017

186 increased 6.0% and 9.3%, respectively, between 2015 and North American Arctic, respectively. Based on remotely-sensed land surface temperatures (LST) and 2016. MaxNDVI in North America increased by from the same sensors as those providing the NDVI - 6.3% compared to 5.4% in Eurasia. The first substan tial annual increase in TI-NDVI for North America values, the summer warmth index (SWI: sum of mean since 2010 occurred in 2016, potentially due to the monthly temperatures >0°C) for the Arctic as a whole high growing season temperatures that year. and for the Eurasian Arctic was greater in 2016 than All NDVI values for 2016 were greater than in any other year of the satellite record (since 1982). their respective mean values for the 35-year record. For the North American Arctic, the 2016 SWI was the second highest on record (very close to the high MaxNDVI values ranked second, third, and first for - the Arctic, Eurasian Arctic, and North American est value in 1994). Arctic, respectively. TI-NDVI values ranked first, Even though the past two years have seen large increases in tundra NDVI, there are still regions of first, and second for the Arctic, Eurasian Arctic, the Arctic that have experienced browning over the length of the satellite record. There have also been substantial periods of tundra browning even within a general greening trend. While research on tundra browning is still relatively sparse, there has recently been greater attention given to this phenomenon. Bjerke et al. (2017) report on extensive vegetation dieback in northern Norway (including Svalbard) in 2014 and 2015. They attributed this dieback largely to F . 5.22. (a) MaxNDVI and (b) TI-NDVI for Eurasia ig . 5.21. (a) Magnitude of the trend in (a) MaxNDVI ig F (top), the Arctic as a whole (middle), and North and (b) TI-NDVI for 1982–2016 America (bottom) for 1982–2016. | S166 AUGUST 2018

187 SIDEBAR 5.3: WILDLAND FIRE IN BOREAL AND ARCTIC NORTH A. YORK, U. BHATT, R. THOMAN, AND R. ZIEL AMERICA — Despite the low temperatures and short growing seasons of northern ecosystems, - wildland fire is the dominant ecological dis turbance in the boreal forest, the world’s largest terrestrial biome. Wildland fire also affects adjacent tundra regions. This sidebar, with a focus on the 2017 Alaska fire season, addresses the history and variability of fire disturbance in Alaska (US) and Northwest Territories (Canada), outlines how short-term weather conditions (temperature, precipita- tion, convection, and wind) influence area burned, and discusses projections for future tendencies in fire susceptibility. Beyond immediate threats to lives and property, fire impacts include compromised human health and limited visibility due to smoke. Fire disturbance affects terrestrial - ecosystems at multiple scales, including car bon release through combustion (Kasischke et al. 2000). About 35% of global soil carbon is stored in tundra and boreal ecosystems (Scharlemann et al. 2014) that are potentially vulnerable to fire disturbance (Turetsky et al. 2015). Other impacts include interactions with vegetation succession (Mann et al. 2012; Johnstone et al. 2010), biogeochemical cycles (Bond-Lamberty et al. 2007), energy balance (Rogers et al. 2015), and hydrology (He. Liu et al. 2005). Combustion of the insulating surface organic layer can destabilize underly - ing permafrost. Because permafrost impedes drainage and ice-rich permafrost settles upon . SB5.3. Annual area burned (ha) each year since 1980 in (a) Alaska ig F - thawing (thermokarst), accelerating degrada and (b) Northwest Territories (Canada), including both boreal and - tion of the permafrost may have large conse tundra regions. Note that high fire years are not coincident in these quences for northern ecosystems (Jorgenson subregions, indicating the importance of local weather and other et al. 2010; Jones et al. 2015). conditions (e.g., fuels, ignition). Category definitions used here are - Weather is a dominant control of fire ac from the fitted log-normal distribution to the observed 1980–2017 tivity on a year-to-year basis. Over the longer area burned; below normal is the 0–33rd percentiles, near normal is term, high-latitude fire regimes appear to be the 33rd–66th percentiles, above normal is the 66th–90th percentiles, much above is greater than the 90th percentile. responding rapidly to environmental changes associated with the warming climate. Although highly variable, area burned has increased since the 1960s in sischke and Turetsky 2006). Figure SB5.3 shows area burned much of boreal North America (Kasischke and Turetsky 2006; each year since 1980 in Alaska and Northwest Territories, Gillett et al. 2004). Over that time, both the number and size including both boreal and tundra regions. of individual fire events has increased, contributing to more - Although highly variable, high-latitude fire seasons gen frequent large fire years in northwestern North America (Ka - erally begin and end earlier than in more temperate areas | S167 AUGUST 2018 STATE OF THE CLIMATE IN 2017

188 WILDLAND FIRE IN BOREAL AND ARCTIC NORTH SIDEBAR 5.3: CONT. AMERICA — A. YORK, U. BHATT, R. THOMAN, AND R. ZIEL (Fig. SB5.4). Depending on weather, fire danger can increase as soon as areas are snow-free in April and May; season-ending rains typically fall in July or August, but their absence can extend the season into September, as in the record years of 2004 (2.67 million ha) and 2005 (1.88 million ha) in Alaska. Recent large fire seasons in high latitudes include 2014 in Northwest Territories (Fig. SB5.3), where 385 fires burned 3.4 million ha, and 2015 in Alaska (Fig. SB5.3), where 766 fires burned 2 million ha—the latter was more than half the total area burned in the entire United - States (NWT 2015; AICC 2015). North ig F . SB5.4. Average (gray line) and climatological range (gray shading) of ern communities threatened or damaged BUI between 1 Apr and 30 Sep in Alaska’s boreal interior for 1994–2017, by recent wildfires include Fort McMurray, compared to the 2017 average (solid purple line) and the 2017 predictive service area AK02 (Upper Yukon and surrounding uplands, centered located in the boreal forest in Alberta, around the Arctic Circle; dashed purple line). While the boreal interior 0 - 00 people were evacu Canada, where 88 average BUI for 2017 (purple line) was similar to the historic average BUI ated and 2400 structures were destroyed - (gray line), the Upper Yukon Zone (dashed purple line), where the major in May 2016 (Kochtubajda et al. 2017). The ity of the hectares burned in the territory in 2017, showed a significant 2007 Anaktuvuk River Fire is the largest elevation in BUI from mid-Jun to mid-Aug. 00 ha) and longest-burning (almost (104 0 3 months) fire known to have occurred on the North Slope North America since 1975 and were a major contributor in of Alaska and initiated widespread thermokarst development the extreme 2014 Northwest Territories and 2015 Alaska fire (Jones et al. 2015). seasons. In addition, Partain et al. (2016) found that human- Most area burned in northern ecosystems occurs during induced climate change—manifested as a combination of high sporadic periods of high fire activity. Half of the area burned - surface air temperatures, low relative humidity, and low pre in Alaska from 2002 to 2010 was consumed over just 36 days cipitation—increased the likelihood of the extremely dry fuel - (Barrett et al. 2016). Recent analyses have identified a tempera conditions seen in Alaska in 2015 by 34%–60%. ture threshold in Alaska with a much greater likelihood of fire The snow-free season has increased by approximately 5 −1 in Alaska since 1979 (Liston and Hiemstra 2011). occurrence within a 30-year period at locations where mean days decade In response, in 2006 Alaska’s fire management agencies shifted July temperatures exceed 13.4°C (Young et al. 2017). Large fire the statutory start of fire season ahead by a month, from 1 events require the confluence of warm and dry weather condi - May to 1 April, to better prepare for early season events. In tions with a source of ignition (often lightning from convective addition to adapting to long-term trends, managers in Alaska thunderstorms) and fuels that can carry fire. High latitude - and Canada must track day-to-day variability in threats to dis ecosystems are characterized by unique fuels, in particular, persed populations with limited resources. Managers in both fast-drying beds of mosses, lichens, and accumulated organic regions use the Canadian fire weather index (FWI) system on material (duff) that underlie resinous shrubs and dense, highly a daily basis to estimate the spatial and temporal distribution of flammable conifers. These understory fuels dry rapidly during wildfire potential from observed and forecast weather condi - periods of warm, dry weather and the long day lengths of June tions (Lawson and Armitage 2008). Among the FWI indices, and July. Consequently, extended drought is not required to the buildup index (BUI), based on cumulative scoring of daily increase fire danger to extreme levels. temperature, relative humidity, and precipitation, represents Historically, lightning is responsible for the majority of the seasonal variability in fuel availability and flammability (Fig. acreage burned in high latitudes, as lightning-ignited fires occur SB5.4). A BUI threshold of 80 has been identified as a critical in more remote locations and thus are subject to lower levels indicator of fire growth potential in Alaska (Ziel et al. 2015). of suppression than human-started incidents. Veraverbeke et al. (2017) showed that lightning ignitions have increased in boreal | S168 AUGUST 2018

189 In 2017, the typical area burned in Alaska (264 2 21 ha; Under a range of climate change scenarios, analyses Fig. SB5.3) was reflected in a fairly normal BUI across the using multiple approaches project significant increases (up boreal region that essentially paralleled the climatological to four-fold) in area burned in high latitude ecosystems by average (Fig. SB5.4). However, the impact of a “normal” the end of the 21st century (French et al. 2015; Young et al. season can fall disproportionately on specific areas in a 2017; Yue et al. 2015, and references therein). In addition, - landscape this large. In 2017, while there were no signifi annual lightning frequency is projected to increase by 12% cant peaks in the BUI, local conditions in the Upper Yukon ± 5% per °C of warming in the contiguous United States zone in northeast Alaska were significantly warmer and (Romps et al. 2014) and may increase correspondingly drier. Consistent with the Upper Yukon BUI trend (Fig. in high latitudes. Because specific fire events depend on SB5.4), the fire season was extended and fairly severe in multiple interacting factors, the resulting changes in high that large region of the state, with periods of high fire latitude fire regimes will vary greatly over space and time, danger (BUI �80) from mid-June to mid-August near and but all evidence indicates that northern ecosystems will 00 ha (63% of 0 north of the Arctic Circle. More than 160 become increasingly susceptible to burning. the 2017 Alaska total) burned in the Upper Yukon area during this period. changes in winter weather, specifically reductions in et al. 2015; http://climate.rutgers.edu/snowcover/; Fig. - snow cover areal extent due to winter warming events, 5.23). For the first time in over a decade, 2017 Eur which left the ground exposed to subsequent freezing asian Arctic spring SCE was above average relative to the 1981–2010 reference period. April and May SCE and desiccation (Vikhamar-Schuler et al. 2016). Insect anomalies were positive, including the second high - outbreaks were identified as a secondary contributor to vegetation mortality (Bjerke et al. 2017). est May SCE over the period of satellite observations. These are the first positive SCE anomalies observed C. Derksen, R. Brown, in May over the Eurasian Arctic since 2005; June SCE i. Terrestrial snow cover in the Arctic— L. Mudryk, K. Luojus, and S. Helfrich anomalies were positive across the Eurasian Arctic Satellite-derived estimates of snow cover extent for the first time since 2004. SCE anomalies over the North American Arctic were negative all spring but (SCE) over Arctic land areas date back to 1967 and did not approach the series of record-breaking low have revealed dramatic reductions since 2005. These SCE values observed in recent years. changes are important to the Arctic system because Snow cover duration (SCD) departures were spring snow cover over land areas significantly inf luences the surface energy budget (snow is calculated from the NOAA Interactive Multisensor Snow and Ice Mapping System (IMS; Helfrich et al. highly ref lective of incoming solar energy), ground 2007) product to identify differences in the onset of thermal regime (snow is an effective insulator of the snow cover in fall and melt of snow cover in spring underlying soil), and hydrological processes (the relative to a 1998–2010 reference period. While there snowpack stores water in solid form for many months before spring melt). Changes in snow cover also have was evidence of earlier snow cover onset over much of midlatitude Eurasia in autumn 2016 (consistent with the potential to impact fauna living above, in, and cold surface air temperature anomalies), Arctic land under the snowpack, vegetation, biogeochemical areas (with the exception of Alaska) had near-normal activity, and exchanges of carbon dioxide and other trace gases (Brown et al. 2017). snow onset timing (Fig. 5.24a). Later-than-normal Spring (April–June) SCE anomalies for the Arctic snow melt onset across Eurasia (Fig. 5.24b), also ref lected in the positive SCE anomalies (Fig. 5.23), (land areas north of 60°N) were regionally computed for North America and Eurasia using the NOAA snow was consistent with colder-than-normal surface air chart climate data record, which extends from 1967 temperatures across this region (especially in May to present (maintained at Rutgers University; Estilow and June). Spring snow melt across the Canadian | S169 AUGUST 2018 STATE OF THE CLIMATE IN 2017

190 3 2 F . 5.23. (a) Monthly SCE anomaly (× 10 ) for Arctic land areas (> 60°N) from the NOAA snow chart CDR ig km for (a) Apr, (b) May, and (c) Jun from 1967 to 2017. Anomalies are relative to the average for 1981–2010 and standardized (each observation is differenced from the mean and divided by the standard deviation and is thus unitless). Solid black and red lines depict 5-yr running means for North America and Eurasia, respectively. Solid symbols denote anomalies for 2017. Arctic was slightly earlier than normal, coincident with warmer-than-average surface temperatures in May and June. Snow depth anomalies derived from the Canadian Meteorological Centre daily gridded global snow depth analysis (Brasnett 1999) showed predominantly positive anomalies over high latitude regions of Siberia and North America in April (Fig. 5.24c) and mainly negative anomalies outside the Arctic. By late spring (June), the anomalies exhibited contrasting continental patterns, with Eurasia char - acterized by extensive positive snow depth anomalies, while the North American Arctic was dominated by negative snow depth anomalies (Fig. 5.24d), consis - tent with the region of earlier snow melt (Fig. 5.24b). Four independent products were integrated to generate a multidataset snow water equivalent (SWE; the amount of water stored in solid form as snow) anomaly time series (1980–2017) for April (typically the month of maximum SWE across the Arctic; Fig. 5.25). The datasets were derived from: (1) modern - atmospheric reanalysis (the Modern-Era Retrospec tive Analysis for Research and Applications version 2; Reichle et al. 2017); (2) reconstructed SWE driven by ERA-Interim meteorology described by Brown et al. (2003); (3) the physical snowpack model Crocus driven by ERA-Interim meteorology (Brun et al. 2013); and (4) the European Space Agency GlobSnow product derived through a combination of satellite . 5.24. SCD anomalies (%) from the NOAA daily ig F passive microwave measurements and climate station IMS snow cover product (Helfrich et al. 2007; rela - observations (Takala et al. 2011). While there is a high tive to 1998–2010 base period because of the shorter degree of interannual variability in the multidataset available time series but higher spatial resolution IMS SWE anomalies, they predominantly show a negative data). IMS data record for the (a) 2016 autumn season trend since 2000 (Fig. 5.25). North American Arctic and (b) 2017 spring season. Snow depth anomaly (% of SWE was again negative in 2017 while Eurasian SWE 1999–2010 average) from the CMC snow depth analysis for (c) Apr and (d) Jun 2017. anomalies were positive, indicating a deeper-than- | S170 AUGUST 2018

191 (altitude of approximately 15 to 25 km), which lead to the formation of polar stratospheric clouds (PSCs). These clouds act as a catalyst to transform inactive forms of chlorine-containing substances (e.g., HCl and ClONO ) to active, ozone-destroying chlorine 2 species (e.g., ClO). Temperatures in the Arctic stratosphere between late November and late December 2016 were about 5°C higher than the average temperature of the observational record (1979–2015); temperatures in late November 2016 were near the highest values on record for this period. Temperatures dropped below the threshold for the formation of PSCs only in late December. (The onset of PSC formation is typically in early December with the earliest onsets observed in mid-November.) Temperatures remained low enough to sustain PSCs through mid-February 2017. ig F . 5.25. Mean Apr SWE anomalies for Arctic land Starting in late December, modest chlorine activa - areas calculated from four independent products for Aura tion was measured by the Microwave Limb North American (black) and Eurasian (red) sectors of Sounder (MLS). From late January to mid-February the Arctic. Anomalies are relative to the average for 2017, active chlorine (ClO) concentrations were, on - 1981–2010 and standardized (each observation is dif average, 45% higher than the mean concentration ferenced from the mean and divided by the std. dev. and is thus unitless). Solid black and red lines depict - calculated from the MLS data record (2005–16) be 5-yr running means for North America and Eurasia, cause stratospheric temperatures during this period respectively, and shading indicates the interdataset were about 4°C below average. Ozone decreases via anomaly spread (± 1 std. dev.). Solid symbols denote - destruction by activated chlorine started in late Janu anomalies for 2017. ary and continued through mid-March 2017. After average snowpack in early spring was a precursor to mid-March, chlorine was deactivated and chemical the above-average snow extent that followed later in ozone destruction ceased. Between December 2016 and mid-January 2017, the season. ozone mixing ratios (a measure of ozone concentra - Despite the long-term decline in Arctic spring SCE tions) were close to the upper limit of values from the driven by increasing temperature trends, negative snow anomalies are not consistently observed in every observational record (2004–17) (Fig. 5.26). At the end season (nor in all regions). Off-trend anomalies, such of January, mixing ratios started to decline and fell below average in March and April 2017. However, in as those observed in the Eurasian Arctic in 2017, are - driven by natural variability in atmospheric circula tion patterns which drive regional temperature and precipitation anomalies. The rebound in Eurasian SCE during May and June 2017 was consistent with winter and spring season circulation patterns which generally favored colder surface temperatures, en - - hanced precipitation, and above-average snow ac cumulation across northern Eurasia. j. Ozone and UV radiation— G. H. Bernhard, V. E. Fioletov, J.-U. Grooß, I. Ialongo, B. Johnsen, K. Lakkala, G. L. Manney, F ig . 5.26. Average ozone mixing ratios (ppmv) mea - and R. Müller MLS at an altitude of ~18 km for the Aura sured by This report emphasizes the November 2016 to area bounded by the polar vortex. Data from 2016/17 April 2017 period because chemically-induced loss (red), 2015/16 (green), and 2010/11 (blue) are compared of polar ozone occurs predominantly during winter with the average (solid white) and minimum/maximum and spring (WMO 2014). Chemical processes that range (gray shading) from 2004/05 to 2014/15, exclud - drive ozone depletion are initiated at temperatures ing 2010/11. Gaps in the record for 2010/11 are due to missing data. below about 195K (−78°C) in the lower stratosphere | S171 AUGUST 2018 STATE OF THE CLIMATE IN 2017

192 of ozone occurs) broke up and air from high and mid - comparison to 2010/11 and 2015/16 (the years with latitudes started to mix, departed by less than ±10% the largest chemical ozone loss in the observational from the historical average, and ozone anomalies for record), mixing ratios in 2016/17 remained well above values observed in those record years. May through November 2017 were unremarkable. UV radiation is quantified with the UV index The evolution of the Arctic total ozone column (UVI), which is a measure of the ability of UV ra - (TOC; i.e., ozone amounts integrated from the diation to cause erythema (sunburn) in human skin surface to the top of the atmosphere) is used here to (WHO 2002). In addition to its dependence on TOC, compare 2017 measurements to the observational record (1979–2016). Specifically, March TOC is evalu the UVI depends on the sun angle, cloud cover, and - surface albedo (Weatherhead et al. 2005). In the Arc - ated because chemically induced Arctic ozone loss tic, the UVI scale ranges from 0 to about 7, with sites is typically largest in this month (WMO 2014). The minimum Arctic daily TOC measured by satellites in closest to the North Pole having the smallest peak March 2017 was 345 Dobson units (DU), which was radiation and UVI values < 4 all year. UVI values ≤ 5 indicate low to moderate risk of erythema (WHO 7.7% (29 DU) below the average of the observational record (374 DU) and 5.4% (20 DU) below the 2005–16 2002). UVI anomalies are assessed using satellite in - average when MLS data are also available (Fig. 5.27). struments (OMI) and ground-based measurements, Spatial deviations of monthly average TOCs from with the former providing the better spatial coverage historical (2005–16) averages (Figs. 5.28a,b) were and the latter providing greater regional accuracy - estimated with measurements from the Ozone Moni (Bernhard et al. 2015). Figures 5.28c,d quantify the toring Instrument (OMI), which is co-located with spatial differences in monthly average noontime UVIs from historical (2005–16) averages and are MLS on the Aura satellite. Average TOCs for March based on OMI measurements. Figures 5.28c,d also 2017 were up to 15% higher over the Norwegian Sea, indicate anomalies calculated from ground-based Greenland, and northern Canada, and up to 20% lower over northern Siberia relative to the long-term measurements at ten research stations located throughout the Arctic and Scandinavia. average (Fig. 5.28a). This spatial pattern is similar to Compared to the historical mean, average noon - a recently described Eurasia–North America dipole time UVIs for March 2017 were larger by up to 25% mode for the month of February, consisting of a shift over northern Siberia and smaller by up to 20% over to negative ozone anomalies over Eurasia and positive Greenland and the Davis Strait (Fig. 5.28c). Areas anomalies over North America (Zhang et al. 2018). with high UVIs roughly match areas with low TOCs Monthly average TOCs for April 2017, the month and vice versa, but UVI anomalies have a larger spa when the polar vortex (the low-temperature cyclone - in which most of the springtime chemical destruction tial variability because of their added dependence on cloud cover. While relative UVI anomalies can be high, absolute anomalies remained below 1 UVI - unit because solar elevations in March in the Arc tic remain low. Anomalies derived from OMI and ground-based measurements agree to within ±7%. Anomalies for April 2017 differed by less than ±15% from the historical average (not shown), except at the western coast of Alaska and the Bering Strait, where . 5.27. Area-averaged minimum total ozone col ig - F OMI measured anomalies of up to 50%. umn (DU) for Mar that are calculated poleward of 63° - Ground-based UV measurements at all sites var equivalent latitude (Butchart and Remsberg 1986). ied within historical bounds from July to November. Open circles represent years in which the polar vortex However, UVIs at Alert, Eureka, and Resolute in broke up before Mar, resulting in relatively high values due to mixing with lower latitude air masses and a northernmost Canada and at Summit, Greenland, lack of significant chemical ozone depletion. Red and were unusually high between 15 May and 15 June blue lines indicate the average TOC for 1979–2016 despite only small negative TOC anomalies (Fig. and 2005–16, respectively. Data are adapted from 5.28b). At Alert, Resolute, and Summit, positive UVI Müller et al. (2008) and WMO (2014), updated using anomalies of between 5% and 10% measured at the ERA-Interim reanalysis data (Dee et al. 2011a). Ozone - ground were in good (±3%) agreement with the satel - data from 1979–2012 are based on the combined to lite data. At Eureka, heavy snowfall in mid-May led tal column ozone database version 3.0 produced by to high surface albedo and high UVIs until mid-June. Bodeker Scientific (www.bodekerscientific.com/data /total-column-ozone). Data for 2013–17 are from OMI. Measurements from the ground indicated a positive | S172 AUGUST 2018

193 F . 5.28. (a) Anomalies of TOC (%) and (c) noontime UVI (%) for Mar 2017. (b) and (d) as in (a) and (c) but for ig 15 May–15 Jun. Anomalies are relative to 2005–16 averages. Maps are based on OMTO3 Level 3 total ozone product (Bhartia and Wellemeyer 2002). (c) and (d) also compare UVI anomalies from OMI (first value in parenthesis) with ground-based measurements at 10 locations (second value presented). Gray shading indicates areas where no OMI data are available. UVI anomaly of 25%, while OMI reported a negative anomalies derived from OMI and ground-based anomaly of −40%. This large inconsistency can be measurements at Finse, Norway, for the same period can be attributed to snow cover disappearing 20 days attributed to systematic errors in the OMI dataset, earlier in 2017 compared with the average snow disap which are caused by a mismatch of the actual high - pearance date for 2005–16 (Fig. 5.28d). Differences - surface albedo and the albedo climatology (Tans - between satellite and ground-based measurements kanen et al. 2003) used by the OMI UV algorithm. Be cause of this mismatch, the high ref lectivity observed at Eureka and Finse illustrate that UV estimates from space due to snow was misinterpreted as cloud from space require verification with ground-based measurements, in particular during months when cover, resulting in erroneously low UVIs reported by OMI. The relatively large difference (12%) of UVI snowmelt occurs. | S173 AUGUST 2018 STATE OF THE CLIMATE IN 2017

194 | S174 AUGUST 2018

195 6. ANTARCTICA T. Scambos and S. Stammerjohn — shallower ocean mixed layers, and delays in the a. Overview— T. Scambos and S. Stammerjohn Eds. autumn ice edge advance over the West Antarctic Last year we reported on an unusual Antarctic ocean sector. The seasonal melt extent and melt climate and sea ice anomaly that developed dur - index over the continent were the second highest ing late winter-early spring 2016 (Stammerjohn since 2005, mostly due to strong positive anoma - and Scambos 2017; see also Schlosser et al. 2018; lies of air temperature over most of the West Ant - Stuecker et al. 2017). This anomaly pattern coincided arctic coast. In contrast, the East Antarctic Plateau with record-breaking negative southern annular recorded record low mean temperatures in March. mode (SAM) index values. Antarctic-wide climate Over the coast and adjacent ocean, conditions were anomalies, including weakened westerly winds, high near average. continental surface pressures and temperatures, and In contrast to autumn, winter ushered in anoma • - the lowest spring sea ice extent on record, stood in lously low surface pressures and temperatures over stark contrast not only to the beginning of 2016, but the continent, and a pronounced zonal wave-three also to the previous four record-breaking high sea pattern existed over midlatitudes from June to Sep - ice years (2012–15). This unusual climate anomaly tember. However, by spring much of the continent - pattern continued until late summer 2017 (Febru experienced near-average pressures and slightly ary–March), after which SAM index values slowly above-average temperatures from October to shifted towards positive for the remainder of the year December, with an East Antarctic station setting (except for a short-lived reversal in October). Sea ice a record high temperature in October. Exceptions extent remained low for most of 2017, as discussed were a record low continental surface pressure in Section 6e. recorded for November on the West Antarctic Ice In general, 2017 was notable for its strong regional Sheet and a low-pressure anomaly centered over climate anomalies. An anticyclone pattern in January the northwestern Weddell Sea (~60°S and ~40°W), in the South Pacific abruptly shifted to an intense which contributed to anomalously warm surface - cyclonic anomaly in late summer–autumn (Febru conditions over the central-eastern Weddell Sea ary–May). A strong zonal wave-three pattern then °W to ~30°E). The latter was coincident with (~3 0 emerged in winter (June–September). Sea ice extent the Maud Rise polynya, which expanded and remained below average for the entirety of 2017, with persisted into early December. record lows persisting for the first four months, fol - The brief appearance of the Maud Rise polynya • lowed by the re-emergence of the Maud Rise polynya in 2016 and its greater presence in 2017 is signifi - in the Weddell Sea in mid-September. The year 2017 cant, as it appears to announce a revival of deep was also distinguished by the second smallest Antarc - ocean convection in the eastern Weddell Sea (see tic ozone hole observed since 1988. (Note: throughout Sidebar 6.1). the chapter, anomalies and stan - dard deviations are with respect to the 1981–2010 climatological mean, unless otherwise specified.) Additional highlights for 2017: • In association with the strong cyclonic pattern in the South Pacific during February–May, anomalously warm near-sur - face atmosphere conditions persisted over much of West Antarctica, including over the ocean areas from the Ross Sea to the Bellings-hausen Sea. Record maximum surface temperatures were observed at several western Peninsula stations in March and on the Ross Ice Shelf in May, along with anomalously F ig . 6.1. Map of stations and other regions discussed in the chapter. warm summer–autumn SSTs, | S175 AUGUST 2018 STATE OF THE CLIMATE IN 2017

196 • The mid-depth Southern - Ocean (~500–1500 m) con tinued to warm at rates up -1 while the sur to 0.02°C yr - face ocean continued to cool -1 by −0.015° to −0.05°C yr , freshen, and acidify. The state of Antarctica’s cli - mate, weather, ice, ocean, and - ozone in 2017 are presented be low. Place names used through - out this chapter are provided in Fig. 6.1. b. Atmospheric circulation and surface observations— K. R. Clem, S. Barreira, R. L. Fogt, S. Colwell, C. Costanza, L. M. Keller, and M. A. Lazzara. Atmospheric circulation pat - terns are the main drivers of all other aspects of this climate summary, affecting sea ice, pre - cipitation, weather records, and even ozone loss. Moreover, long- term changes in climate are impacting ocean circulation (e.g., Schneider et al. 2012) and surface melt patterns (e.g., van den Broeke 2005) and thereby affecting the ice mass balance of the Antarctic ice sheet. For this detailed analysis of the atmospheric circulation and temperature and pressure anomalies, the European Cen - tre for Medium-Range Weather ig . 6.2. Area-averaged (weighted by cosine of latitude) monthly anomalies F Forecasts interim reanalysis over the southern polar region in 2017 relative to 1981–2010: (a) polar cap (ERA-Interim) is utilized as it (60°–90°S) averaged geopotential height anomalies (contour interval is is shown to be the most reliable 50 m up to ±200 m with additional contour at ±25 m, and 100 m contour interval after ±200 m); (b) polar cap averaged temperature anomalies - representation of Antarctic tro (contour interval is 1°C with additional contour at ±0.5°C); (c) circumpolar pospheric pressure and tempera - 1 − (50°–70°S) averaged zonal wind anomalies (contour interval is 2 m s ture among all modern global 1 − with additional contour at ±1 m s ). Shading depicts std. dev. of monthly reanalyses (Bracegirdle and anomalies from the 1981–2010 climatological average as indicated by Marshall 2012). Figure 6.2 shows color bar at bottom. (Source: ERA-Interim reanalysis.) Red vertical bars the monthly geopotential height indicate the four climate periods used for compositing in Fig. 6.3; the (Fig. 6.2a) and temperature (Fig. dashed lines near Dec 2016 and Dec 2017 indicate circulation anomalies wrapping around the calendar year. Values from the NOAA CPC Antarctic 6.2b) anomalies averaged over Oscillation index (herein referred to as the SAM index) are shown below the polar cap (60°–90°S) and (c) in black (positive values) and red (negative values). the monthly circumpolar zonal wind (Fig. 6.2c) anomalies aver - aged over 50°–70°S. Anomalies are contoured and shading. The year was grouped into four periods - the standard deviation level is indicated by colored characterized by relatively consistent climatic fea | S176 AUGUST 2018

197 tures: January, February–May, June–September, and October–December. These periods are indicated by vertical red bars at the bottom of each panel in Fig. 6.2. Anomalies for the four groups from their respec - tive group climatological mean are shown in Fig. 6.3, with surface pressure anomalies shown on the left and 2-m temperature anomalies shown on the right. Monthly temperature and pressure anomalies during 2017 are also displayed in Fig. 6.4 for three staffed stations (Amundsen–Scott, Casey, and Rothera) and three automatic weather stations (AWS; Byrd, Dome C II, and Ferrell) to examine the monthly variability and extreme events for the surface conditions across the continent. January 2017 was distinct from the rest of the year with positive pressure anomalies over the continent and primarily negative pressure anomalies between 40° and 60°S (Fig. 6.3a) and slightly weaker-than- average circumpolar westerlies throughout the troposphere and lower stratosphere (Fig. 6.2c). The January circulation pattern is consistent with the negative phase of the southern annular mode (SAM) which continued from late 2016 [the SAM index from NOAA’s Climate Prediction Center (CPC) in January was −0.98], and it marks the transition of the late 2016 circulation to opposite sign anomalies in autumn 2017. At the surface, a strong high pressure anomaly was present over the South Pacific poleward to the Amundsen Sea, which through altered temperature advection and sea ice conditions produced negative temperature anomalies of ~1°C across the Antarctic Peninsula (Figs. 6.3a,b). These features weakened after January, and from February through May they were replaced by an anomalously deep Amundsen Sea Low centered over the northwest Amundsen Sea into the South Pacific (Fig. 6.3c; 6–9 hPa and 2–3 standard deviations below average). The anoma - lous cyclonic circulation, in conjunction with an anticyclone anomaly in the northwest Weddell Sea, produced well-above-average temperatures across much of West Antarctica during late summer and autumn spanning the western Antarctic Peninsula, ig . 6.3. (left) Surface pressure anomalies and (right) F Amundsen Sea Embayment, Marie Byrd Land, and 2-m temperature anomalies relative to each group’s Ross Ice Shelf (Fig. 6.3d; 2°–5°C and >3 standard 1981–2010 climatological average for (a) and (b) Jan deviations above average). Temperatures at the Byrd 2017; (c) and (d) Feb–May 2017; (e) and (f) Jun–Sep AWS in central West Antarctica were 2°–5°C above 2017; and (g) and (h) Oct–Dec 2017. Contour interval for surface pressure anomalies is 3 hPa and 1°C for average during February–May, and record maximum 2-m temperature anomalies. Shading depicts std. dev. monthly mean temperatures were observed on the of anomalies relative to the 1981–2010 average of each Antarctic Peninsula in March at both Marambio group. (Source: ERA-Interim reanalysis.) (−2°C) and Rothera (1.2°C; Fig. 6.4c); Dome C II AWS, on the East Antarctic plateau, reported record low monthly mean temperatures in March (−57.1°C, Fig. 6.4e), nearly 5°C below average. | S177 AUGUST 2018 STATE OF THE CLIMATE IN 2017

198 . 6.4. Monthly Antarctic climate anomalies during 2017 at six representative stations [three staffed (a)–(c), F ig and three automatic (d)–(f)]. Anomalies for temperature (°C) are shown in red and MSLP/surface pressure (hPa) in blue are shown, with filled circles denoting record anomalies for a given month at each station in 2017. All anomalies are based on differences from the monthly 1981–2010 averages. Observational data start in 1957 for Amundsen–Scott, 1959 for Casey, 1976 for Rothera, 1980 for Byrd AWS, and 1981 Dome C II and Ferrell AWS. During April (not shown), negative pressure across the Ross Ice Shelf, and Ferrell AWS reported anomalies developed over Wilkes Land and Queen a record monthly mean maximum temperature for Maud Land, and the associated northerly f low was an May of −23.1°C, 7.4°C warmer than the climatological average (Fig. 6.4f ). important contributor to the late advance of sea ice in the western Ross and eastern Weddell Seas; these During the winter months (June–September), a cyclonic circulation anomalies were part of a larger pronounced zonal wave-three pattern developed, shift in the Antarctic circulation as captured to some characterized by three anomalous ridges along 50°S centered at 90°E, 150°W, and 30°W and a deep low extent in the polar-cap-averaged geopotential height (Fig. 6.2a) and circumpolar zonal wind anomalies pressure anomaly over the Antarctic Peninsula - (Fig. 6.3e). Temperatures were generally below aver (Fig. 6.2c), both of which changed sign in April and intensified in May. At Casey Station, located in Wilkes age across the continent, especially on the eastern side of the midlatitude ridges/western side of the Land in coastal East Antarctica, pressure anomalies - - troughs, where southerly f low produced cool con were the lowest in April, consistent with this circu lation shift. On the eastern side of this low pressure ditions. Colder-than-average winter temperatures were also observed throughout the troposphere and system, the northerly f low increased temperatures | S178 AUGUST 2018

199 −1 stratosphere, accompanied by negative geopotential (10.6 m s ), and Dome C II had a record in October −1 (5.4 m s ). Relay Station tied its record low wind speed height anomalies and stronger-than-average circum- −1 polar westerlies in winter (Fig. 6.2); the stratospheric ). The record high wind speeds in November (5.5 m s vortex exhibited the greatest positive anomalies of ref lect the incidence of lower than normal pressure −1 above average during June. East Antarctica 4–6 m s for much of the year (Figs. 6.3, 6.4). experienced its most negative temperature anomalies )— E during 2017 in June (2°–6°C below average), with – c. Net precipitation (P D. H. Bromwich and S.-H. Wang Precipitation minus evaporation/sublimation Casey (Fig. 6.4b) and Dome C II AWS (Fig. 6.4e) – P ( both setting record low monthly mean temperatures ) closely approximates the surface mass balance E over Antarctica (e.g., Bromwich et al. 2011; Lenaerts in June (−19.4°C and −57.1°C, respectively). West Antarctica saw its strongest cold anomalies during and van den Broeke 2012), except for near-coastal July (2°–4°C below average) and September (4°–8°C areas where wind-driven transport of snow and meltwater runoff can become significant factors. below average; see Byrd AWS temperature anomalies in Fig. 6.4d). The colder-than-average temperatures Precipitation variability is the dominant term for P in September were partially due to the low pressure changes at regional and larger scales over the E – Antarctic continent. Precipitation and evaporation anomaly over the Antarctic Peninsula, ref lected in fields from the Japanese 55-year reanalysis (JRA-55; the record negative monthly mean pressure of 971.3 - Kobayashi et al. 2015) were examined to assess Ant hPa at Rothera (Fig 6.4c), more than 16 hPa below the E P arctic net precipitation ( – ) behavior for 2017. JRA- climatological average. - A positive temperature/geopotential height anom 55, the second generation of JRA, has incorporated aly developed in the stratosphere during September many improvements compared to its predecessor JRA-25 (Onogi et al. 2007; Bromwich et al. 2007). The and propagated downward into the lower troposphere - JRA-55 is used here because of these improvements during October (Fig. 6.2). Positive pressure and tem and its low latency, rather than ERA-Interim used perature anomalies developed at the surface across elsewhere. Because of the highly uneven distribution much of the continent in October, ref lected in the of E characteristics (from large Peninsula and observations at Amundsen Scott in Fig. 6.4a; the P – −1 to very strongest positive surface air temperature anomalies coastal West Antarctica values >1000 mm yr −1 during October (not shown) were over interior por - low values <50 mm yr in the high interior), only an - nual changes are shown in Fig. 6.5. tions of East Antarctica along the Transantarctic – E P Figure 6.5 shows the JRA-55 2017 and 2016 annual Mountains reaching 2°–4°C (> 3 standard deviations) – (Figs. 6.5a,b) and mean sea level E P anomalies of above average, and Vostok Station in the central East Antarctic plateau set a record high monthly mean pressure (MSLP; Figs. 6.5c,d) departures from the – P 1981–2010 average. In general, annual temperature in October of −51.1°C, 1.7°C higher anomalies E than the previous record set in 2015. Averaged over over the high interior of the continent were small −1 the October–December period (Figs. 6.3g,h), the (within ±50 mm yr ), and larger anomalies were strongest positive temperature anomalies were over observed along the coast, consistent with the low and high net precipitation accumulation in these Queen Maud Land, while the rest of the continent regions. From JRA-55, the 2016 positive anomalies experienced near-average to slightly-above-average located along the coast between Queen Maud Land temperatures and near-average pressure to close out 2017; exceptions include the Ross Ice Shelf where and Mac Robertson Land (between 5°W and 60°E) - below-average temperatures were observed during became weak negative anomalies in 2017, most pro nounced near 60°E. The weak negative anomalies November due to enhanced southerly f low from the development of an anomalous cyclone in the South over the American Highland (between 70° and Pacific that was consistent with the late austral spring 90°E) in 2016 became strongly positive in 2017. Both La Niña conditions (see Section 4b) and a record Queen Mary Land and Wilkes Land (between 90° and 125°E) remained strongly negative. The strong monthly mean low surface pressure value at the Byrd positive anomalies over Adélie Land and Victoria AWS in November (787.7 hPa). There were several record high monthly-mean Land (between 125° and 175°E) became near-zero in wind speeds recorded at various AWS during the 2017. The positive anomaly over the eastern Ross Ice year. Ferrell had record high wind speeds in May (9.7 Shelf in 2016 evolved into a larger positive anomaly −1 −1 −1 that extended into interior Antarctica in 2017. The ), and August (10.6 m s m s ), July (9.5 m s ) and largest positive anomalies that were located over Marble Point had a record high wind speed in March −1 (5.7 m s ). Byrd had a record high wind speed in May the Bellingshausen Sea and the southern Antarctic | S179 AUGUST 2018 STATE OF THE CLIMATE IN 2017

200 P F P − E ) and MSLP anomalies: (a) 2017 . 6.5. (a–d) Annual precipitation minus evaporation ( − E anomaly (mm); ig (b) 2016 P − E anomaly (mm). Antarctic regions with > ±30% departure from the reference mean are hatched; vertical denotes negative anomaly and horizontal is positive. (c) 2017 MSLP anomaly (hPa); and (d) 2016 MSLP anomaly (hPa). All anomalies are calculated with respect to the 1981–2010 means. (e) Monthly total P − E (mm; dashed green) for part of West Antarctica bounded by 75°–90°S, 120°W–180°, along with index trends for EQ- SOI (dashed blue, from NOAA CPC) and SAM (dashed red, from Marshall 2003). Centered annual running means are plotted as solid lines. | S180 AUGUST 2018

201 Peninsula (between 110° and 70°W) in 2016 became the conventional station-based SOI, EQ-SOI is less the second largest negative anomalies in 2017. Similar susceptible to weather noise and better captures the equatorial trade wind events (see www.climate.gov to 2016, the two sides of the Antarctic Peninsula had /news-features/blogs/enso/why-are-there-so-many- opposite anomalies but with a reversal of signs in enso-indexes-instead-just-one). 2017: positive in the east, negative to the west. The The EQ-SOI and SAM were in phase (same sign) Ronne Ice Shelf anomalies remained negative but P in most months were weaker during 2017. E – but have opposite behavior to anomaly features are gener - from 2010 to mid-2011 (Fig. 6.5e). From then on, P These annual – E ally consistent with the mean annual atmospheric EQ-SOI and SAM were out of phase (opposite sign) circulation implied by the MSLP anomalies (Figs. through early 2016. Both EQ-SOI and SAM were offsetting factors modulating precipitation, result - 6.5c,d). In 2017 (Fig. 6.5c), the MSLP annual anoma - lies surrounding Antarctica were less regionalized E ing in little overall change of P – . From late 2016 to early 2017, the MSLP anomalies over the Ross Sea and were weaker than in 2016 (Fig. 6.5d) with strong - shifted from positive (December–February) to nega seasonal variation during 2017 (e.g., Fig. 6.3). The tive (March–May; e.g., Figs. 6.3a,c). A combination of largest positive anomaly center in 2016 over the Drake a weak La Niña pattern and a negative SAM resulted Passage (~75°W) became the largest negative anomaly amounts in this region during early P in 2017 and extended into the Weddell Sea as the in higher – E 2017. As the seasons progressed into late 2017, both seasons progressed through the year, with a peak in ENSO and SAM indices strengthened and became in September–November (SON; e.g., Figs. 6.3 e,g). The began to show signs of a observed negative anomaly centers over the South phase with each other. P – E large decrease in late 2017. Pacific (160°~110°W) and Indian Ocean (105°~165°E) strengthened in the first half of 2017 (e.g., Figs. 6.3a,c). d. Seasonal melt extent and duration— Both anomalies expanded and changed to positive L. Wang and H. Liu values that covered nearly two-thirds of the Southern Surface melt contributes to accelerated iceberg - calving, hence the retreat of ice margins and conti Ocean (between 15°E and 90°W) later in the year nental ice mass loss (Scambos et al. 2013; Rignot et al. (SON; e.g., Fig. 6.3e). These seasonal MSLP changes 2004). The intensity, duration, and spatial extent of resulted in negative–positive–negative anomaly cen - surface melt (Tedesco et al. 2013) contribute directly ters along the East Antarctic coast in the Southern to the enlargement of ice crevasses (Scambos et al. Ocean (Figs. 6.3e, 6.5c). Two secondary negative 2000), accelerated glacier ice f low (Zwally et al. 2002), MSLP anomaly centers located over coastal Ross Sea and disintegration of buttressing ice tongues and ice (between 165°E and 160°W) and the Amery Ice Shelf shelves (van den Broeke 2005; Massom et al. 2018). (~65°E) produced stronger onshore wind f lows and Surface melt on the Antarctic continent during P resulted in greater than 30% higher – E in the inte - the 2016/17 austral summer season was estimated rior of the Antarctic continent (Fig. 6.5a). By contrast, – E anomalies (~120°E and ~80°W; from daily passive microwave brightness temperature P two negative Fig. 6.5a) were associated with strong offshore f low data at the 19 GHz frequency acquired by the Special in 2017 (Fig. 6.5c). Sensor Microwave Imager/Sounder (SSM/IS) onboard Earlier studies (e.g., Cullather et al. 1998) show the Defense Meteorological Satellite Program (DMSP) that almost half of the moisture transport into inte satellite in the ascending passes. The data were - F17 preprocessed and provided by the U.S. National Snow rior Antarctica occurs in the West Antarctic sector. and Ice Data Center (NSIDC) at level-3 EASE-Grid Antarctic moisture transport has large interannual format (Armstrong et al. 1994) and were analyzed us - variability, often associated with variations of ENSO ing a wavelet transform-based edge detection method (e.g., Bromwich et al. 2004) and the southern annular (Ho. Liu et al. 2005). The algorithm delineates each mode (SAM; e.g., Fogt et al. 2011). Figure 6.5e shows melt event in the time series by tracking its onset and the time series, with 12-month running means, of E over Marie Byrd Land–Ross Ice end dates, with the onset day of the first melt event monthly total P – - Shelf (75°–90°S, 120°W–180°) and the monthly equa being the start day of the melt season (Fig. 6.6a) and torial Southern Oscillation index (EQ-SOI) and SAM the end day of the last melt event being the end day of indices. The NOAA CPC EQ-SOI is a standardized the melt season (Fig. 6.6b). The melt duration is then sea level pressure difference centered on the equator the total number of melting days per pixel during the defined melt season (excluding any refreezing events between the east Pacific (5°N–5°S, 80°–130°W) and Indonesia (the west Pacific; 5°N–5°S, 90°–140°E) and that may have occurred during this period; Fig. 6.6c). is negative during warm events. In comparison to The melt extent and melt index are metrics useful for | S181 AUGUST 2018 STATE OF THE CLIMATE IN 2017

202 . 6.6. Estimated surface melt for the 2016/17 austral summer: (a) melt start day, (b) melt end day, (c) melt ig F duration (days), and (d) melt duration anomalies (days). quantifying the interannual variability in surface melt events on the Wilkins Ice Shelf extended to late melt (Zwally and Fiegles 1994; Liu et al. 2006). Melt March 2017 (Fig. 6.6b). Figure 6.6c shows the melt 2 duration in the austral summer of 2016/17 (Fig. 6.6c). ) is the total area that experienced surface extent (km melt for at least one day during the melt season. Melt Areas with intensive melt (> 45 day duration in or - 2 index (day. km ) is the sum of the duration (days) of ange–red) were the Larsen, Wilkins, and Shackleton ice shelves, and some coastal areas of Wilkes Land the melt pixels in the study area that describes the strength of melt as the accumulative melt days in a and Queen Maud Land. The Shackleton Ice Shelf had year. The anomaly map (Fig. 6.6d) was created by an abnormal prolonged melt season this year, which referencing to the mean melt duration computed over could have been related to the higher-than-average the 1981–2010 period (see also Fig. 3 in Liu et al. 2006). temperature in November and record-breaking low monthly mean pressure recorded at the nearby Casey Figure 6.6a shows that the earliest melt events - Station (Keller et al. 2017). Areas with moderate in occurred on the Wilkins Ice Shelf during the austral summer of 2016/17. The early melt area also extended tensity of melt (16–45 day duration in green–yellow) included coastal Queen Maud Land and the Amery to the tip of the Antarctic Peninsula. Some late but Ice Shelf; short-term melt (< 16 day duration in blues) short melt events occurred on the Ross Ice Shelf. The | S182 AUGUST 2018

203 interface within the high southern latitudes (Bourassa et al. 2013). It also acts as a buffer for ice shelves against ocean processes (Williams and Squire 2007; Massom et al. 2018). Net sea ice extent (SIE; the area enclosed by the ice edge consisting of a range in sea ice concentration) and sea ice area (SIA; the actual area covered by sea ice) were well below the 1981–2010 average for all of 2017 (Fig. 6.8b). Following the record low seasonal sea ice cover in November and December 2016 (Reid et al. 2017; Stuecker et al. 2017; Turner et al. 2017; Schlosser et al. 2018), the first four months of 2017 6 2 also had record low net SIE, followed by sporadic ig F . 6.7. (a) Melt index (× 10 d a y· k m ) from 1978/79 to 1 − 2 periods of record low SIE into September. Overall, 2016/17, showing a negative trend (265 800 day· km yr , 2 6 not significant at 95%). (b) Melt extent (× 10 ) from km 130 days of record low SIE occurred during 2017, 1978/79 to 2016/17, showing a negative trend (10 2 00 with 57 individual days of record low SIA between 1 2 − km yr , p < 0.05%). The year on the x-axis corresponds - January and early October. The month of Febru to the start of the austral summer melt season, that ary 2017 recorded the lowest monthly mean SIE on is, 2008 corresponds to summer 2008/09. record (Schlosser et al. 2018). Other records in 2017 included the lowest observed daily value of SIE in the occurred on the Ross Ice Shelf and small portions of continuous satellite record (since 1978) on 1 March coastal Queen Maud Land. Almost half of the Ross Ice 6 2 (not shown; previous lowest was 2017 of 2.1 × 10 Shelf experienced melt, albeit brief ly, in the summer km 6 2 of 2016/17. Compared to the previous year, melt on the on 27 February 1997). The annual daily 2.3 × 10 km maximum was also later than previously observed, on Ross Ice Shelf was less extensive. Overall, the 2016/17 9 October (previous latest maximum was 3 October melt season was slightly longer than the historical 1988). Record low sea ice cover during 2016/17 is in average (Fig. 6.6d), indicating an above-average melt contrast to the long-term (1979–2016) positive trend year for Antarctica. in net SIE (Turner and Comiso 2017), as discussed > 0.05) p Figure 6.7a shows a non-significant ( −1 2 further below. yr negative trend (265 800 day·km ) in melt index Regionally, early 2017 (January through mid- since 1978, highlighted by the record low melt season observed during austral summer 2008/09. The trend - April) sea ice coverage followed on from the predomi lines were fit using a linear regression between the nantly low net sea ice coverage in late 2016. However, melt indices and year number. Before adding 2017 to high concentrations of sea ice were observed along the regression, the negative trend in melt index was much of the coast of East Antarctica (~80°–160°E) significant ( p and in parts of the Weddell Sea (~30°–60°W), for < 0.05; Wang and Liu 2017). The marked example (Fig. 6.8c). Synoptically, in late 2016 and increase in melt index for 2017 was due to the inten - January 2017, winds around East Antarctica and the sive melt (> 90 days) on the Wilkins Ice Shelf. The Weddell Sea were anomalously easterly, causing an negative trend of the melt extent, however, remained p < 0.05; Fig. 6.7b), because half of the initial southward compaction of the sea ice via Ekman significant ( transport while largely retaining the sea ice extent in Ross Ice Shelf did not melt (Fig. 6.6c) as it did in 2016, which reduced the 2017 melt extent as compared to these regions during the summer season (Figs. 6.8a,c). 2016 (Fig. 6.7b). Nonetheless, both the melt extent and Consequently, sea ice advanced early across much of melt index were the second highest since 2005. The East Antarctica. Elsewhere around the coast in early 2017, sea ice coverage was either non-existent or well negative trends are consistent with previous reports below average (e.g., Fig. 6.8c). Ocean SSTs around (Liu et al. 2006; Tedesco 2009; Tedesco et al.2009). Antarctica through early 2017 were anomalously high e. Sea ice extent, concentration, and seasonality— (e.g., Fig. 6.8c; Section 6f) and coincident with regions P. Reid, S. Stammerjohn, R. A. Massom, J. L. Lieser, S. Barreira, of suppressed ice formation, particularly in the Ross, Bellingshausen, and Amundsen Seas and the eastern and T. Scambos Weddell Sea. The suppressed ice formation led to Antarctic sea ice performs important roles in considerably later ice edge advance in these regions, the climate system through the formation of dense by as much as 50 days later in most of the Amundsen oxygen rich Antarctic Bottom Water (Johnson 2008) and modulating f luxes across the ocean/atmosphere | S183 AUGUST 2018 STATE OF THE CLIMATE IN 2017

204 - (~70°E), promoting sea ice ad vance (albeit later than normal) in the Ross (~120°W–180°) and Weddell (~10°–30°W) Seas. Through May and into early June, SIE continued to be above average across much of East Antarctica but below average - in the western Ross, Amund sen, and Bellingshausen Seas and across the eastern Weddell Sea and Indian Ocean sectors (~10°W–80°E). During mid-June, a station - ary wave-three atmospheric - pattern began to develop (Sec tion 6b), with broad low-pres - sure centers to the north of the Bellingshausen Sea (~80°W), East Antarctica (~140°E), and Dronning Maud Land (~40°E) that broadly correspond to the SIE anomalies (Fig. 6.8a). This pattern increased southerly cold air outf low in the eastern Ross Sea, far eastern Wed - dell Sea, and north of Wilkes Land (~120°E), increasing ice coverage and contributing to a positive anomaly in SIE in these ig F . 6.8. (a) Hovmöller plot of daily SH sea ice extent anomalies for 2017 regions (Fig. 6.8a). Conversely, 3 2 (× 10 km per degree of longitude; from the 1981–2010 mean); (b) net sea intervening warm air advection ice extent anomaly (blue) and sea ice area anomaly (red) (from 1981–2010 and higher SSTs associated with mean); thin blue lines represent the historical daily values of extent for 1979–2015, while the thin black lines represents ±2 std. dev. of extent. (c) this zonal wave-three pattern and (d) sea ice concentration anomaly (%) and SST anomaly (°C; Reynolds et were observed in the western al. 2002; Smith et al. 2008) for (c) Feb. and (d) Sep. 2017. Based on satellite Weddell and Ross Seas and passive-microwave ice concentration data [Cavalieri et al. 1996, updated to the north of the Amery Ice yearly, for climatology; and Maslanik and Stroeve (1999) for the 2017 sea Shelf (~60°–100°E), suppressing ice concentrations]. ice expansion and producing a negative SIE anomaly (Fig. 6.8a). Interestingly, Sea, as ref lected in the negative duration anomaly in - this region (Fig. 6.9a). while the atmospheric zonal wave-three pattern sub The atmospheric circulation pattern changed sided during September, the zonal wave-three pattern during April with deep low pressure systems de - within both the patterns of SIE and SST anomalies veloping north of the Weddell Sea (~30°W), Wilkes persisted through early November (Figs. 6.8a,d). It Land (~125°E), and well north of the Amundsen Sea was during this period (September onwards; e.g., Fig. (~100°W; see Section 6b). This pattern enhanced ice 6.8d) that, again similar to 2016 (Mazloff et al. 2017; coverage predominantly within the western Weddell Reid et al. 2017), the Maud Rise polynya opened up - Sea while continuing to suppress extent in the Bell (see Sidebar 6.1). ingshausen, Amundsen, and Ross Seas through warm - Early November saw another change in the cir air advection and higher-than-normal SSTs. A zonal cumpolar atmospheric circulation pattern, with the development of a deep Amundsen Sea low pressure wave-two atmospheric pattern developed in May (not shown), with synoptic lows centered in the eastern system and an associated zonal wave-three pattern. Ross Sea (~140°W) and north of the Amery Ice Shelf The change in atmospheric circulation inf luenced | S184 AUGUST 2018

205 Sea and the Weddell Sea through the end of March. f. Southern Ocean— S. Swart, K. Johnson, M. R. Mazloff, A. Meijers, M. P. Meredith, L. Newman, and J.-B. Sallée In the climate system, the South - ern Ocean is disproportionately im - portant when it comes to its storage of heat and carbon. Modification of the upper Southern Ocean could have significant implications for the rate and magnitude of air–sea f luxes (of heat and carbon) and for the ventilation of the ocean interior, thereby altering the effects of climate warming on the ocean - system as a whole. Here, we evalu ig . 6.9. Maps showing (a) duration anomaly for the 2016/17 sea ice season F 1 − ate the state of the Southern Ocean in days and (b) duration trend for 1979/80–2016/17 in days yr . The black in 2017 by first assessing the upper < 0.01 contour in (b) delineates those trends with significance at the p significance level. ocean as the interface between the atmosphere and ocean interior. We then discuss the changes in intermediate to deep the regional rate of ice retreat, particularly in the - water masses, which are critical pathways to moving Bellingshausen–Amundsen (60°–120°W) and Wed heat and carbon to the ocean interior where it will dell (30°W–30°E) Seas where slower and faster retreat (in general) remain for decades to centuries. Lastly, occurred, respectively (Fig. 6.8a). we report on the status of ocean acidification in the These austral springtime sea ice distribution Southern Ocean using newly available biogeochemi - changes are consistent with the inf luence of the rela - cal observations. tively weak La Niña developing within the tropical Pacific in early November 2017 (see Section 4b), which changed the position of the higher-latitude southern 1) u pper ocean By utilizing all available 2017 hydrographic profiles - jet streams and hence the cyclonicity around the Ant (40 16 from Argo f loats and 11 16 from tagged seals), 9 arctic continental edge (Yuan 2004; Stammerjohn et 8 al. 2008). Thus, SIE towards the end of the year and anomalies of mixed layer depth (MLD) and mixed within the Weddell Sea was, in some areas, more than layer (ML) temperature and salinity (Figs. 6.10a–c) were computed from the climatological (2000–2010) six standard deviations below average. Elsewhere, seasonal cycle (see Pellichero et al. 2017). During 2017, SIE was close to average, although small pockets of greater-than-average SIE existed within the Amund - the most significant observation is the shallower MLDs - (negative anomalies) occurring in the Pacific sector, sen Sea and western Pacific sector (~110°–150°E) as sociated with wind-driven compaction of the sea ice - particularly within the Antarctic Circumpolar Cur - rent (ACC), where MLDs are more than 100 m shal cover and lower-than-normal SSTs near these regions. - lower than the climatology. Meanwhile, the Atlantic The long-term trend for Antarctic sea ice is re and Indian sectors are characterized by mixed MLD gionally and seasonally variable: increased SIE and longer seasonal duration within the Ross and Weddell anomalies. Seas, and decreased SIE and shorter duration in the In contrast to anomalies reported for 2015 and Bellingshausen–Amundsen Seas (e.g., Fig. 6.9b for sea 2016 (Sallée et al. 2016; Mazloff et al. 2017), a markedly warmer ML (Fig. 6.10b) was observed throughout most ice duration trends over 1979/80–2016/17; see Comiso et al. 2017 for sea ice extent trends). For SIE these of the Southern Ocean in 2017, except for the northern subantarctic region of the Atlantic sector. In further changes are largest during January–May (Hobbs et al. contrast, positive Southern Ocean ML temperature 2016). Apart from some areas of the Amundsen Sea, the regional pattern of sea ice coverage during 2017, anomalies in 2015 were juxtaposed against negative - anomalies to the north, indicating a north–south di described above, was in contrast to this long-term pole separated by the ACC (Sallée et al. 2016), while in trend (Fig. 6.9a), particularly in much of the Ross | S185 AUGUST 2018 STATE OF THE CLIMATE IN 2017

206 F ig . 6.10. (a) Mixed layer depth anomaly (m) in 2017 from the climatological seasonal cycle. The thin black contours represent the main ACC fronts from north to south: northern Subantarctic Front (SAF), main SAF, Polar Front (PF). The thick black contour is the Sep. climatological sea ice extent. (b) Same as (a) but for mixed layer temperature (°C). (c) Same as (a) but for mixed layer absolute salinity. (d) Circumpolar average trend 1 − in potential temperature (in °C yr ) from Argo float data (seasonal cycle removed), oriented along constant dynamic height and isopycnal surfaces. The thick solid black contour is the T-min layer and the white contour is the S-min layer. Constant pressure surfaces are indicated. Vertical lines indicate the position of the PF (left) and SAF (right). Dots indicate trends in potential temperature significant at the 95% level. (e) Observed upper 30-m pH (black outlined colored dots) from the GLODAPv2 database (Key et al. 2015) spanning 1992–2013 and the SOCCOM pH observations in 2017 (colored dots without outlines). The 3000-m bathymetry contour is shown in black. (f) Changes in pH via two different methods: black diamonds are annual averaged GLODAPv2 comparisons (1992–2013) to all SOCCOM float data (2014–17), while red/blue pluses denote the discrete GLO - DAPv2 comparisons to only 2017 float data and all other float data, respectively (based on criteria explained in the text). The black and red lines are the linear fit to the respective color markers. | S186 AUGUST 2018

207 −3 2016, ML temperature anomalies defined a quadrupole 27.6–28 kg m ) of the densest Antarctic Intermediate delimited by ocean basins (Mazloff et al. 2017). Water (AAIW) and Circumpolar Deep Water (CDW) In 2017 warmer ML temperatures north of the throughout the Southern Ocean over 2002–16. This may be related to changes in westerly winds (due to mean September ice edge (thick black contour in Figs. long-term increases in the SAM; Böning et al. 2008). 6.10a-c) generally coincided with shallower MLDs (negative anomalies; Fig. 6.10a) and increased ocean Lying above these depths, two fairly distinct negative stratification (not shown). However, deeper MLDs trends were observed, one north of the Polar Front (positive anomalies) appearing in the Indian sector of - (> 0.8 dyn m), which indicates cooling of the up per AAIW and Subantarctic Mode Water (SAMW; °E near the mean the Southern Ocean (centered on 60 −1 −0.01°C y r September ice edge) coincided with anomalously ), and the other south of the Polar Front, - which indicates relatively strong cooling and fresh saline mixed layers (Fig. 6.10c) and reduced ocean stratification (not shown). South of the ACC (and ening of winter and surface water (e.g., Haumann −1 mean September ice edge) there are fewer observa . et al. 2016) ranging from −0.015° to −0.05°C yr - The cooling of SAMW in the northern ACC does tions, but overall the data suggest a negative MLD not contradict the general warming trend observed anomaly (shallower) in the West Antarctic sector beyond the northern ACC, as this may be related to together with strongly negative ML salinity (fresher) increased volume and hence heat content of SAMW and positive (higher) ML temperature anomalies. These fresher MLs may be linked with a long-term and AAIW predominantly caused by wind-driven increase in wind-driven transport of freshwater changes, namely increased wind stress curl (Gao et al. 2018). northward (Haumann et al. 2016) and/or increased sea ice melt in summer (February) 2017 (Section ontin u e d : c Geoche mic al io 6e). From the Maud Rise region (~3 3) statu s °E) towards East oce an b acidification Antarctica positive ML salinity anomalies (Fig. 6.10c) The new Southern Ocean Carbon and Climate were observed, which resulted in weak stratification. These changes may be linked to the anomalously low - Observations and Modeling project (SOCCOM) ar ray currently has 105 active profiling f loats, allowing sea ice conditions experienced in the eastern Weddell Sea (Section 6e) together with the recent re-emergence a characterization of ocean pH variability at shorter of the Maud Rise polynya and its associated impacts time scales and higher spatial resolution. Ocean pH on the upper ocean via enhanced air–sea exchanges enters the ocean is decreasing as anthropogenic CO 2 and ventilation of warmer, saltier interior water and forms carbonic acid (causing ocean acidification). (see Sidebar 6.1). Ocean acidification challenges the viability of organ masses - shells and has fundamental isms producing CaCO 3 ntermediate 2) i impacts on the ocean carbon cycle. ocean Observed pH in the upper 30 m from the GLO - Significant thermohaline changes are occurring DAPv2 database (1992–2013 observations as black below the surface layers of the Southern Ocean. Due outlined colored dots in Fig. 6.10e; Key et al. 2015; to the slow time scales of these changes (unlike the Olsen et al. 2016) is compared to SOCCOM pH - more temporally sensitive surface mixed layer proper observations in 2017 (colored dots without black out - ties just described), it is more appropriate to discuss multi-year changes (2002–16). The gravest empirical lines). It is qualitatively apparent that the GLODAPv2 estimates have higher pH. Two methods were used to mode (GEM; see methods in Meijers et al. 2011; Swart et al. 2010), a highly effective method to reconstruct quantify these differences. For the first method, all upper 150-m GLODAPv2 observations from 1992 to subsurface property fields from sea surface height, is used to map thermal changes at intermediate depths 2013 were used, and f loat observations (from 2014 to 2017) that were within 20-km and 5-m depth of the (thermocline depth to 2000 m). The GEM uses a current inventory of all Argo f loat profiles (2002 GLODAPv2 observations were identified. Differences to 2016) to derive circumpolar-averaged potential in pH between GLODAPv2 and f loat observations were determined and bin-averaged for each year temperature trends approximately oriented along stream-following dynamic height contours, here of GLODAPv2 observations (black diamonds in shown on constant isopycnal surfaces (Fig. 6.10d). Fig. 6.10f); any year with less than five matches was eliminated. (The additional data point for 2016 shows Such a coordinate choice removes aliasing and trends due to frontal movements and vertical heave. the mean offset between f loat pH and hydrocast pH, the latter acquired during f loat deployments; Overall, there was a consistent warming of up to −1 (Fig. 6.10d) and salinification (not shown: 0.02°C yr Johnson et al. 2017). A second method identified all | S187 AUGUST 2018 STATE OF THE CLIMATE IN 2017

208 SIDEBAR 6.1: RETURN OF THE MAUD RISE POLYNYA: CLIMATE S. SWART, E. C. CAMPBELL, C. H. HEUZÉ, K. JOHNSON, — LITMUS OR SEA ICE ANOMALY? J. L. LIESER, R. MASSOM, M. MAZLOFF, M. MEREDITH, P. REID, J.-B. SALLÉE, AND S. STAMMERJOHN The Maud Rise polynya is a persistent area of open water within the sea ice cover of the Southern Ocean, which overlies an area of elevated topography called Maud Rise (66 °S, 3 °E) located in the eastern sector of the Weddell Sea (Fig. SB6.1a). It is termed a “Weddell polynya” if it grows and migrates west - ward into the central Weddell Sea. This larger sized polynya was first observed in satellite data in 1974 and recurred for each of the two subsequent austral winters (Zwally and Gloersen 2 00 km , meant that 1977; Carsey 1980). Its large size, ~300 0 it could contribute strongly to the transfer of heat from the ocean to the atmosphere in winter and, hence, instigate dense water production and the renewal of deep ocean waters in the Weddell Sea (Gordon 1978). The amount of deep water formed via this route was estimated at 1–3 Sverdrups (Martinson et al. 1981). The 1974–76 polynya may have been responsible for up to 34% of observed warming of the deep Southern Ocean (Zanowski et al. 2015). Smaller features, perhaps associated with topographically driven upwelling of warm waters, have been observed subsequently (Comiso and Gordon 1987), but a large polynya had not re-appeared until recently and unex - pectedly during austral winters 2016 and 2017. Following the Maud Rise polynya development in 2016 (Mazloff et al. 2017), mid-September 2017 saw the opening of a longer lived and larger polynya over the same region. The 2017 polynya grew quickly but its size remained quite static 2 0 00 km at approximately 50 until 3 November, after which it more than tripled in size over a period of a week. The sudden expansion is possibly the result of a considerable change in - . SB6.1. (a) Circumpolar map of AMSR2 sea ice con ig F atmospheric circulation due to the development of a La Niña centration (in %) on 8 Nov 2017, with the red shading in early November (Section 6e), combined with an anomalously marking polynya locations, including the largest—the earlier spring ice edge retreat (see Section 6e). The polynya Maud Rise polynya. (b) Location of the polynya on 14 continued to expand over the following month (Fig. SB6.1b) Oct 2017 from AMSR2 sea ice concentration (Spreen et 2 and reached its maximum size of 295 (larger than New 0 00 km al. 2008). The black line represents the polynya size on Zealand) on 2 December 2017 before coalescing with the open - 29 Nov 2017, at its largest extent just prior to coalesc ocean. Overall, it contributed to a significantly large negative ing with the open ocean. The yellow and cyan stars anomaly in sea ice concentration (see Section 6e). represent the location of the SOCCOM floats 5904471 and 5904468, respectively. The magenta contour shows Two under-ice biogeochemical profiling floats from the a 20-yr mean location of the polynya as depicted in the - Southern Ocean Carbon and Climate Observations and Model MPI-ESM-LR model. - ing (SOCCOM) project were present at Maud Rise before, dur ing, and after the 2016 and 2017 polynyas. Both floats surfaced and transmitted data within the 2017 polynya (Fig. SB6.1b). mixing during the 2017 polynya event. Additionally, enhanced These data show the appearance of cold and fresh subsurface biogeochemical responses to the polynya’s presence were anomalies in late 2016 (extending from ~100 to 300 m depth observed with approximately a 2-month earlier (September in Figs. SB6.2a,b), indicating that deep ventilation may have oc - 2017) increase in chlorophyll fluorescence (phytoplankton) and curred during the brief 2016 polynya. This modified subsurface pH (Figs. SB6.2c,d) compared to the two previous years, which water mass persisted into 2017 and was punctuated in October were ice covered. Hydrographic measurements collected near and November by warm and salty intrusions indicative of deep S.A. Maud Rise during two research expeditions on the R/V | S188 AUGUST 2018

209 SAM (Gordon et al. 2007), and/or reduced sea ice concentra - tion and upper-ocean instability from upwelling of warm and salty waters on the flanks of Maud Rise (Gordon and Huber 1995; Lindsay et al. 2004; de Steur et al. 2007; Cheon et al. 2014, 2015). Triggering mechanisms remain less clear but may include transient eddies or other topography–mean flow in - teractions associated with Maud Rise (Holland 2001) or small positive salinity anomalies at the surface caused by anomalous wind and/or sea ice conditions (Cheon et al. 2014; Heuzé et al. 2015; Kjellsson et al. 2015). A prolonged period of strong westerly winds (coincident with positive SAM) might also ex - plain the 2016 and 2017 openings that may have responded to the wind-induced Ekman transport and associated upwelling of warmer water (Cheon et al. 2014; Ferreira et al. 2015). In the lead-up to the 2016 and 2017 polynyas, the SAM index was indeed strongly positive with three of its ten highest monthly values since 1957 recorded in 2015 and 2016, including the largest value in February 2015—coinciding with the annual sea ice minimum. It is quite possible that strong winds and an associated enhanced Weddell Gyre were the catalyst for these polynya events. A contributing mechanism during both years may be anomalously warm waters advecting south from the Indian and western Atlantic sectors of the Southern Ocean. More research is needed to better understand the respective roles of large-scale modes (SAM) versus regional circulation anomalies, in addition to needing more highly resolved data in space and time (e.g., Schlosser et al. 2018). Global coupled models generally exhibit a greater frequency . SB6.2. Sections of (a) potential temperature (°C), F ig of Maud Rise polynya occurrence compared to observations (b) salinity (PSS-78) from SOCCOM float 5904471, (c) (e.g., Heuzé et al. 2013; Fig. SB6.1b) and have thus been a − 3 chlorophyll- (mg m a ), and (d) pH from SOCCOM float valuable source of information regarding their causes and oc - 5904468, from within the polynya over 3 years. Gray currences. Models suggest a preconditioning is needed by the dashed lines represent the start and end dates of the slow accumulation of subsurface heat over several decades 2017 polynya. Gray shading indicates absence of data. (Martin et al. 2013; Dufour et al. 2017), heat that would be lost in December 2017 and January Agulhas II Polarstern and R/V - after years of the polynya remaining open, possibly explain 2018, respectively, when fully processed and analyzed, may ing why polynyas on the scale of the 1974–76 event have not lend additional insights regarding the ocean impacts from the been seen in 40 years. Alternatively, models also suggest that 2016 and 2017 polynyas. increased freshening at the ocean surface, caused by increased The research community continues to speculate on the ice sheet/iceberg melt for example, may increase stratification causes of the 2017 polynya and whether it is related to the and reduce the frequency of polynya formation (Kjellsson et al. 2016 event. It is possible that the 2017 polynya was caused by 2015). The extent to which such models robustly reproduce - persistent subsurface ocean conditions that were initiated dur the real ocean is largely unknown due to the comparatively ing the 2016 polynya, and/or it was caused by preconditioning short observational record, but such results highlight the need that resulted from anomalous sea ice divergence occurring late to better understand this intermittent but important mode of spring 2016 (Schlosser et al. 2018). Preconditioning mecha - deep ocean ventilation. nisms may include a build-up of subsurface heat (Martin et al. 2013), a precipitation deficit caused by prolonged negative | S189 AUGUST 2018 STATE OF THE CLIMATE IN 2017

210 f loat observations in the upper 30 m at intervals of a dip due to stratospheric warming in mid-August. 5 days, 5-m depth, 3° latitude, and 10° longitude of It increased until another stratospheric warming stopped the expansion in mid-September, reaching the GLODAPv2 1992–2013 observations (Fig. 6.10f, 2 peak on 11 September, and then red and blue pluses for 2017 and 2014–16 f loat data a 19.6 million km comparisons, respectively). declined slowly into October and disappeared on 19 - The two estimates reveal consistent trends (deter November. The warmings prevented the hole from - mined by least-squares fit) in pH, hence ocean acidifi cation. The first approach resulted in an acidification −1 ; for the second method the rate rate of −0.0023 yr −1 −1 for the 2014–16 and was −0.0025 yr and −0.0028 yr 2017 f loat data comparisons, respectively, implying either the 2017 f loats sampled lower pH (than was - sampled in 2014 to 2016) or the 2017 f loats cap tured different spatial variability. These results are nevertheless consistent with previous observations based on individual hydrographic lines (Rios et al. 2015; Williams et al. 2015) or based on predictions from coupled models. Faster acidification rates in the Southern Ocean compared to the global average −1 ) expected due to low carbonate ion (~ −0.0017 year concentrations in the Southern Ocean (McNeil and Matear 2008; Orr et al. 2005). Nonetheless, there is considerable spatial and temporal variability in surface ocean pH, both in observed and as predicted with coupled climate models (Russell et al. 2018), but as the f loat record expands and lengthens, both the assessment and prediction of the spatial and temporal variability in acidification rates will improve. g. 2017 Antarctic ozone hole— N. Kramarova, P. A. Newman, E. R. Nash, S. E. Strahan, C. S. Long, B. Johnson, M. L. Santee, I. Petropavlovskikh, G. O. Braathen, and L. Coy Severe ozone depletion in the Antarctic strato - sphere has been observed every austral spring since - the early 1980s (WMO 2014) and is caused by het erogeneous chemical reactions with human-made chlorine- and bromine-containing compounds. As much as 98% of the ozone in the lower stratosphere F ig . 6.11. Antarctic vortex-averaged concentrations around 70 hPa is destroyed in September–October. MLS (updated Aura of: (a) ClO and (b) ozone from As a result of regulations set in place by the Montreal from Manney et al. 2011). These MLS averages are Protocol and its amendments, levels of chlorine from made inside the polar vortex on the 440-K isentropic ozone depleting substances have gradually declined, surface (~18 km or 65 hPa). (c) Temperature on the and springtime Antarctic ozone is beginning to show 440-K isentropic surface over Antarctica (60°–90°S) signs of recovery (WMO 2014). from MERRA-2 (Gelaro et al. 2017). (d) Ozone hole The 2017 Antarctic ozone hole was the second area based on Ozone Monitoring Instrument (OMI) 2 smallest since 1988, with an area of 17.4 million km and Ozone Mapping and Profiler Suite (OMPS) satel - 2 lite observations. Four years are shown: 2010 (orange), (or 6.7 million mi ) averaged from 7 September to 2012 (pink), 2015 (blue), and 2017 (black). The white 13 October, the period of greatest ozone depletion. line shows the daily average and the gray shading shows The ozone hole area is defined as the area with total the daily ranges for 2005–16. The vertical solid lines column ozone values less than 220 Dobson units indicate the averaging period for Fig. 6.13, while the (DU). Figure 6.11d displays the daily areal coverage vertical dashed lines indicate the dates of maximum of the ozone hole for 2017 (black curve). The area wave forcing for the stratospheric warming events started expanding at the beginning of August, with in 2017. | S190 AUGUST 2018

211 In July 2010, a wave event warmed the stratosphere, followed by little wave ac - tivity until a mild event early in September of that year. The 2010 ozone hole slowed its areal growth but continued to develop in a normal manner with reduced values. In 2012, the development of the ozone hole proceeded as in 2017, but large wave events happened in late September into October. In contrast to these ac - tive years, 2015 was a year with little planetary wave activity and consistently lower-than-average tem - peratures throughout the austral winter and spring. Consequently, 2015 had . 6.12. Altitude vs. time cross sections from balloon observations at South Pole F ig - severe Antarctic ozone de station in 2017 for: (a) temperature profiles (°C), (b) ozone profiles (ppmv), and pletion and a larger ozone anomalies from the 2005–16 average normalized by std. dev. of (c) temperature hole area. and (d) ozone. Balloon ozone and temperature observations at South Pole station (Fig. growing further, accounting for the low average area 6.12) revealed the record high temperatures in the compared to previous years. stratosphere above ~15 km in August and September, The extent of the seasonal ozone depletion over soon after the wave events. These temperatures were - Antarctica is controlled by the total inorganic chlo 2–4 standard deviations higher than the average rine and meteorological conditions in the lower seasonal values derived from the balloon observations stratosphere. Colder temperatures facilitate formation of the polar stratospheric clouds (PSC) and transfor - over the period 2005–16 (Fig. 6.12c). The above- mation of inorganic chlorine to active chlorine that average stratospheric temperatures over the South Pole led to weak ozone depletion this year (Fig. 6.12d). eventually lead to ozone loss. There were two key planetary wave events in the lower stratosphere that Even though the ozone values dropped below 0.1 ppm in October between 12 and 18 km (Fig. 6.12b), slowed ozone depletion (and thus the areal expansion of the ozone hole). The first occurred between 11 and the anomalies show that compared to previous years, ozone concentrations were 3–6 standard deviations 21 August, and the second on 13 September. These above the mean, consistent with smaller ozone loss. disturbed the polar vortex and warmed the lower The weaker 2017 ozone depletion has further Antarctic stratosphere (Fig. 6.11c). Satellite observa - strengthened the long-term downward trend seen in tions acquired by the NASA Microwave Limb Aura Sounder (MLS) show that ClO (chlorine monoxide) the annual ozone hole area since the early 2000s (Fig. levels rose until the 13 September planetary wave 6.13). Since 1988, the only ozone hole smaller than the event and then rapidly declined (Fig. 6.11a, black 2017 hole was observed in 2002, when the only major curve), stalling this year’s ozone depletion (Fig. 6.11b). stratospheric warming on record rapidly warmed the polar vortex in late September and drastically The seasonal evolution of the stratospheric ozone limited ozone depletion. The 2017 wave events noted concentration, ozone hole area, ClO, and temperature in 2017 is similar to those in 2010 and 2012—two above, while strong, were still smaller in amplitude other years with unusually strong wave activity than those in 2002. Moreover, the 2017 wave events that resulted in higher-than-average stratospheric occurred earlier in September, stalling the typical temperatures and smaller-than-average ozone holes. seasonal evolution of ozone depletion. Because of | S191 AUGUST 2018 STATE OF THE CLIMATE IN 2017

212 that, the maximum daily area in 2017 (the top of the vertical gray bars in Fig. 6.13) was the smallest since 2 1988, at 19.6 million km . MLS observations for 2004–16 show that inorganic chlorine (Cl ) levels in the Antarctic lower y stratospheric vortex have declined on average 25 ppt −1 , directly attributable to the Montreal Protocol yr and its amendments (Strahan et al. 2014; Strahan and Douglass 2017). Year-to-year meteorological variability can cause dynamically driven multiyear Cl increases in the Antarctic, as occurred from y 2013 to 2017, in spite of the overall downward trend in stratospheric chlorine loading. Year-to-year Cl y variations and large temperature variability in late September and October complicate the attribution of the decline in hole area (Fig. 6.13) to declining . 6.13. Average Antarctic ozone hole area calculated ig F chlorine levels. The appearance of a downward areal between 7 Sep and 13 Oct (dots), along with the range trend in the last decade is mostly driven by higher of daily values over this period (gray vertical bars). The value for 2017 is highlighted (orange dot and horizontal spring temperatures in the lower stratosphere. line). The years with significantly cold temperatures In Fig. 6.13, ozone holes in the last two decades since 1995 are highlighted (blue dots) and the horizontal with September Antarctic lower stratospheric blue band indicates the range of the associated ozone temperatures one standard deviation below the hole areas. Data for 1979–92 are from Total Ozone average are highlighted in blue. It is apparent that ; 1993 –94 Mapping Spectrometer (TOMS) Nimbus- 7 these cold years produce ozone hole areas with similar are from TOMS Meteor- 3 ; 1996–2004 are from; 2005– size and show no clear trend in area. Depletions in 15 are from Aura OMI; and 2015–17 are from Suomi National Polar-orbiting Partnership (SNPP) Ozone recent years are consistent with current knowledge of Mapping and Profiler Suite (OMPS). There were no the photochemical destruction driven by chlorine and satellite total ozone observations for 1995. bromine compounds and stratospheric temperatures and circulation. | S192 AUGUST 2018

213 7. R EGIONAL CLIMATES — P. Bissolli, C. Ganter, T. Li, A. Mekonnen, and A. Sánchez-Lugo, Eds. a. Overview This chapter provides summaries of the 2017 tem - perature and precipitation conditions across seven broad regions: North America, Central America and the Caribbean, South America, Africa, Europe, Asia, and Oceania. In most cases, summaries of notable weather events are also included. Local scientists provided the annual summary for their respective regions and, unless otherwise noted, the source of the data used is typically the agency affiliated with the au - thors. Please note that different nations, even within . 7.1. Annual average temperature anomalies (°C; F ig the same section, may use unique periods to define 1981–2010 base period) in Canada for 1948–2017. Red their normals. Section introductions will typically line is the 11-year running mean. (Source: Environment define the prevailing practices for that section, and and Climate Change Canada.) exceptions will be noted within the text. In a similar way, many contributing authors use languages other mer, and autumn mean temperatures near or below average across the country. Precipitation measured than English as their primary professional language. To minimize additional loss of fidelity through re- at 28 available stations indicates wetter-than-average interpretation after translation, editors have been spring conditions across the country and drier-than- average summer conditions mainly in southern Brit - conservative and careful to preserve the voice of the ish Columbia. author. In some cases, this may result in abrupt transi - tions in style from section to section. (i) Temperature b. North America The annual average temperature in 2017 for This section is divided into three subsections: - Canada was 0.7°C above the 1981–2010 national aver age, its tenth warmest year since nationwide records Canada, the United States, and Mexico. All anomalies began in 1948 (Fig. 7.1). Four of the ten warmest years are with respect to the 1981–2010 base period, unless have occurred during the last decade, with 2010 being otherwise noted. the record warmest (+2.2°C). The national annual Much of North America had warmer-than-average average temperature has increased by 1.8°C over the conditions during 2017. The annual temperatures past 70 years. Spatially, annual departures above for each country were among the 10 warmest years for their respective records, with Mexico having its +2.0°C were recorded in the north (Fig. 7.2a), which resulted in two provinces/territories reporting an - warmest year on record. Precipitation varied greatly across the continent, with the United States and nual average temperatures among their ten highest: Mexico recording near-average national precipita- Northwest Territories (fifth highest) and Nunavut tion totals. Annual precipitation across Canada was (seventh highest). mostly near to below average, with only parts of the Seasonally, winter (December–February) 2016/17 was 1.8°C above average—the seventh warmest winter east experiencing above-average conditions. Warm, on record. The national winter average temperature dry conditions across the west contributed to the development of one of the earliest and largest fires has increased by 3.4°C over the past 70 years. Winter ever recorded in Canada. Over the course of the year, anomalies above +3.0°C were recorded from the the U.S. experienced 16 weather and climate events northwest to the Atlantic coast, and five provinces/ that each caused over $1 billion (U.S. dollars), tying territories had winter average temperatures among with 2011 as the highest number since records began their ten highest: Northwest Territories (third high - est), Nunavut (fourth highest), Ontario (fifth highest), in 1980. Manitoba (sixth highest), and Saskatchewan (ninth 1) c anada — highest). During the spring (March–May), near- to L. A. Vincent, R. Whitewood, D. Phillips, and V. Isaac below-average temperatures were recorded from the Pacific to the Atlantic coast across southern Canada In Canada, 2017 was characterized by higher- than-average winter mean temperatures from the while above-average temperatures were observed in Yukon to Atlantic Canada, followed by spring, sum - the north. The nationally averaged temperature for | S193 AUGUST 2018 STATE OF THE CLIMATE IN 2017

214 average temperatures were experienced in the western provinces, from southern Yukon to western Ontario. The national autumn temperature has increased by 1.7°C over the past 70 years. December 2017 was 2.0°C above average with most of the north experiencing above-average conditions, while Ontario and south- ern Quebec had below-average temperatures. (ii) Precipitation Over the past decade, precipitation monitoring technology has evolved and Environment and Cli - mate Change Canada and its partners implemented a transition from manual observations to using automatic precipitation gauges. Extensive data inte - gration is required to link the current precipitation observations to the long-term historical manual observations. While this data reconciliation due to changing monitoring technology and methods is in progress, this report presents the analysis based on only 28 stations which have sufficient precipitation observations from similar instrumentation over the period 1981–2017; most of these stations are located in the southern regions of the country. Annual precipitation was near to below average F ig . 7.2. Annual (a) average temperature anomalies across western Canada, with near to above-average (°C) and (b) total precipitation (% of average) in Canada precipitation across eastern Canada (Fig. 7.2b). - for 2017. Base period: 1981–2010. (Source: Environ Seasonally, drier-than-average conditions were ex - ment and Climate Change Canada.) perienced at several stations located in the western - spring 2017 was 0.3°C below the 1981–2010 aver provinces during the winter 2016/17 and summer age and the 27th highest in the 70-year record. The 2017; wetter-than-average conditions were observed national spring temperature has increased by 1.7°C at most stations across the country during the spring over the past 70 years. None of the provinces/territo - 2017; near-average conditions were found at most ries experienced an average spring temperature that stations otherwise. ranked among their ten highest or lowest on record (iii) Notable events and impacts (si nce 1948). - In 2017, the southern British Columbia interior ex Summer (June–August) was 0.4°C above average perienced its longest and most severe wildfire season and the 13th warmest since 1948. Most of the Yukon, - Northwest Territories, and southern Nunavut experi in the province’s history. After a wet spring, the region - enced summer anomalies greater than +1.0°C; North had its driest summer on record. One of the earliest and largest fires ever recorded in Canada burned west west Territories and Yukon reported their seventh and - of Kamloops in the Ashcroft–Cache Creek–Clinton eighth warmest summer on record, respectively. Sum mer temperatures were below average for the regions area. The towns of Ashcroft, Kamloops, and Kelowna extending from southern Manitoba to the Atlantic each received less than 10 mm of total precipitation provinces, and were near average for the remainder during the entire summer. A province-wide state of emergency, the first in 15 years and the province’s of the country. The national summer temperature has longest one, began on 7 July and lasted until 15 increased by 1.5°C over the past 70 years. Autumn September. In total, the British Columbia Wildfire (September–November) was 0.6°C above average Service reported 1265 fires that destroyed 1.2 million and the 19th highest since 1948. Above-average tem - hectares of timber, bush, and grassland, exceeding peratures were experienced in the north and in the eastern provinces, from eastern Ontario to Atlantic the previous record for burned land by 30%. Total Canada, which resulted in two Maritimes provinces, firefighting costs exceeded half a billion Canadian dollars and insured property losses reached close New Brunswick and Nova Scotia, each having their to $130 million Canadian dollars ($103 million U.S. third warmest autumn since 1948. Near- or below- | S194 AUGUST 2018

215 −1 dollars). This memorable season follows the equally , with the trend increasing since 1970 to 0.3°C decade −1 memorable extreme Fort McMurray wildfire in May decade . The nationally averaged precipitation total 2016 in neighboring Alberta province (Kochtubajda during 2017 was 104% of average, the 20th wettest year in the historical record. The annual CONUS et al. 2017). In May, eastern Ontario and southern Quebec each precipitation total is increasing at an average rate of −1 . Outside the CONUS, Alaska had experienced one of their worst spring f looding events 4.3 mm decade its seventh warmest year (+1.2°C departure) since on record. Several rivers exceeded the maximum amount of water released in the past and overf lowed statewide records began in 1925, and near-median from Gananoque to Gaspésie. In Montréal, April precipitation (104% of average). Complete U.S. rainfall totaled 156.2 mm—its second wettest April temperature and precipitation maps are available at in 147 years. Both Ottawa and Montréal had their www.ncdc.noaa.gov/cag/. wettest spring on record—with 400 mm or more at each location (records date back to the 1870s). Spring (i) Temperature f looding forced 4000 people to evacuate their homes For the CONUS, ten months in 2017 were warmer from the Ottawa region to near Quebec City. Many than their respective 1981–2010 average. Every state, - - towns and cities declared states of emergency, includ except Washington, had a warmer-than-average an nual temperature (Fig. 7.4a). Arizona, Georgia, New ing Gatineau, Laval, and Montréal. According to the Mexico, North Carolina, and South Carolina were Insurance Bureau of Canada, spring f looding in April and May resulted in 15 750 claims and $223 million each record warm. - Canadian dollars ($177 million U.S. dollars) in prop erty damages. In total, more than 5000 residences were f looded, 550 roads were washed or swept away by f loods or landslides, and—tragically—on 6 May, two people were swept away by the swollen Sainte-Anne River in the Gaspé region. nited s J. Crouch, A. Smith, C. Fenimore, and u 2) — tates R. R. Heim Jr. The annual average temperature in 2017 for the contiguous United States (CONUS) was 12.5°C or 1.0°C above the 1981–2010 average—its third warmest year since records began in 1895, 0.2°C cooler than 2016 and 0.4°C cooler than 2012 (Fig. 7.3). The annual CONUS temperature over the 123-year period of record is increasing at an average rate of 0.1°C F . 7.4. Annual (a) average temperature anomalies ig . 7.3. Annual mean temperature anomalies (°C; ig F (°C) and (b) total precipitation (% of average) in the 1981–2010 base period) for the contiguous United contiguous United States for 2017. Base period: 1981– States for 1895–2017. Red line is the 10-year running 2010. (Source: NOAA/NCEI.) mean. (Source: NOAA/NCEI.) | S195 AUGUST 2018 STATE OF THE CLIMATE IN 2017

216 The winter (December–February) 2016/17 and Mid-Atlantic. A record-breaking f lood event CONUS temperature was sixth highest at 1.3°C impacted the mid-Mississippi Valley in late April. The Northern Plains were drier than average with above average, driven largely by the second warmest drought conditions developing by the end of the sea - February on record. The Rockies to the East Coast were warmer than average, while the Northwest was son. Summer precipitation for the CONUS was 112% of average, its 13th wettest on record. Above-average cooler than average. The CONUS spring (March– precipitation fell across the Southeast, Great Lakes, May) temperature was 0.9°C above average, its and Northeast. In August, Hurricane Harvey brought eighth warmest spring on record. Above-average record rainfall to parts of Louisiana and Texas (see temperatures spanned the nation with near-average Sidebar 4.3 for more details). Below-average precipi - conditions in the Northwest and Northeast. The tation fell across the Northwest, Northern Rockies, summer (June–August) CONUS temperature was 0.4°C above average, its 15th warmest summer on and Plains. For autumn, the CONUS precipitation total was 94% of average, which is near the median record. Above-average conditions were observed in value. Above-average precipitation fell across the the West and along parts of the East Coast. California Northwest, Northern Rockies, Midwest, and North and Nevada experienced a record-warm summer. The - east. Above-average precipitation also fell in Florida south-central CONUS was cooler than average. The where Hurricane Irma made landfall in September autumn (September–November) temperature was (see Sidebar 4.1 for more details). Below-average 0.9°C above average, the tenth warmest such period on record for the CONUS. Record warmth occurred autumn precipitation occurred across parts of the Southwest, Southern Plains, and Lower Mississippi in parts of the Southwest and Northeast. December 2017 was 0.6°C above average with the first half of Valley. Arkansas had its driest autumn on record. the month having record and near-record warmth By the end of the season, drought covered much of across much of the nation and a significant cold wave the southern CONUS. December 2017 was the 11th impacting the East the last week of the month. driest on record for the CONUS and driest since 1989 with 68% of average precipitation. Drier-than-average conditions stretched from coast to coast with nearly (ii) Precipitation one-third of the CONUS having precipitation totals Locations across the West, Great Plains, Great below the 10th percentile. Lakes, Deep South, Midwest, and Northeast had a wetter-than-average year in 2017, while areas of the Northern Rockies and Plains were drier than (iii) Notable events and impacts T average (Fig. 7.4b). Six states had annual precipita - here were 16 weather and climate events with tion totals above their 90th percentile, including losses exceeding $1 billion (U.S. dollars) each across the United States (Fig. 7.5) in 2017, including three Michigan, which was record wet, while only North Dakota was below its 10th percentile. Areas of the tropical cyclones, eight severe storms, two inland f loods, a crop freeze, drought, and wildfires. The 2017 West, particularly California, experienced significant drought relief in early 2017, with a multiyear drought total tied with 2011 as highest annual number of U.S. billion-dollar disasters (adjusted for inf lation) since nearly eradicated due to the heavy winter precipita - tion. However, the wet winter allowed vegetation to records began in 1980. Cumulatively, these events led to 362 fatalities and caused $306 billion U.S. dollars f lourish, creating an abundance of fuels for wildfires in total, direct costs—a new U.S. annual cost record. during the subsequent dry season. In the Northern The previous costliest year for the U.S. was 2005 with Plains, a dry spring and summer set the stage for a losses of $215 billion. One of the more noteworthy rapidly expanding and intensifying drought. The year began and ended with about one-quarter of the events included the western wildfire season, with total costs of $18 billion, tripling the previous U.S. contiguous U.S. in drought. The CONUS winter precipitation was 120% of annual wildfire cost record set in 1991. Overall, average, its wettest since 1997/98 and ninth wettest wildfires burned over 4.0 million hectares across the on record. Above-average winter precipitation oc United States during 2017, which is well above the - 2000–10 average of 2.7 million hectares. Hurricane curred across the West and parts of the Northern Harvey had total costs of $125 billion, second only Plains and Midwest. Nevada and Wyoming each had their wettest winter. Spring 2017 was tenth wettest to Hurricane Katrina in the 38-year period of record for the CONUS, with 119% of average precipita - for billion-dollar disasters. Hurricanes María and tion. Above-average precipitation occurred across Irma had total costs of $90 billion and $50 billion, the Northwest, Central Plains, Midwest, Northeast, respectively. Hurricane María now ranks as the third | S196 AUGUST 2018

217 F ig . 7.5. Map depicting date, approximate location, and type of the 16 weather and climate disasters in the U.S. in 2017 with losses exceeding $1 billion U.S. dollars. (Source: NOAA/NCEI.) costliest weather and climate disaster on record for the nation, and Irma ranks as the fifth costliest. Tornado activity during 2017 was above average for the first time since 2011 with 1400 tornadoes con - firmed, compared to the 1991–2010 annual average of approximately 1250. There were 34 tornado-related fatalities, well below the 30-year average of 110. exico — 3) M R. Pascual Ramírez and A. Albanil The 2017 mean temperature for Mexico was the highest since national temperature records began in . 7.6. Annual mean temperature anomalies (°C; ig F 1971, marking the fourth consecutive year that a new 1981–2010 base period) for Mexico for 1971–2017. The national annual temperature has been tied or broken. red line represents the linear trend over this period. Precipitation during 2017 varied greatly across the (Source: Meteorological Service of Mexico.) country; however, the 2017 national precipitation total was near average at 99.4% of normal. (i) Temperature The 2017 mean temperature for Mexico was the highest since national temperature records began in 1971 at 22.6°C, or 1.6°C above its 1981–2010 average. This surpassed the previous record set in 2016 by 0.2°C and 2014 and 2015 by 0.5°C, which at the time had been reported as the warmest years on record - (Fig. 7.6). The year 2017 also marks the 14th consecu tive year with an above-average annual temperature. F ig . 7.7. Nationwide daily temperatures (°C; 1981–2010 The national daily mean, maximum, and minimum - base period) for Mexico in 2017. Shaded areas repre temperatures were close to two standard deviations sent the ±2 std dev. Solid lines represent daily values above average during much of January–October (Fig. for the three temperature parameters and dotted lines 7.7), resulting in above-average monthly temperatures are the climatology. (Source: National Meteorological Service of Mexico.) | S197 AUGUST 2018 STATE OF THE CLIMATE IN 2017

218 21.7% of the annual rainfall. During the month, four tropical cyclones (Tropical Storms Lidia and Pilar from the Pacific; Hurricane Max in the Pacific; and Hurricane Katia in the Gulf of Mexico) impacted the nation with heavy rain. Three of those four tropical cyclones made landfall, while Pilar stayed off shore, along Mexico’s Pacific coastline. The last time four cyclones came close to or made landfall in Mexico was in September 1974. Four is the highest number of cyclones to come close to or make landfall in Mexico for any month, according to available hurricane data since 1949. March is typically the driest month of the year, providing only 1.8% of the annual rainfall; however, February was the driest month of 2017, contributing only 1.6% to the annual rainfall total. Northwestern Mexico typically receives nearly 60% of its total annual rainfall during the four-month - period of June–September. However, in 2017, pre cipitation associated with the monsoon and Tropical Storm Lidia caused the region to receive 60%–68% of its annual rainfall total in just one week. (iii) Notable events and impacts Ten tropical cyclones affected Mexico in 2017, five F ig . 7.8. 2017 annual (a) mean temperature anomalies - fewer than the 1971–2012 average of fifteen. Six tropi (°C) over Mexico and (b) precipitation anomalies (% of cal cyclones were near land or made landfall from the normal;). Base period: 1981–2010. (Source: National Pacific basin, and four from the Caribbean basin/ Meteorological Service of Mexico.) Gulf of Mexico. The Pacific number was fewer than the average of ten, and the Caribbean/Gulf of Mexico for 2017. March, June, and November were each warmest on record for their respective months. number was near the average of five. Of note, Caribbean Hurricane Franklin (Category Temperatures were above average across most of the country, with small areas in the middle of the 1 on the Saffir–Simpson scale) produced the year’s country experiencing cooler-than-average condi - highest 24-hour precipitation total for Mexico when tions (Fig. 7.8a). Eight of Mexico’s 31 states reported 404 mm fell in Veracruz upon landfall on 9 August. This value ranks among the top 20 highest daily pre their warmest year on record. With the exception of - Quintana Roo, located in the south, the remaining cipitation totals recorded in the country, according to the Mexican National Meteorological Service. record-setting states are located across the northern half of Mexico (Durango, Sinaloa, Nuevo León, - Drought conditions, which commenced dur ing spring (March–May) 2016, continued to affect Jalisco, Hidalgo, San Luis Potosí, and Tamaulipas). southern Mexico in 2017, in particular the Isthmus of Tehuantepec in Oaxaca. Drought conditions deterio - (ii) Precipitation rated during the first five months of 2017 due to the Rainfall anomalies varied across Mexico, with above-average conditions in northern Chihuahua and warmer- and drier-than-average conditions affecting - Coahuila, coastal Jalisco, northern Puebla and Vera the area. However, heavy precipitation associated cruz, some areas of Oaxaca, and most of the Yucatan with Tropical Storms Beatriz and Calvin, which made landfall in the affected area, helped ameliorate Peninsula. The rest of the country had below-average the long-term drought. These two storms impacted conditions, with the most notable precipitation deficit the same area within two weeks of each other (1 June of 50% of normal precipitation in Sonora, Sinaloa, and and 12 June, respectively), producing much-needed a portion in the central-west (Fig. 7.8b). precipitation and relief for the agriculture sector, but Climatologically, September is typically the wet - test month of the year, contributing about 18.4% of causing damage to infrastructure, such as damaged the annual total rainfall. September 2017 provided roads and bridges due to landslides. Drought also af - | S198 AUGUST 2018

219 fected southern Sinaloa, in northwest Mexico, causing c. Central America and the Caribbean — J. A. Amador, H. G. Hidalgo, agricultural and livestock losses, and a shortage of 1) c entr al a merica drinking water in more than 400 rural communities. E. J. Alfaro, B. Calderón, and N. Mora - For this region, nine stations from five countries Several heat waves affected eastern Mexico, no were analyzed (Fig. 7.9). Stations on the Caribbean tably the Huastecas (an area that encompasses the states of San Luis Potosi, Hidalgo, and Veracruz) from slope are: Philip Goldson International Airport, Belize; Puerto Barrios, Guatemala; Puerto Lempira, 26–30 April and again from 5–8 June. During both heat waves, the maximum temperature reached 50°C, Honduras; and Puerto Limón, Costa Rica. Stations breaking the previous record of 49°C in Huejutla, located on the Pacific slope are: Tocumen Interna - - tional Airport and David, Panamá; Liberia, Costa Hidalgo, set in April 2013. These heat waves were pro duced by a broad high pressure system located over Rica; Choluteca, Honduras; and Puerto San José, northeastern Mexico, inhibiting cloudiness and thus - Guatemala. The station distribution covers the rel increasing temperature. Another major heat wave evant precipitation regimes located on the Caribbean affected the municipality of Aldama, Chihuahua, and Pacific slopes of Central America (Magaña et al. during 11–20 June. 1999). Precipitation and temperature records for the stations analyzed were provided by Central American ig . 7.9. Mean surface temperature (Tm; °C) frequency (F; days) and accumulated pentad precipitation (AP; F mm) time series are shown for nine stations (blue dots) in Central America: (1) Philip Goldson International Airport, Belize; (2) Puerto Barrios, Guatemala; (3) Puerto Lempira, Honduras; (4) Puerto Limón, Costa Rica; (5) Tocumen International Airport, Panamá; (6) David, Panamá; (7) Liberia, Costa Rica; (8) Choluteca, Hon - duras; and (9) Puerto San José, Guatemala. The blue solid line represents the 1981–2010 average values and the red solid line shows 2017 values. Vertical dashed lines show the mean temperature for 2017 (red) and the 1981–2010 period (blue). Vectors indicate July wind anomalies at 925 hPa (1981–2010 base period). Shading depicts regional elevation (m). (Sources: NOAA/NCEI and CA-NWS.) | S199 AUGUST 2018 STATE OF THE CLIMATE IN 2017

220 National Weather Services (CA-NWS) or by NOAA. ing 2017. Low-level circulations in the region showed - Anomalies are reported using a 1981–2010 base pe a slightly stronger-than-average Caribbean low-level riod and were calculated using CA-NWS data. The jet (Amador 1998) during summer (July vectors in Fig. 7.9), a condition usually associated with wetter (drier methodologies used for all variables can be found in and more intense mid-summer drought) conditions Amador et al. (2011). in the Caribbean (Pacific) slope of Central America. (i) Temperature (iii) Notable events and impacts The mean temperature (Tm) frequency distribu - tion for the climatology and for 2017 for all stations - Tropical storms were very active in the Carib is shown in Fig. 7.9. Five stations on the Caribbean bean basin (6°–24°N, 92°–60°W) during 2017. There were eight named storms: five tropical storms (Bret, slope and northern Central America (Tm2, Tm3, Franklin, Harvey, Nate, and Phillipe) and three Tm5, Tm8, and Tm9) had a higher annual mean tem - major hurricanes (Irma, José, and María). Tropical peratures than the base period. The largest annual Storm Nate made landfall in Nicaragua and crossed mean temperature occurred at Puerto San José and Honduras on 5–6 October. Nate induced indirect Choluteca (Tm8 and Tm9, respectively), which were cyclonic circulations (Peña and Douglas 2002) over about 1.0°C above normal. Three stations (Tm1, Tm4, and Tm6) had a mean annual temperature similar to the isthmus, impacting the Pacific slope of Costa - the reference period, and the Liberia Station (Tm7) Rica. According to the Costa Rica National Emer gency Commission (CNE, its Spanish acronym), mean annual temperature was colder by 2.0°C. On the Caribbean side, three stations (Tm1, Tm2, and Nate caused more than $540 million U.S. dollars in damages, the highest amount in the country’s Tm3) depicted a bi-modal temperature distribution documented history of natural disasters since 1996. during 2017. This information is based on a CNE study (Hidalgo 2017) of economic loses including Tropical Storms (ii) Precipitation The accumulated pentad precipitation (P; mm) Alma (2008) and Nate (2017) and Hurricanes Cesar (1996), Mitch (1998), Tomas (2010), and Otto (2016). time series for the nine stations in Central America are presented in Fig. 7.9. Puerto San José (P9) was As with Tropical Depression 12-E in 2011 (Amador et al. 2012), the relative position of Nate with respect - close to normal until pentad 55, when storms pro duced above-average conditions that continued to highly vulnerable areas in Central America is as through pentad 59, followed by a sparse rain period important as tropical storm intensity. Tropical Storm that lasted for over 2 months. This was sufficient - Selma developed in the eastern tropical Pacific and af fected Central America during 27–28 October. Selma to yield above-normal precipitation accumulations made landfall in El Salvador on 28 October, marking at the end of the year. Choluteca (P8) was generally near-normal all year but had a light mid-summer the first time on record a tropical storm made landfall in El Salvador. For additional information on regional drought from pentad 35 to 41. Liberia (P7) started with significantly above-average conditions during impacts from hydrometeorological events during the the first part of its rainy season, then experienced year, please refer to Online Table 7.1. a deep midsummer drought (Magaña et al. 1999) 2) c and a near-normal second part of the rainy season T. S. Stephenson, M. A. Taylor, A. R. Trotman, — aribbean C. J. Van Meerbeeck, V. Marcellin, K. Kerr, J. D. Campbell, that resulted in near-normal annual accumulations. J. M. Spence, G. Tamar, M. Hernández Sosa, and K. Stephenson During most of the year, David (P6) recorded slightly- above-average conditions, while Puerto Barrios (P2) and Tocumen (P5) were wetter than normal during (i) Temperature Normal to above-normal annual mean tempera - most of the year, and extremely wet from pentad 32, tures were recorded across the Caribbean in 2017 with values that surpassed the normal average at the 95% confidence level. Belize (P1) had considerable (Fig. 7.10a). Some locations in the northern Caribbean (including southern Cuba and Bahamas) experienced - rainfall deficit until pentad 35, after which it recu below-normal surface temperatures during Janu perated due to wetter-than-average conditions and - remained normal until the end of the year. Lempira ary–June. In the latter half of the year, above-normal (P3) recorded conditions during most of the year that surface temperatures (+0.2° to +1.0°C) were spread across the entire region. were significantly higher than normal at the 95% confidence level, while Puerto Limón (P4) was the Trinidad reported its tenth warmest annual mean - only station that had below-average conditions dur temperature (28.0°C) since records began in 1946; | S200 AUGUST 2018

221 second highest mean maximum temperature in August (33.6°C), which tied with August 2015 and 2016; and highest daily maximum temperature for August (35.8°C) set on 23 August. San Juan, Puerto Rico, had its third warmest mean temperature in both February (26.4°C) and September (29.2°C) since records began in 1898. Grenada had its highest May mean maximum temperature on record as - temperatures soared to 31.1°C in Point Sa lines. Several locations across the Caribbean had annual maximum temperatures among their nine highest on record (Table 7.1). (ii) Precipitation - The year brought normal to above-nor mal annual rainfall totals to much of the Caribbean (Fig. 7.10b). This was observed in association with above-normal annual and seasonal Caribbean SSTs (Chen and Taylor 2002; Taylor et al. 2002; Spence et al. 2004). During the first quarter of the year, most islands experienced predominantly near-normal conditions. However, some islands—including Tobago, Aruba, Curacao, Dominica, parts of Puerto Rico, Dominican Republic, and Jamaica—observed above- normal rainfall, while severely dry conditions were observed in some areas of Puerto Rico. For the April–June period, apart from Tobago - where moderately dry conditions were re corded in some areas, rainfall over the islands of the eastern Caribbean was normal to above normal. Mixed conditions were observed over the northern islands. Notably, extremely wet F ig . 7.10. (a) 2017 Annual mean temperature anomalies (°C; conditions were observed in central regions 1981–2010 base period) and (b) 2017 annual rainfall pattern as characterized using the standardized precipitation index across of Jamaica. the Caribbean. [Source: Caribbean Climate Outlook Forum Above-normal rainfall dominated much of (CariCOF) and NCEP/NCAR Reanalysis Data. Prepared by the the Caribbean between July and September. Caribbean Institute for Meteorology and Hydrology (CIMH).] This was likely related to the pas - t AB le 7.1. Extreme annual maximum temperatures (°C) for some sage of a number of storms through Caribbean locations. the region, including Hurricanes Station Irma (see Sidebar 4.1), Jose, and Start Year of Max temp Country 2017 Rank Name/ María (see Sidebar 7.1), and favor - (°C) Records Location able atmospheric and oceanic con - 32.1 1985 Bea 9 Aruba ditions in the region enabled by a 5 Bahamas Freeport 1971 29.0 La Niña event in the Pacific Ocean. 3 30.1 1971 LPIA Bahamas Barbados, Dominica, Guadeloupe, 8 30.5 1971 Airport Belize St. Kitts, northern Dominican 1973 Jamaica 32.0 4 Sangster Republic, and eastern Cuba were 2 Jamaica Worthy Park 1973 30.8 extremely wet. In contrast, western 30.7 Martinique 4 1971 Lamentin areas of Jamaica were extremely 7 Trinid ad Piarco 1946 32.4 dry. During the final three months | S201 AUGUST 2018 STATE OF THE CLIMATE IN 2017

222 of the year, mixed rainfall conditions were experi - records commenced in 1951. The September extreme enced across the region. Parts of Trinidad and Tobago anomalies were observed in relation to the passage of Hurricanes Irma and María. and central Jamaica experienced very wet conditions, while parts of Martinique and Guadeloupe were severely dry. (iii) Notable events and impacts Category 5 Hurricane Irma severely impacted Two locations (Cave Valley, Jamaica, and Sainte Marie, Martinique) each observed their wettest the Caribbean during 5–8 September. Some of the impacts of Irma on the islands included: 14 deaths year using records available since 1971, with 2961.2 mm and 2923.0 mm of precipitation, respectively. and over 50 000 residents without electrical power in Port-au-Prince (Haiti) recorded its driest year (588.3 the Turks and Caicos; one death and total destruction mm) using records available since 1971. Cyril E. in Barbuda; several deaths reported in St. Martin; one death and severe damage in Anguilla; damage King Airport in St. Thomas had its second wettest March (148.1 mm). Jamaica experienced its seventh to property in St. Kitts; five deaths and extensive heaviest mean rainfall across the island in March damage in the U.S. Virgin Islands; four deaths and (248.0 mm) using records available since 1881. San severe impacts in the British Virgin Islands; major Juan International Airport, Puerto Rico, recorded its power outages over eastern Puerto Rico; more than 2000 homes damaged in the Dominican Republic; wettest September (401.1 mm) since records began in and f looding in some northern coastal areas in Cuba. 1898. Christiansted, Henry E. Rohlsen Airport, U.S. Virgin Islands, experienced its wettest March (162.6 See Sidebars 4.1 and 7.1 for more detailed information mm) and second wettest September (282.4 mm) since about Irma. S I D E B A R 7.1: IMPACTS FROM HURRICANES IRMA AND MARIA IN — O. MARTINEZ-SÁNCHEZ THE CARIBBEAN - September 2017 featured the passage of two major hur ricanes across the Caribbean: Irma and María. Both hurricanes caused extensive to catastrophic damages across the eastern and northeastern Caribbean islands, where buildings, roads, homes, and the electrical grids were left in ruins. −1 ) Hurricane Irma had sustained winds of 160 kt (82 m s with higher wind gusts, torrential rain, and destructive storm surge just as it made landfall on the islands of Barbuda, Saint Martin, and the British Virgin Islands (BVI). Reports indicated that at least 95% of Barbuda’s infrastructure was damaged or destroyed. The catastrophic damage that occurred in Barbuda - forced a mandatory evacuation of the entire island, with resi ig . SB7.1. F Satellite image of the center of Hurricane - Maria located southeast of St. Croix, USVI on 20 Sep dents brought to the island of Antigua. In the aftermath of Irma, tember 2017. (Source: NOAA/NWS.) 22.5% of the population in Tortola (BVI) was displaced. Even though the center of Irma passed just north of St. Thomas and María before making landfall, resulting in no land-based wind St. John (U.S. Virgin Islands; USVI), wind gusts greater than 117 −1 ) were reported as the southern eyewall clipped observations that would record the maximum winds affecting kt (60 m s the USVI, causing catastrophic damage and five confirmed the island. María’s strong winds also destroyed the FAA-NWS deaths. Hurricane Irma delivered the first powerful punch to radar, which was designed to endure maximum sustained winds −1 ). Although the lack of observations was an the electrical grid, structures, and roads across the northern of 116 kt (60 m s issue for the post-hurricane assessment, there is no doubt that USVI and eastern Puerto Rico. María was much more severe than Irma as the center moved Two weeks later, Hurricane María made landfall in Dominica west-northwestward from southeastern Puerto Rico through as a category 5 storm. María maintained category 5 strength as the interior and into the northwestern sections of the island. it continued its path towards the USVI and Puerto Rico (Fig. SB Most trees were defoliated, and many were either broken or 7.1). María made landfall on the southeastern coast of Puerto - uprooted. Citizens reported the ground and their houses shak Rico as a category 4 hurricane with sustained winds of 134 kt −1 ). Unfortunately, most wind sensors were damaged by ing, and most were amazed by the force of the unprecedented (69 m s | S202 AUGUST 2018

223 4 2 Hurricane María made landfall in Dominica as was 175 × 10 kt —the highest value for any month a Saffir–Simpson category 5 intensity level storm for the Atlantic basin since 1851. Multiple extreme rainfall events were observed in on 18 September and struck southeastern Puerto Jamaica throughout the year. A surface to upper-level Rico at category 4 intensity on 20 September (see trough over the western Caribbean resulted in heavy Sidebar 7.1). In Dominica approximately 15 deaths were associated with María, with an additional 20 rain over parts of Jamaica during 13–15 May, causing - major f looding and landslides. Impacts include de persons missing. The hurricane destroyed much of stroyed bridges, multiple damaged roads, and stranded the island’s infrastructure, removed vegetation, cut off communication and access to the island, and - communities. (Clarendon was the worst affected par ish; f looding was also observed in nine other parishes.) resulted in food and water shortages. Approximately 80% of agriculture crops were ruined in Puerto Rico On 8–10 September, a trough induced by Hurricane Irma across the western Caribbean resulted in heavy and the power grid was destroyed, leaving 3.4 million residents without electricity. thundershowers and f looding over the eastern and Remarkably Hurricanes Irma, Jose (which peaked central parishes. An accompanying lightning strike on the Jamaica Civil Aviation Authority facility in Kings as a top end category 4), and María traversed the - region over a two-week period. Largely as a result ton on 8 September resulted in damage to radar and of these three hurricanes, the accumulated cyclone communication equipment, resulting in the shutdown energy (ACE) index (Bell et al. 2000; see also Section of Jamaica’s airspace for more than 24 hours and for 12-hour periods on 10–11 September. 4f2 for an explanation of ACE) for September 2017 strong winds. While most structures across the island are built in widespread landslides across the Island, making thousands of of concrete and are generally strong enough to withstand roads impassable, especially across the mountainous areas of strong winds, countless homes and buildings suffered some Puerto Rico. The blocked roads disrupted the ability of rescue type of structural damage. Nearly all commercial signs, traffic workers to distribute food, water, medical supplies, and fuel lights, and roads signs were destroyed. All communications— for stranded communities. The damage was so bad that <8% of cellphones and landlines, radio, and television—were largely roads were open and usable a month following María’s passage disrupted; an estimated 95% of the cell towers were out of over Puerto Rico. Damage due to beach erosion and coastal service in Puerto Rico. The electrical grid was also destroyed, flooding was also observed along the shorelines, particularly causing 100% of the island to lose electric power. The damage across western Puerto Rico, where waves destroyed dozens to the electrical grid was so extensive that 5 months later 25% of houses. Storm surge observations across the local islands of residents in Puerto Rico were still without power. ranged from 2–3 meters with wave heights greater than 6 The flash flooding due to Hurricane María’s extreme heavy meters. Winds, waves, and the storm surge across eastern rainfall was catastrophic. The 48-hour rainfall accumulations Puerto Rico sank more than 300 boats. were generally between 380 and 500 mm with isolated higher María was the strongest hurricane to impact Puerto Rico amounts. As a result, 30 rivers reached major flood stage, - since 1928, when Hurricane San Felipe II (also known as Hur with 13 of those at or above record-flood stage. Numerous ricane Okeechobee) made landfall over the island as a category many bridges were destroyed by the strong currents, isolating 5. The official death toll for María stands at 64, although rural communities. The La Plata River, across north central many believe the number is much higher. The catastrophic damage caused thousands of Puerto Ricans to move to the - and northeastern Puerto Rico, flooded its entire alluvial val U.S. mainland after the storm. NOAA’s National Centers for ley, including the municipality of Toa Baja where hundreds Environmental Information (NCEI), in consultation with the of families had to be rescued from their rooftops in Barrios National Hurricane Center (NHC), classified Hurricane María Ingenio and Levittown. Across northwestern Puerto Rico, as the third costliest U.S. tropical cyclone on record, with $90 excessive runoff moving across the dam at the Guajataca Lake billion U.S. dollars in damages across Puerto Rico and the U.S. compromised the stability of the dam, resulting in communities Virgin Islands (www.nhc.noaa.gov/news/UpdatedCostliest.pdf). along the Guajataca River below the dam being displaced due to the risk of dam failure. The excessive rainfall also resulted | S203 AUGUST 2018 STATE OF THE CLIMATE IN 2017

224 d. South America est positive maximum and minimum temperature Warmer-than-normal conditions engulfed much anomalies (+2°C) across the region. of South America during 2017, with anomalies +1.0°C Cooler-than-normal conditions were limited to Colombia and Venezuela during January and March. - or higher. However, below-normal minimum tem peratures were observed across Suriname, French However, on 8 February, Bogota, Colombia, set a new maximum temperature of 25.1°C, surpassing the Guiana, a small area in northern Colombia, and across parts of southern Brazil. During 2017, wetter- previous record of 24.9°C set in 1995. than-normal conditions prevailed over much of the region, with the largest positive anomalies across the (ii) Precipitation coast of Peru. Drier-than-normal conditions persisted Most of northern South America had above- normal precipitation during 2017 (Fig. 7.12). During across northeastern Brazil and across parts of south - January–March 2017, the presence of the coastal El ern South America. Niño caused above-normal precipitation in the coast Anomalies in this section are all with respect to the - al region of Ecuador and southern Colombia (Fig. 1981–2010 average, unless otherwise noted. 7.13). The extreme rainfall events triggered deadly — n orthern s outh a merica 1) R. Martínez, L. López, landslides (see Notable events and impacts section). These locations received 150–300 mm (180%–230%) D. Marín, S. Mitro, R. Hernández, E. Zambrano, and J. Nieto of their normal precipitation from January to March. The northern South America region includes Ecuador, Colombia, Venezuela, Guyana, Suriname, and French Guiana. (i) Temperature - Most of northern South America had above-nor mal temperatures during 2017. Colombia, Ecuador, Suriname, and Venezuela had annual maximum temperatures that were 0.5°–1.5°C above normal and, in some isolated areas, greater than +2.0°C. Below- normal maximum temperatures for 2017 were limited to small areas across northern South America (Fig. - 7.11a). Most of northern South America also experi enced above-normal annual minimum temperatures that were +1.5°C or more, although Suriname, French Guiana, and a small area in northern Colombia ob - served below-normal minimum temperatures during - 2017 (Fig. 7.11b). During 2017, August had the high F ig . 7.11. Annual anomalies of 2017 (a) maximum and . 7.12. Annual anomalies of 2017 precipitation ig F - (b) minimum temperature (°C; 1981–2010 base pe (%; 1981–2010 base period). (Source: Data from the riod). (Source: Data from the NMHSs of Argentina, - NMHSs of Argentina, Colombia, Chile, Brazil, Ecua Colombia, Chile, Brazil, Ecuador, Paraguay, Peru, dor, Paraguay, Peru, Suriname, and Venezuela; pro - Suriname, and Venezuela; processed by CIIFEN, 2018.) cessed by CIIFEN, 2018.) | S204 AUGUST 2018

225 ing f loods in more than a decade in the states of Bolívar and Delta Amacuro. The 5-day accumulated rainfall of 120 mm at the end of August caused f lash f loods and a landslide in Río Mercedes (State of Aragua), affecting hundreds of people and causing four fatalities. outh c entr al s 2) a merica — J. A. Marengo, J. C. Espinoza, L. M. Alves, J. Ronchail, J. Báez, K. Takahashi, and W. Lavado-Casimiro The central South America region includes Brazil, Peru, Paraguay, and Bolivia. (i)Temperature The first half of 2017 was characterized by extreme high temperatures (2°–3°C above normal) in Bolivia, Paraguay, northern Peru, and southern Brazil. Warmer-than-normal conditions continued to affect the region from June through September, with tempera - 1 − tures ranging from 1° to 3°C above normal ; 1981–2010 F ig . 7.13. Precipitation anomalies (mm month over Bolivia, Paraguay, and northeastern Bra - base period) during Jan–Mar 2017. (Source: UCSB CHIRPS v2; processed by CIIFEN, 2018.) zil. Near-normal temperatures were recorded across the region during October–December. (iii) Notable events and impacts Several cold episodes occurred from April During the first quarter of 2017, regional climate through July. The passage of a cold front on 20 June was highly inf luenced by sea surface temperature brought cold temperatures to the southern half of warming of the coastal El Niño (Sidebar 7.2). The - Brazil, with some regions recording minimum tem sudden warming in the eastern equatorial Pacific was peratures <0°C. São Joaquim and Bom Jesus (located different from the typical development of El Niño in the state of Río Grande do Sul; climatologies of events. Although its impacts in the Andean countries 5.9°C and 8.0°C, respectively) reported minimum varied, the most significant effects of the intense and temperatures as low as −3°C and −2.6°C, respectively. quick coastal El Niño were mainly associated with A polar air intrusion during 17–19 July (see Notable extreme precipitation events and subsequent f looding events and impacts section) brought cooler-than- and landslides. normal conditions to parts of southern and eastern From January to April, rainfall exceeded normal Brazil and in western Amazonia, resulting in monthly conditions in a large part of the coastal region of minimum temperatures 1°–3°C below normal. - Ecuador and most of Colombia, Suriname, and Ven ezuela. Heavy rain during February–April produced (ii) Precipitation f loods in Ecuador, which were responsible for more The first half of 2017 was characterized by below- 00 people than two dozen fatalities and over 127 normal precipitation in Bolivia and west-central and 0 affected in the provinces of Guayas and Manabi. northeastern Brazil. Above-normal precipitation was Some locations set new precipitation records during observed in northwestern Amazonia, southern Brazil, and along the northern Peruvian coast during the March. In Mocoa, Colombia, extreme rainfall (130 second half of the year. mm in 3 hours) in late March fell in areas that were already saturated by heavy rain earlier in the month, The dry conditions observed in 2016 in Bolivia causing f lash f loods and a landslide that killed more and northeastern Brazil (Marengo et al. 2017) per - than 250 people and left over 300 people injured. sisted through 2017. Most of central South America east of the Andes experienced below-normal rain During March–May, devastating f loods affected - −1 fall (100–150 mm month ; Fig. 7.14) from January the departments of Antioquia, Cundinamarca, and through April, with only weak episodes of the South Choco in Colombia. Atlantic convergence zone (SACZ)—a summertime In Venezuela, above-normal precipitation fell dur - circulation pattern associated with rainfall in the - ing August–September, triggering the most devastat | S205 AUGUST 2018 STATE OF THE CLIMATE IN 2017

226 1 − ; 1981–2010 base period) for Jan–May 2017. [Source: . 7.14. Monthly rainfall anomalies (mm month ig F Climate Hazards group Infrared Precipitation with Stations (CHIRPS) dataset.] region. The extreme dry conditions were ideal for between Lima and the Andean cities was inaccessible the development of wildfires. According to the for several days. Brazilian National Institute for Space Research, the The January 2017 precipitation total for the city of São Paulo was 453.8 mm, 179% of normal for the total number of wildfires for the Amazon region in 00, the highest number since records month, and its wettest January since 2011. The copi 0 2017 was 272 - ous rain prompted f lash f loods in several locations 0 began in 1999, which burned over 986 00 hectares. across the city. In December, a SACZ episode caused heavy rainfall - In the city of Maceio, located on the coast in over southeastern Brazil, Paraguay, and Bolivia, re sulting in high river levels for Amazonas-Solimões the state of Alagoas in northeast Brazil, a state of emergency was declared due to torrential rains that and Negro River basins. Floods affected the cities of Trinidad and Santa Cruz de la Sierra (Bolivia), as produced landslides and f lash f looding on 27 May, well as soybean crops and livestock in the lowlands killing three people. By 29 May, over 8400 families of Bolivia. 00 people were left 5 were affected, and more than 16 homeless. Total rainfall in May 2017 was 742.4 mm (more than twice the monthly normal of 344.7 mm), (iii) Notable events and impacts Heavy rains that fell in Peru during January–May with 169.6 mm recorded on the 27th. - During the first two weeks of June, well-above- (Fig. 7.14) were triggered by the coastal El Niño pres ent in the eastern tropical Pacific Ocean (see Sidebar normal rainfall was observed in the eastern portion 7.2). Torrential rainfall triggered f lash f loods and of the state of Santa Catarina (southern Brazil) due to 00 people in the landslides that affected over 625 0 the passage of a cold front. Torrential rains affected more than 28 8 00 people and, in some districts, a state regions of Tumbes, Piura, Lambayeque, La Libertad, Ancash, Ica, and Arequipa and claimed nearly 100 of emergency was declared due to f loods. The same lives. Losses include 242 bridges, 13 cold front caused heavy rainfall and f lash f loods in 27 km of rural 2 and main roads (1.5% of the national road system), Rio de Janeiro, and the total rainfall measured on 20 45 35 km of agricultural irrigation channels, and 60 3 June was almost 247 mm (June climatology is 461.8 mm). This event affected public transportation in the 400 ha of crops. In the suburbs of Lima, landslides (“huaycos”) destroyed houses, and the highway city and f looded some neighborhoods. | S206 AUGUST 2018

227 F ig . 7.15. (a) Rainfall anomalies in west-central Brazil during Jan–Apr 2017; the city of Brasilia is marked with a black dot; (b) Monthly rainfall (mm) in Brasilia from Oct 2016 to Apr 2017; 2017 monthly totals are depicted in light blue, while the 1981–2010 normals are in dark blue; (c) Time series of annual rainfall (mm) from 1965 to 2017; normal annual value is depicted with a red line. (Source: GPCP and INMET.) - West-central Brazil, particularly Brasilia (Distrito Paulo, the maximum temperature was 8°C (climatol Federal), has been affected by dry conditions since ogy of 11.7°C) on 18 July, and one person died due to 2015. The drought conditions, which continued into exposure to the cold temperatures. From mid-July to 2017, were the worst in the last 57 years. In April - mid-August, a cold front in Peru produced tempera 2017, Brasilia received only 20% of its normal April tures as low as −20°C at 4000 meters above sea level precipitation, which is 125 mm; in fact, during the (the record-coldest value is −25°C set on 6 July 1968 at Macusani station in Puno region), and snow fell in peak of the rainy season (October 2016–April 2017), the Andes of Peru and Altiplano. only February had above-normal monthly rainfall (Fig. 7.15). This prompted a state of emergency and mandatory water restrictions. J. L. Stella and L. S. Aldeco — merica a outh 3) s outhern s This region includes Argentina, Chile, and Uru - The most intense cold episode during austral winter 2017 occurred during 17–19 July. A polar g uay. air mass affected the Andes, bringing cooler-than- (i) Temperature normal conditions to the western Amazonia regions Above-normal temperatures were observed of Brazil, Peru, and Bolivia. On 17 July, minimum across southern South America (SSA) during 2017, temperatures as low as 10°C were recorded in the Bo - livian Amazon and in Puerto Maldonado, Peru (July with annual mean temperatures 0.5°–1.5°C above climatology of 18°C), while on 18 July the western normal. The national mean temperature anomaly Brazilian Amazon saw temperatures drop to 7.2°C in for Argentina and Uruguay was +0.68°C and +1.0°C, - Campo Verde (located in the state of Mato Grosso; respectively, placing 2017 as the warmest year on re cord since 1961 for both countries. The five warmest climatology of 21.2°C), 11.3°C in Epitaciolândia years on record for Argentina have all occurred since (located in the state of Acre; climatology of 19.0°C), 2012 (Fig. 7.16). The mean temperature anomaly by and 11.1°C in Guajará-Mirim (located in the state of Rondonia; climatology of 20.0°C). In the city of São decade since the 1960s (Fig. 7.17) shows an increase | S207 AUGUST 2018 STATE OF THE CLIMATE IN 2017

228 across central and northern Argentina during 2001–2010 and a significant rise across the country as a whole during the decade to date (2011–2017). Summer (December–February) 2016/17 was particularly warm over most of SSA, with mean temperatures 1°–2°C above normal. Chile had its second warmest summer since 1964. At the end of the season, a heat wave affected a large area in central Argentina. The maximum duration of extreme heat, defined here as minimum and maximum temperatures . 7.16. Annual mean temperature anomalies (°C; 1981–2010 base F ig surpassing the 90th percentile, ranged period) for Argentina for 1961–2017. (Source: Argentina’s National between five and eight days, and for some Meteorological Service.) locations these conditions extended into the beginning of March, resulting in one of the latest heat waves recorded in that area. Below-normal maxi - mum temperatures and above-normal mini - mum temperatures during austral autumn (March–May) resulted in near-normal mean temperatures across much of SSA (±0.5°C). Winter (June–Au - gust) 2017 was extremely warm over much of the - region, with tempera tures 1°–3°C above nor - mal across the eastern and northern parts of - SSA. This was the warm est winter on record for Uruguay and second warmest for Argentina in their 46-year records. Several individual lo - cations in eastern Ar - gentina reported their warmest winter on re - cord. A new national maximum temperature record was set on 17 June when the temperature soared to 40°C at Tino - - gasta (northwestern Ar gentina), marking the F ig . 7.17. Decadal mean temperature anomalies (°C; 1981–2010 base period) across first time on record the Argentina from the 1960s through 2017. (Source: Argentina’s National Meteoro - logical Service.) | S208 AUGUST 2018

229 temperature reached 40°C between May and July. - During May, intense daily rainfall (>100 mm) af Meanwhile, a few cold outbreaks during June and July fected the Coquimbo region in central Chile; it was considered the most extreme event since the 1950s. also impacted the region. Bariloche (northwestern La Serena, also in central Chile, received 200% of its Patagonia, Argentina) broke its absolute minimum normal precipitation for the month. temperature record when temperatures dropped to As La Niña conditions emerged in October, the −25.4°C on 16 July. The previous record was −21.1°C precipitation pattern across SSA changed abruptly. set on 30 June 1963. These cold spells also caused heavy Most of the SSA region had below-normal rainfall, snowfalls over southern Argentina and Chile and broke particularly during October and November, with several minimum temperature records across Uruguay during June and July. 50–150 mm below-normal precipitation reported in northern Argentina and 25–50 mm below normal in Spring (September–November) was characterized central Patagonia of Chile and Argentina. by below-normal temperatures. The cooling in the Pacific Ocean during spring contributed to the change (iii) Notable events and impacts in temperature pattern across SSA. During January, an extraordinary heat wave - affected central Chile and Argentina. The Chil (ii) Precipitation ean cities of Antofagasta and Curicó recorded the Much of eastern and southern SSA had above- - most prolonged warm periods with extreme high normal annual rainfall during 2017. The most sig nificant precipitation totals were observed in central temperatures for 14 and 17 days, respectively. The Argentina, Uruguay, eastern Patagonia, and southern temperature at Santiago de Chile rose to 37.4°C, the highest value recorded in the 104-year record. The Chile (between 40° and 50°S). Conversely, central locations of Chillán (41.5°C), Los Angeles (42.2°C), Chile and northwestern Patagonia (32°–42°S) had and Curicó (37.3°C) also broke their maximum below-normal rainfall during 2017. The estimated an - temperature records. In Argentina, the temperature nual precipitation anomaly for Argentina was 109.8% reached 43.4°C on 27 January at Puerto Madryn, the of normal, the fourth consecutive year with above- normal rainfall after a long dry period (2003–13). At highest temperature ever recorded so far south (43°S) anywhere in the world. Isla de Pascua, Chile, 2017 was the second driest year on record since 1950. - Drought, combined with high temperatures, trig gered devastating forest fires in large areas of central The beginning of 2017 was particularly dry in and southern Chile in January. More than 600 000 central Argentina and central Chile, causing severe drought that contributed to the development of - hectares were burned, with thousands of people af fected. Central Argentina had a similar situation with wildfires in both countries. Meanwhile, northeastern forest fires affecting La Pampa province, leading to Argentina and Uruguay had above-normal rainfall during summer 2016/17. This pattern intensified more than 1 million hectares burned and cattle and crops losses. and extended to central Argentina during autumn and winter. Intense precipitation affected central On 30 March, Comodoro Rivadavia reported an and northeastern Argentina and Uruguay, triggering impressive daily rainfall amount of 232.4 mm, close f loods in large parts of the region. One of the most to the city’s annual normal precipitation total. The heavy rain produced severe f lash f loods that affected extreme precipitation events occurred in Comodoro the region. A few days later, the city was impacted Rivadavia, a city located in eastern Patagonia (see - Notable events and impacts section). Heavy precipi once again by heavy rainfall (more than 60 mm in a tation events also occurred in April 2017, affecting few hours), leaving most of the city destroyed. northern, eastern, and southern Uruguay, with On 15 July, Santiago, Chile’s capital, experienced several locations recording new daily precipitation its heaviest snowfall since 1922, with 3–10 cm of snow. Meanwhile, the same synoptic system produced 40 records. The copious rainfall triggered f loods and cm of snow over the city of Bariloche, Argentina—its caused road interruptions. heaviest snowfall in 20 years. | S209 AUGUST 2018 STATE OF THE CLIMATE IN 2017

230 S I D E B A R 7. 2 : K. TAKAHASHI, V. ALIAGA-NESTARES, — THE 2017 COASTAL EL NIÑO G. AVALOS, M. BOUCHON, A. CASTRO, L. CRUZADO, B. DEWITTE, D. GUTIÉRREZ, W. LAVADO-CASIMIRO, J. MARENGO, A. G. MARTÍNEZ, K. MOSQUERA-VÁSQUEZ, AND N. QUISPE (Fig. SB7.2b). The coastal city of Piura (5.2°S, 80.6°W), located The original concept of El Niño consisted of anomalously at the core of the ITCZ extension, had a February–March high sea surface temperature and heavy rainfall along the arid precipitation total of 723 mm, which is nearly seven times its northern coast of Peru (Carranza 1891; Carrillo 1893). The normal amount of 106 mm. The largest precipitation anomalies concept evolved into the El Niño–Southern Oscillation (ENSO; were observed at low and medium elevations on the western Bjerknes 1969), although the original El Niño and the Southern slope of the Andes, triggering several floods and mudslides Oscillation do not necessarily have the same variability (Deser along the Peruvian coast. Mean January–March 2017 river and Wallace 1987), and the strong El Niño episode in early discharge was around 250% of normal in the Santa (9.01°S, 1925 coincided with cold-to-neutral ENSO conditions (Taka - 77.76°W), Rímac (11.77°S, 76.46°W), and Cañete (12.77°S, hashi and Martínez 2017). To distinguish the near-coastal El 75.83°W) River basins. Niño from the warm ENSO phase, Peru operationally defines Impacts along the coast were severe. In the northern re - the “coastal El Niño” based on the seasonal Niño 1+2 SST gions, a total of 50 927 houses were damaged with close to anomaly (ENFEN 2012; L’Heureux et al. 2017). While recent 1.2 million people affected by flooding, and over 76 000 ha of attention has been brought to the concept of ENSO diversity crops were damaged. As is common with El Niño, this event (e.g., “central Pacific” vs “eastern Pacific” events; Capotondi affected marine resources, primarily the anchovies (Ñiquen and et al. 2015), the coastal El Niño represents another facet of Bouchon 2004; Ñiquen et al. 1999), resulting in decreased fat ENSO that requires further study in terms of its mechanisms content and early spawning as a reproductive strategy (IMARPE and predictability. 2017). The estimated growth of the Peruvian gross domestic A strong coastal El Niño developed off the coast of Peru product in 2017 was 1.3% lower than expected (BCRP 2017). from January to April 2017 (ENFEN 2017; WMO 2017a,b; The coastal El Niño appears to have been initiated by Takahashi and Martínez 2017; Ramírez and Briones 2017; westerly anomalies in the equatorial far-eastern Pacific in Garreaud 2018). The changes were dramatic within the cool January, the largest for that month since 1981, with a northerly coastal upwelling region, as daily SST at Puerto Chicama (7.8°S, component near the coast (Fig. SB7.3a). At upper levels, the 79.1°W) increased abruptly from ~17°C by mid-January to a Bolivian high (Lenters and Cook 1996) was located west of peak of 26.9°C in early February (ENFEN 2017). The mean its normal position, and a subtropical ridge spread from the maximum/minimum air temperature anomalies along the coast Northern Hemisphere, resulting in easterly anomalies and ranged between +1.0°C and +2.3°C across the north, central, divergence favorable for convection over northwestern Peru and southern regions during February–March. (Kousky and Kayano 1994; Vuille et al. 2000). The Madden– Convective precipitation is activated in the eastern Pacific Julian oscillation (MJO) had its highest amplitudes in the second when SST exceeds a threshold of ~26°–27°C (Takahashi and half of January and was dominated by the MJO phases 1 to 3, Dewitte 2016; Jauregui and Takahashi 2017). With SST well in which feature westerly anomalies in this region, according to excess of 27°C, the southern ITCZ branch (Huaman and Taka - the Real-time Multivariate MJO index (RMM; Wheeler and hashi 2016; Fig. SB7.2a) was very strong between February and Hendon 2004; see Section 4c). The northerly component March 2017 and extended into the South American continent 1 − . SB7.2. Feb–Mar SST (contours, interval 1°C) and rainfall (shading, mm day ig F ): (a) 1981–2010 climatology, (b) 2017 observations, and (c) 2017 anomalies (contour interval: 0.25°C). (Sources: SST: ERSST v5; rainfall: CMAP.) | S210 AUGUST 2018

231 was probably associated with the negative mean sea level pressure anomalies in the Fig. S B7. 3a ). The latter southeast Pacific ( could have been associated with Rossby- wave teleconnections from the western Pacific (Garreaud 2018), but the SLP anomalies also extended zonally uniformly across the subtropical South Pacific (Fig. SB7.3a), consistent with the negative phase of Antarctic Oscillation (M 2000a), while the subtropical anomalies closer to the coast of South America were probably partly a response to preexisting positive SST anomalies in that region. In early February, rainfall in the south - ern ITCZ became active, and the subse - quent growth and maintenance of the event was consistent with the ocean–atmosphere mechanisms proposed for the 1925 coastal El Niño (Takahashi and Martínez 2017), that is, positive feedback between surface warming to the south of the equator, enhanced southern branch of the ITCZ, and reinforced near-equatorial northerly surface wind anomalies (Figs. SB7.3b,c; e.g., Xie and Philander 1994). The strong coastal ocean warming off northern Peru (> +2°C) was limited to a shallow layer of about 30 m until the end of February, consistent with local surface forcing (Garreaud 2018). This, jointly with the smaller regional basin scale, explains the much faster timescale of this event (Takahashi and Martínez 2017). The termination of the event in April (Fig. SB7.3d) was also abrupt, as the insolation- driven seasonal sea surface cooling (Taka- hashi 2005) deactivated the southern branch of the ITCZ, shutting down the feedback mechanism. We should note that ig F - . SB7.3. Mean sea level pressure (shading, hPa), 10-m wind (vec 1 − − 1 − 1 tors, m s , in black), and precipitation [contours: 3 mm day ; >1 m s toward the end of March, the subsurface solid (dashed) contours represent positive (negative) anomalies. Zero warming off northern Peru became deeper not shown]. (Sources: MSLP and wind: ERA-interim; precipitation: (down to 180 m; ENFEN 2017) and per - C M A P.) sisted until May, probably associated with 2017, but only once the event started in late January, since local ocean–atmospheric Bjerknes feedback (Takahashi international climate models provided little indication that and Martínez 2017; Dewitte and Takahashi 2017), although such an event would occur (ENFEN 2017). Extending the warm ENSO conditions did not materialize (L’Heureux et lead time and accuracy of the prediction of coastal El Niño al. 2017; also see Section 4b), similar to 1925. events is a critical challenge for Peru and requires increased The knowledge of the basic mechanism of the 1925 understanding and improved models for this region. coastal El Niño guided the official Peruvian forecasts in | S211 AUGUST 2018 STATE OF THE CLIMATE IN 2017

232 e. Africa In 2017, most of Africa experienced above-normal air temperatures, with slightly lower-than-normal temperatures in a few areas of West and southern Africa (Fig. 7.18). For the continent as a whole (except for a few areas in southern Africa and in the deep Sahara, stretching between Niger and Libya), 2017 was above normal by about 0.3°–1.8°C. Annual mean rainfall was above normal over boreal summer rain - fall areas in West, central, and parts of East Africa. Below-normal rainfall was recorded in equatorial and southern Africa between 10° and 20°S (Fig. 7.19). Extreme events like heavy rainfall and f looding were reported in many parts of the region. These reports include heavy downpours in February and December 2017 in Morocco and during August and September in Nigeria, The Gambia, and Niger. Tropical cyclone events in the Mozambique Channel affected Mozambique and Zimbabwe. An additional tropical cyclone over the southern Indian Ocean af - fected Réunion Island and Madagascar. This report was compiled using observational records from the meteorological and hydrological services of Morocco, Egypt, Nigeria, Ethiopia, South − 1 F ig ; . 7.19. Annual 2017 rainfall anomalies (mm day 1981–2010 base period) over Africa. (Source: NOAA/ NCEP.) Africa and the southern Indian Ocean Island coun - tries of Madagascar, Seychelles, Mayotte (France), La Réunion (France), Mauritius, and Rodrigues (Mauritius). Reanalysis data from NCEP/NCAR and the ECMWF and rainfall from version 2 of Climate Hazards Group Infrared Precipitation with Stations data (CHIRPS) were also used. Precipitation fields from reanalyses are problematic, but unfortunately observations from many African nations are not available. fatalities and f looding hazards are as re - ported in news outlets. The climatological base period used is 1981–2010. 1) n orth a frica — K. Kabidi, A. Sayouri, M. ElKharrim, and A. E. Mostafa Countries considered in this section include Mo - rocco, Mauritania, Algeria, Tunisia, Libya, and Egypt. In 2017, the region was marked by unusually high temperatures, below-normal rainfall, and extreme events including heavy rainfall and f looding that caused loss of life and property damage. In Morocco, extreme events were pronounced during February . 7.18. Annual 2017 mean temperature anomalies F ig and December. (°C; 1981–2010 base period) over Africa. (Source: NOAA/NCEP.) | S212 AUGUST 2018

233 (i) Temperature of Morocco, northern and northwestern Mauritania, extreme southern Algeria, and over Libya and Egypt. Temperatures in 2017 were above normal over most of the region. However, January was cooler Extreme minimum air temperatures of 0.2°C at El- than normal over most of North Africa (Fig. 7.20a). Arish (northern Egypt) on 10 January and 1.3°C at Asyut (Upper Egypt) on 10 February were recorded. Minimum temperatures of about 3°C below normal Minimum and maximum temperatures over were observed over the mountains and southern parts of Morocco. Mean temperatures of about 2°–3°C Morocco were among the highest in the period of re - below average were also observed over most of Alge cord starting in 1901. The mean annual temperature - ria, southern Tunisia, and western parts of Libya in anomaly was 0.65°C above normal (exceeding the value recorded in 2016 by +0.2°C), with anomalies January (Fig. 7.20a). January–February mean tem - peratures were also below normal over southern parts of +0.6°C for the annual maximum temperature and +0.7°C for the annual minimum temperature. The largest monthly mean temperature anomalies of about 3.7°C were recorded in May and October in Morocco. Summer temperatures were above normal over most of the region and up to 3°C above normal over most of Morocco (Fig. 7.20b). Autumn (September– November) temperatures were above normal over Morocco, with anomalies exceeding 2°C over central regions, and in Algeria. Eastern Algeria, southern Tunisia, and most of Libya and Egypt were colder than normal (Fig. 7.20c). (ii) Precipitation Annual precipitation over Morocco was 61% of normal. Morocco received 40% less precipitation in 2017 than 2016. Moroccan precipitation was char - acterized by strong spatial and temporal variability. About 13% of normal precipitation was recorded in the south at Smara (Saharan region), while close to normal rainfall (~89% of normal) was received in the - north at Rabat. About half (46%) of the annual precip itation was received in February and December 2017. January–February precipitation over the region was largely below average due to a prevailing anti - cyclonic circulation that settled over North Africa’s Atlantic coast and western Europe. However, heavy rainfall was recorded in Morocco on 22–24 February (Fig. 7.21). Heavy rainfall events were observed at many places, including 91 mm in 24 hours at Agadir in southwest Morocco. Above-normal rainfall was observed over northern Algeria in February. Heavy rainfall was also reported from Egypt in January. The spring season (December–March) was gener - ally characterized by rainfall deficits over most of the region. In Morocco, the seasonal precipitation was - . 7.20. (a) Northern Africa 2017 surface air tem ig F 47% of normal. Seasonal totals of 60%–80% of normal perature anomalies (°C; 1981–2010 base period) for - were recorded along the northern coastal areas of Al Jan. (b) Northern Africa 2017 surface air temperature geria. Some eastern and southern Moroccan stations anomalies (°C; 1981–2010 base period) for Jun–Aug. received less than half of their normal precipitation, (c) Northern Africa 2017 surface air temperature with some recording zero precipitation. Spring pre - anomalies (°C; 1981–2010 base period) for Sep–Nov. (Source: NOAA/NCEP.) cipitation was below normal, especially in northern | S213 AUGUST 2018 STATE OF THE CLIMATE IN 2017

234 exceeding 40°C. These were associated with eastern continental winds and caused significant forest fires, especially in Morocco and Algeria. About 325 forest fires were reported in Morocco, causing the destruc - tion of about 2056 hectares of forested land. a 2) w est frica — S. Hagos, I. A. Ijampy, F. Sima, S. D. Francis and Z. Feng In this section, West Africa refers to the region between 17.5°W (eastern Atlantic coast) and ~15°E (the western border of Chad), and from 5°N (near the Guinean coast) to 20°N. It is typically divided into . 7.21. Northern Africa 2017 rainfall rate anomalies F ig two climatically distinct subregions; the semiarid − 1 (mm day ; 1981–2010 base period) over 21–24 Feb. Sahel region (north of about 12°N) and the relatively (Source: NOAA/NCEP.) wet Coast of Guinea region to the south. The rainy and east-central Algeria. In Algeria, Skikda received - period over the region is associated with the latitu no precipitation, and Annaba received only 11% of its dinal movement of the convective zone referred to as normal March rainfall. Alexandria received a record the West African monsoon which typically occurs rainfall of 50 mm in 24 hours on 21 April, the highest during June through September. 24-hour rainfall in Egypt during 2017. (i) Temperature Precipitation during summer (June–September) was generally below normal over the entire region; The annual mean temperature over western parts however, convective events in August produced of West Africa was higher than average, with much - of the region about 0.5°C above normal (Fig. 7.18). monthly rainfall that was 192% of normal in Mo rocco. Al Hoceima observed 44.8 mm, an order of During September (Fig. 7.22), much warmer-than- average conditions— ~0.5°C above normal—were magnitude greater than its normal 4 mm. In Algeria, reported with record-warm conditions over Ghana, summer rainfall was highly variable, with deficits Ivory Coast, Burkina Faso, and southern Nigeria. across the northern part of the country. Western There was, however, significant regional variability. coastal stations observed amounts on the order of 40%–85% of normal. Stations in the plains and inland For example, daily temperatures >40°C were recorded - between January and June in parts of Nigeria, es basins of Algeria received just 10%–40% of normal, while Saïda (in the northwest) received 166% (66% pecially over its northern states. The highest daily temperatures of April, March, and May were 45.3°, above normal) of normal precipitation. Farther south, 44.2°, and 44°C, recorded at Maiduguri, Yelwa, and at Saharan Atlas stations of Algeria, the seasonal to - tals varied, from 20%–60% of normal near El Bayadh Nguru, respectively. Other cities, such as Sokoto and Mechria, to 125%–150% of normal at AinSefra and Naâma. In contrast to summer, precipitation during autumn was 38% of normal in Morocco. However, precipitation for November was above normal over northern Tunisia, leading to f looding on 10–11 No - vember at Gabès, killing 5 people and causing more than 117 evacuations. (iii) Notable events and impacts Flooding in both February and December 2017 - caused loss of life and damage to property in Mo rocco. Cold spells, ranging from 0.3° to 7.0°C below normal, were reported in January and December in Morocco. An all-time heavy rainfall of 119.2 mm was recorded on 23 February at Rabat, Morocco. ig F . 7.22. Temperature anomalies (°C; 1981–2010 base Extended heat waves occurred over the region period) for West Africa for Sep 2017. (Source: NOAA/ NCEP.) during May and June with maximum temperatures | S214 AUGUST 2018

235 and Yola, also recorded daily high temperatures of drier conditions dominated much of the region, with - - significant regional variability. According to the Ni 43.0° and 43.5°C, respectively. The Gambia also ex perienced warmer-than-normal conditions during gerian Meteorological Agency, most of the country recorded normal rainfall conditions, while the cities 2017. The annual mean maximum temperature over in central and northwestern Nigeria recorded below- the country was 35.6°C, about 4.5°C above normal. normal rainfall. The Gambia experienced early onset Daily maximum temperatures exceeded 40°C in some places, such as the 47.7°C recorded at Kaur, in the and cessation of the rains, but overall seasonal rainfall was near normal. However, the timing of rainfall had Central River region of The Gambia. an uneven distribution, with prolonged dry spells and f looding, leading to crop failure in some parts of the (ii) Precipitation country. Most of the rain during the 2017 season was Rainfall totals for June to September (JJAS), dur - ing which the West African monsoon provides much in July–September, with the highest amounts of 130.8 of the annual precipitation, are presented in Fig. mm recorded in August at Jenoi in the lower river 7.23. During JJAS, the northern Sahel was wetter, region of The Gambia. and coastal regions were drier, than normal. This (iii) Notable events and impacts was consistent with above-normal SSTs over the northern tropical Atlantic early in the season. Early- On 6 July, f looding and wind storms occurred at Jarra Bureng and Jasobo, in The Gambia’s Lower Riv - season monsoon precipitation over the Sahel region, particularly northern Nigeria and southern Niger, er region, destroying hundreds of homes and farms, and affecting 20 00 individuals; wells and latrines was observed. Later in the season, the southern and 0 central tropical Atlantic was warmer, and relatively were also affected. The event lasted for 4 hours. The event affected 94 households and 857 people across 5 communities. On 12 July, a wind storm at Kerewan, North Bank region, caused one casualty and affected 222 households. A wind storm claimed two lives the same day in The Gambia’s West Coast region. In Nigeria, heavy rain during August and Septem - ber caused the Niger and Benue Rivers to overf low, causing f looding in the Benue and Kogi States. In 0 00 people were Benue State, it was reported that 100 displaced by f looding, 12 local governments within the state were affected, and around 4000 homes dam - aged. No fatalities were reported. Flooding in Kogi State came just days after thousands of people were - displaced by f loods in Benue. The Kogi f lood dis placed over 10 0 00 people. The worst affected area was the state capital, Lokoja, which lies at the conf luence of the two rivers. Other affected areas included Ibaji, Igalamela-Odolu, Ajaokuta, Bassa, and Koton-Karfe. A bridge at Tatabu village along Mokwa-Jebba road, in the Kwara State, collapsed after a heavy rainfall. The road is the major link between the northern and southern parts of the country; motorists were advised to use alternative routes. The popular Ahmadu Bello way in Victoria Island, Lagos State, was temporarily closed on 7 September by the Lagos State Police Com - mand due to f looding. On 27 August, several hours of rainfall caused f loods in Churchill Town, Bakoteh, and Ebou town ig . 7.23. Jun–Sep 2017 precipitation (mm) for West F in The Gambia’s west coast region, and Tabanani Africa: (a) total accumulation; 100-mm isohets (red and Sare Molo in its Central River region. Two dashed line), 600-mm isohyets (red solid line). (b) lives were lost, more than a thousand homes were Departure from 1981–2010 climatology. (Source: NOAA/NCEP.) damaged, and about 4000 people were displaced. | S215 AUGUST 2018 STATE OF THE CLIMATE IN 2017

236 On 12 and 29 August, f looding at Kuntaur Niani, 7.24b). During June–August (JJA), above-normal temperatures covered large parts of GHA, extending in the Central River region, caused significant internal displacements, damage to public and private to the Ethiopian highlands and northwestern Sudan properties, including a bridge, and submerging of Republic (Fig. 7.24c). However, South Sudan and western Ethiopia enjoyed near-normal temperatures farmlands. There were five fatalities and around 8000 during this season. During September–November people affected. In Niger, heavy rain on 26 August caused f looding (SON), cold anomalies of up to −2°C prevailed over in the capital city of Niamey and surrounding areas; the northern half of Sudan Republic, southwestern Ethiopia, and adjoining South Sudan and northern around 100 mm of rain fell in Niamey. The UN Of - Uganda, extending across western Kenya towards fice for the Coordination of Humanitarian Affairs (UNOCHA) reported that two people had died and northern Tanzania (Fig. 7.24d). The rest of the GHA region experienced above-normal mean four were injured. According to Niger’s government, 219 houses were destroyed and over 1000 people were temperatures. left homeless in Gabagoura and other villages around Niamey. a — f r i c a a s t e r n e 3 ) G. Mengistu Tsidu Eastern Africa, also known as the Greater Horn of Africa (GHA), is a region comprised of South Sudan, Sudan Republic, Ethiopia, Somalia, Eritrea, Kenya, Uganda, Rwanda, Burundi, and Tanzania. Despite its location across the equator, the region has a relatively cool climate due to its generally high altitude. Some parts of the region are also characterized by bimodal seasonal rainfalls. In general, the GHA experienced above-average temperatures in 2017. (i) Temperature The December 2016–February 2017 (DJF) mean temperature was below normal to normal over central Kenya, the Ethiopian highlands, eastern Ethiopia, much of northern Somalia, and the northwestern Sudan Republic (Fig. 7.24a). Above- normal anomalies, up to +3°C, were observed over the rest of the GHA. During MAM, the GHA remained warmer than normal, except for the northern Ethiopian highlands, northwestern Sudan Republic, part of central Kenya, north-central and southwestern - . 7.24. Eastern Africa seasonally averaged mean temperature anoma F ig Tanzania which had normal to lies (°C; 1981–2010 base period) for (a) DJF 2016/17 and (b) MAM, (c) JJA, and (d) SON 2017 (Source: ERA-Interim.) below-normal temperatures (Fig. | S216 AUGUST 2018

237 (ii) Precipitation its southeastern lowlands, South Sudan, and southern Sudan Republic receive their main rainfall during The northern half of Uganda, northwestern and northeastern Kenya, and adjacent western Somalia JJAS. Normal to above-average rainfall, ranging from 110% to 200% of the seasonal mean, dominated the received normal to above-average rainfall, whereas region in 2017, including unseasonal rain over the southern Uganda, most of Kenya, Tanzania, Burundi, and Rwanda received 50%–90% of the base period southern half of GHA (Fig. 7.25c). Dry conditions mean during DJF (Fig. 7.25a). Some isolated pockets prevailed over most of Kenya, central Tanzania, and in Tanzania and Kenya received less than half of their coastal Somalia during SOND, which is the climato - logical rainy season in this area (Fig. 7.25d). normal DJF rainfall. Rainfall during MAM was below normal over southern Ethiopia and adjacent southern (iii) Notable events and impacts Somalia, all of Kenya, northern Uganda, and most of - Tanzania (Fig. 7.25b). Southeastern Tanzania, in par Heavy rainfall recorded in most parts of the region throughout the 2017 rainy seasons caused f looding. ticular areas along the coast, received above-normal For example, eastern Kenya and Tanzania had steady rainfall during MAM. Ethiopia, with the exception of torrential rain in May. As a result, Mombasa recorded 235 mm on 9 May, which led to f lash f looding. According to news outlets, at least nine people perished. There were also heavy rains in mid-May, with a number of stations in western and central Ethiopia recording 49 mm and above (e.g., Gore: 59.9 mm; Jimma: 53 mm; Addis - Ababa Bole: 49 mm). The subse quent f looding led to a death in the Gambella region of Ethiopia on 18 May. JJAS rainfall was also notably heavy over Sudan and Ethiopia. Resulting f loods af - 00 people in fected more than 53 0 the Gambella and Oromia regions during August and September, according to a UNICEF humani - tarian report. 4 ) o u t h e r n a f r i c a — s G. Mengistu Tsidu, A. C. Kruger, and C. McBride Southern Africa comprises the Republic of South Africa, Angola, Botswana, Zimbabwe, Namibia, - Malawi, Zambia, Lesotho, Swa ziland, and Mozambique. The rainfall assessment is based on rainfall from CHIRPS, and in situ observations from South Africa. (i) Temperature Above-normal temperatures prevailed across Angola, . 7.25. Eastern Africa seasonal total rainfall anomalies (% of normal; F ig northeastern Zimbabwe, western 1981–2010 base period) for (a) DJF 2016/17 and (b) MAM, (c) JJAS, and (d) SOND 2017 (Source: CHIRPS.) Namibia, western South Africa, | S217 AUGUST 2018 STATE OF THE CLIMATE IN 2017

238 Mozambique, and northeastern South Africa remained lower than normal (Fig. 7.26a). Normal to below-normal mean temperatures expanded to include much of Zimbabwe in MAM (Fig. 7.26b). During JJA, warm conditions dominated across the region, with the exceptions of isolated pockets in Botswana, Zimbabwe, Zambia, Mozambique, and southern strips of South Africa, where normal mean temperature conditions were observed (Fig. 7.26c). However, during SON, southern Mozambique and adjoining areas in Zambia experienced temperature anomalies exceeding −1°C, with normal mean temperature prevailing over most of Zimbabwe, Zambia, northern Botswana, northeastern South Africa, and Swaziland. The rest of the region remained moderately warmer than normal (Fig. 7.26d). For the year as a whole, warmer-than-normal conditions throughout 2017 prevailed across the southern part of the region (spatial - . 7.26. Southern Africa seasonally averaged mean temperature anoma ig F lies (°C; 1981–2010 base period) for (a) DJF 2016/17, (b) MAM, (c) JJA, and representation not shown), as (d) SON 2017 (Source: ERA-Interim.) evident from the annual mean temperature anomalies of 26 climate stations in South Africa, which averaged 0.48°C above average (Fig. 7.27). (ii) Precipitation Above-normal rainfall, exceeding 150% of average, was observed over Botswana and adjoining eastern Namibia, eastern Angola, Zimbabwe, Zambia, west - ern Mozambique, and most of South Africa from December 2016 to February 2017. However, most of Angola, western Namibia, western South Africa, F . 7.27. Annual mean temperature anomalies (°C; ig and eastern Mozambique experienced below-normal 1981–2010 base period) of 26 climate stations in South rainfall (Fig. 7.28a). Wet conditions persisted through Africa, as indicated on inset map, for 1951–2017. Red MAM only over central parts of the region, namely line represents the linear trend. (Source: South African - northern Botswana, northern Namibia, Zambia, Mo Weather Service.) zambique, and southern Zimbabwe. Below-normal conditions—as low as 20% of normal—prevailed and Mozambique from December 2016 to February over the rest of the region (Fig. 7.28b). The JJA period 2017. In contrast, northeastern Namibia, all of is typically dry over the region. Compared to this Botswana, Southern Zimbabwe, Zambia, southern | S218 AUGUST 2018

239 - since at least 1951. A Standard ized Precipitation Index (SPI) analysis of South Africa (not shown) indicates that almost the whole Western Cape Province, including the adjacent areas of the northern and eastern Cape Provinces, experienced what can be described as moderate to extreme drought conditions over an extended period of time, (i.e., longer than two years). In addition, an analysis of Cape Town in particular shows that the only other comparable dry period was around 1973/74, with a probability of occurrence lower than 3%. (iii) Notable events and impacts Tropical cyclones impacted the region during early Febru - ary. Cyclone Carlos was active - from 3–10 February. On 13 Feb ruary, Cyclone Dineo formed in the Mozambique Channel and made landfall in southern Mo - zambique on 15 February. The cyclone subsequently tracked westwards, leaving significant f lood-related damage in south - F ig . 7.28. Southern Africa seasonal total rainfall anomalies (% of normal; ern Mozambique and southern 1981–2010 base period) for (a) DJF 2016/17 and (b) MAM, (c) JAS, and (d) Zimbabwe. The impact of Dineo SON 2017 (Source: CHIRPS.) - was also felt as far west as Bo tswana, where daily rainfall of 70 mm was observed at baseline, most of Mozambique received above-normal rainfall (Fig. 7.28c), whereas the rest of the region Kgomokitswa on 19 February and 136 mm at Lobatse remained drier than normal. The wet conditions in on 20 February. These amounts accounted for 70% Mozambique in JJA expanded to Zimbabwe, Zambia, and 66%, respectively, of the total February rainfall and isolated areas in the northeastern part of South at each station. As a result, the Gaborone dam filled Africa by SON (Fig. 7.28d). up due to the intense rains, marking an abrupt end - Analysis of annual total gauge rainfall over South to the multiyear hydrological droughts that have af Africa confirms that large parts of the country fected the region. Other notable but localized events during early received near-normal rainfall. The most notable exception was the region including the largest part summer includes stormy weather over South Africa. of the Western Cape and the western parts of the - In October, a cut-off low system moving southeast ward from northwestern South Africa triggered large Northern Cape, which received less than 75% of its normal annual precipitation (figure not shown). thunderstorms, especially over western Gauteng province, with several tornado sightings. A subse The western Northern Cape was especially dry with - quent statement by the South African Weather Ser - station averages indicating less than half of normal vice reported that severe thunderstorms with heavy rainfall, in agreement with the analysis of seasonal downpours, strong damaging winds, and large hail rainfall from CHIRPS. 2017 was the driest year over hit parts of eastern North-West, Gauteng, eastern Free the last three-year period in the southwestern Cape | S219 AUGUST 2018 STATE OF THE CLIMATE IN 2017

240 State, KwaZulu-Natal, Mpumalanga, and Limpopo region. Figure 7.29 shows temperature and rainfall provinces on 9 and 10 October. Areas that were most anomalies for selected areas. affected were Mogale City, the city of Johannesburg, and Ekurhuleni Metropolitan municipalities. There (i) Temperature were two sightings of tornadoes in Ruimsig (adjacent In Madagascar, the annual mean temperature, to Roodepoort and Krugersdorp) and Eloff, near Del - - based on fifteen stations, in 2017 was 24.1°C; the an nual anomaly was +0.6°C. It was the fifth warmest mas (Mpumalanga), which caused extensive damage year on record since 1978. June recorded the highest to property. Elsewhere in the Free State, a tornado was - monthly anomaly, around +1.4°C. The highest maxi observed near Bethulie. Golf ball– to tennis ball–sized mum temperature was observed in Antsohihy on 10 hail was also reported near Krugersdorp. The system - moved rapidly east, affecting KwaZulu-Natal from 10 October (37.7°C), and the lowest minimum tempera ture (2.0°C) was recorded three times in Antsirabe: October, with severe urban f looding and high winds on 13, 15, and 19 July. resulting in loss of life. In 24 hours, Durban recorded At Seychelles International Airport, monthly 108 mm of rain, where 65 mm fell in less than an hour. Similarly, Virginia in KwaZulu-Natal received 142 mean maximum temperatures were slightly below normal during the first four months of 2017. Above- mm of rain, where 89 mm occurred within an hour. −1 normal temperatures prevailed from June through Maximum sustained wind speeds of 75 and 78 km h were reported in Durban and King Shaka airport to November. The annual mean maximum temperature was 30.3°C and ranks as the seventh highest since the north, respectively. records began in 1972. (The warmest year on record isl and countries — cean o ndian i estern w 5) is 2009 with an annual mean of 30.6°C.) The 2017 G. Jumaux, C. L. Rakotoarimalala, M. Belmont, and extreme daily maximum temperature was 33.7°C K. R. Dhurmea on 22 April. Monthly mean daily minimum tem - This region consists of several island countries: perature departures from the long-term means varied Madagascar, Seychelles, Mayotte (France), La from +0.3° to +1.2°C. The extreme daily minimum Réunion (France), Mauritius, and Rodrigues (Mau temperature for 2017 was 21.6°C, recorded on 6 July. - Finally, the annual mean temperature was +0.4°C rit ius). above normal, ranking as the fourth highest since Overall, the 2017 mean temperature for the region 1972. The highest monthly temperature deviation was well above normal, while precipitation was mixed occurred during October (+1.2°C). across the region. It was the warmest year on record For Mayotte Island, 2017 was the warmest year in the Mascarenes archipelago (Réunion, Mauritius, Rodrigues) because of strong SST anomalies in this since records began in 1961, with an annual mean temperature anomaly of +1.0°C at Pamandzi Airport. Temperatures during January and February were slightly above normal. From March to December temperatures were well above the reference base pe - riod, often more than +1.2°C from June to November. ig . 7.29. Mean annual temperature anomalies (°C; F squares), annual rainfall anomalies (% of average; circles), and their respective deciles for the western . 7.30. Annual mean temperature anomalies (°C; ig F Indian Ocean islands countries in 2017. (Sources: Météo 1981–2010 base period) for Réunion Island, 1968–2017. France; and Meteorological Services of Madagascar, (Source: Météo-France.) Seychelles, and Mauritius.) | S220 AUGUST 2018

241 (ii) Precipitation For Réunion Island, the annual mean temperature, For Madagascar, annual precipitation, based on 21 based on three stations, was +0.9°C above normal in stations, was 107% of normal in 2017. It was the sixth - 2017, ranking as the highest since 1968, far exceed wettest year on record since 1978. Nine months were ing the previous record of +0.7°C in 2011 (Fig. 7.30). above normal, with November the relative wettest Temperatures during January and February, the at 175% of normal for the month. More stations in warmest months of the year climatologically, were western Madagascar were below normal than in the slightly above normal. From March to December they eastern part (Fig. 7.31). The highest percent of normal were well above the reference base period, often by (233%) was recorded at Mananjary and the lowest at more than +1.0°C. Morombe (30%). The highest accumulated precipi - Over the island of Mauritius, the annual mean - maximum temperature anomaly based on two sta tation within a 24-hr period in 2017 was 215.4 mm tions was +0.6°C, and the annual mean minimum recorded in Sambava on 7 March, the eighth wettest day there since 1961. temperature anomaly was +1.5°C, indicating a greater departure during nighttime. July and August were For Seychelles, five months recorded higher-than- among the warmest on record since 1951. The an - normal rainfall during 2017. The second half of the nual mean temperature anomaly over the island was year was dominated by persistent negative anomalies +1.0°C. This makes 2017 the warmest on record since - that impacted the amount of rain received at the be ginning of the rainy season. The outer islands were 1951. Similar observations were made at Rodrigues (Pointe Canon) with an annual mean temperature worst affected during that period. (The rainy season anomaly of +1.2°C. usually starts mid-October and ends mid-April.) The total rainfall amount recorded for the year is 2146.2 mm, which is 91% of the long-term mean, ranking as 18th driest since records began in 1972. The total number of rain days was 195, which is slightly below normal. For Mayotte Island, the annual rainfall amount based on two stations was 118% of average, ranking eighth wettest since 1961. January, May, and October were drier than normal, whereas February, April, and December were among the rainiest on record. For Réunion Island, the annual rainfall based on 34 stations was 91% of the long-term mean, ranking 18th driest since 1969. January was the second driest on record, associated with a late rainy season onset in early February. Total precipitation during the rainy season (January–April) was only 76% of aver - age, ranking ninth driest. During May to November, typically the driest months of the year, the rainfall amount was 133% of average, ranking fifth raini - est. At Plaine des Fougères, 215 mm of rain fell fell in three hours on 29 August, which is very unusual during this season. The year started very dry both at Mauritius and Rodrigues (Fig. 7.32). Dry conditions persisted over Rodrigues until March. May, usually a dry transition month, was very wet on both islands; it was the seventh wettest on record for both Mauritius and Rodrigues. Heavy rainfall and widespread f looding affected both - islands. Winter months (May–October) yielded a posi tive rainfall anomaly over Mauritius. December 2017 was the second driest in Mauritius since 2010 and . 7.31. Annual total precipitation anomalies (% of F ig second driest in Rodrigues (Pointe Canon) since 2003. average; 1981–2010 base period) for Madagascar in 2017. (Source: Madagascar Meteorological Services.) The total annual rainfall over Mauritius amounted to | S221 AUGUST 2018 STATE OF THE CLIMATE IN 2017

242 provided by WMO RA VI Regional Climate Centre on Climate Monitoring (RCC-CM; www.dwd.de /rcc-cm). Anomaly information has been taken from Figs. 7.34–7.37 when national reports are not available. 1) o verview Based on the CRUTEM4 dataset (Jones et al. 2012) dating to 1851, Europe (35°–75°N, 10°W–30°E) observed its fifth warmest year on record with an anomaly of +1.3°C; its five warmest years have all oc - curred since 2011 (Fig. 7.33). NOAA data (not shown) F ig . 7.32. Mean monthly total precipitation anomalies (mm; 1981–2010 base period) over Mauritius in 2017 also ranks Europe as fifth warmest for 2017. (Source: Mauritius Meteorological Services.) - Local temperature anomalies varied mostly be - tween +1° and +2°C and were homogeneously distrib uted across Europe, with local areas in the Ukraine 2110 mm (105% of average). The total annual rainfall amounted to 968 mm at Rodrigues (88% of average). and central Spain above +2°C (Fig. 7.34). While large parts of Europe had near-normal (iii) Notable events and impacts precipitation for the year on average, some regions like northeastern Germany, northern Poland, and Tropical cyclone Carlos passed 130 km off the western Russia recorded above-normal precipitation western coast of Réunion Island on 7 February. A −1 totals up to 167%. This contrasted with 60%–80% was recorded at maximum wind gust of 37 m s of normal precipitation on the Iberian Peninsula, in Bellecombe, and 934 mm of rain was recorded at southern France, and Italy. Especially noteworthy is Grand-Ilet within a 72-hr period. The agricultural - - the Middle East, with 20%–60% of normal precipita sector sustained losses up to 4.9 million U.S. dol tion (Fig. 7.35). lars. Winter 2016/17 was exceptionally mild over In Madagascar, March 2017 was marked by Cyclone Enawo which formed in the Indian Ocean. Scandinavia with temperature anomalies reaching It tracked across the island from 7 to 9 March, more than +4°C, whereas the southern Balkan states, Greece, and Turkey recorded widespread below- producing three-day precipitation totals of 224 mm at Sambava, 210 mm at Antsohihy, 184 mm at average temperature anomalies down to −2°C (Fig. Antananarivo, 159 mm at Fianarantsoa, 291 mm at 7.36a). The 500-hPa heights featured above-average Mananjary, and 178 mm at Taolagnaro which were, heights which allows for an anomalous southwesterly f low of mild marine air masses into northern Europe respectively, 82%, 86%, 112%, 113%, 81%, and 101% of the normal monthly precipitation for March at (dotted in Fig. 7.36a). In particular, January was a cold month over much of central and southeastern each location. Enawo led to 81 fatalities, injured 250 people, and caused significant f looding. Eastern Europe. With respect to precipitation, winter in Madagascar was the most affected. f. Europe and the Middle East— P. Bissolli, M. Demircan, J. J. Kennedy, M. Lakatos, M. McCarthy, C. Morice, S. Pastor Saavedra, M. R. Pons, C. Rodriguez Camino, B. Rösner, S. Sensoy, S. Spillane, K. Trachte, and G. van der Schrier - For this section, 1961–90 is used as the base pe riod for temperature, and 1981–2010 is used as the base period for precipitation, as described in Figs. - 7.33–7.37, unless otherwise specified. European coun tries conform to different standards applied by their individual national weather services, and their specific base periods are noted throughout the subsections as needed. All seasons mentioned in this section refer to ig . 7.33. Annual average land surface air temperature F the Northern Hemisphere. More detailed information - anomaly for Europe (35°N–75°N, 10°W–30°E) rela can be found in the Monthly and Annual Bulletin on tive to the 1961–90 base period. [Source: CRUTEM4 dataset (Jones et al. 2012.)] the Climate in RA VI – European and the Middle East, | S222 AUGUST 2018

243 . 7.34. Annual mean air temperature anomalies (°C, ig F . 7.35. European precipitation totals (% of 1981–2010 ig F 1961–90 base period) in 2017. (Source: DWD.) average) for 2017. (Source: DWD.) Europe was dry with values ranging mostly from 40% to 60% of normal, except for the Norwegian coast, which - was wet with some locations exceed ing 167% of normal (Fig. 7.37a). This was a consequence of both the strong Icelandic low and Azores high (NAO +1.22), favoring westerly f lows induc - ing a mild and wet winter in northern Europe and dry winter conditions over the more southern parts. In spring, along with above-average 500-hPa heights situated over central Europe (dotted in Fig. 7.36b), above- normal temperatures were measured all over Europe, up to +4°C in Spain. However, an unusually warm March was followed by a cold late April, and severe late frosts led to agricultural - losses across many European coun tries. April and May temperatures were well below normal in northwest Russia - and northern Scandinavia, contribut ing to the highest May snow cover extent in this area since 1985. While ig F . 7.36. Seasonal anomalies (1961–90 base period) of 500-hPa eastern Europe, except for central geopotential height (contour, gpm) and air temperature (shading, Ukraine, showed normal to slightly - °C) using data from the NCEP/NCAR reanalysis and DWD, respec above-normal spring precipitation tively, for (a) DJF 2016/17, (b) MAM 2017, (c) JJA 2017, and (d) SON totals, most of the Iberian Peninsula, 2017. Dotted areas indicate regions where 500-hPa geopotential Italy, and the Benelux countries contin - is higher (lower) than the 95th percentile (5th percentile) of the 1961–90 distribution, while hatched areas represent the correspond - ued to be drier than normal. ing thresholds but for air temperature. | S223 AUGUST 2018 STATE OF THE CLIMATE IN 2017

244 quent cyclonic situations, autumn was characterized by some severe storms - accompanied by extreme wind veloci ties and heavy precipitation leading to - widespread severe damage to infra structure and f loods in northeastern Germany, Poland, Slovakia, and the Czech Republic. Therefore, eastern Europe received precipitation up to 250% of normal, while on the Iberian Peninsula precipitation totals were mostly well below 60% of the normal. With temperature anomalies reaching +4°C or more and above- normal precipitation of at least 125%, the year ended very mild and wet for - northeastern Europe, while tempera ture anomalies of around −1°C and below-normal precipitation of locally less than 40% in the Mediterranean region brought a rather cold and dry December. urope e western and entral 2) c This region includes Ireland, the . 7.37. Seasonal anomalies for 2017 (1981–2010 base period) of sea ig F - United Kingdom (UK), the Nether level pressure (hPa) from NCAR/NCEP reanalysis (contours) for (a) lands, Belgium, Luxembourg, France, DJF 2016/17, (b) MAM 2017, (c) JJA 2017, and (d) SON 2017. Colored shading represents the percentage of seasonal mean precipitation for Germany, Switzerland, Austria, Po - 2017 compared with the 1981–2010 mean from the monthly Global land, Czech Republic, Slovakia, and - Precipitation Climatology Centre (Schneider et al. 2015) dataset. Dot Hu nga r y. ted areas indicate regions where SLP is larger (lower) than the 95th percentile (5th percentile) of the 1981–2010 distribution, while hatched (i) Temperature areas represent the corresponding thresholds but for precipitation. Overall, western and central Eu - rope were warmer than normal for 2017, with many With a similar circulation pattern featuring the development of a high pressure bridge during the countries reporting a year ranking among their ten summer months, temperatures in central Europe warmest [UK +0.7°C (fifth; since 1910), Switzerland continued to be high, with anomalies of up to +5°C +0.8°C (sixth; since 1864), Austria +0.8°C (eighth; measured in Italy and the Balkan states (former - since 1767), France +0.8°C (fifth; since 1900); refer Yugoslavia). Characteristically for this blocking ence period is 1981–2010 for all four]. pattern situation, summer in Italy and the Balkan Winter 2016/17 was mostly mild or around nor - mal. In February, several storms in southwestern states was dry, with precipitation totals as low as 20% of normal while northern Europe, western Europe brought warm air masses to central Europe, leading to exceptionally high temperatures, with Russia, Greece, and western Turkey recorded above- France reporting its warmest February and several average precipitation totals of up to 250% under the stations in Switzerland observing record-breaking inf luence of the surrounding lows. Italy in particular daily maximum temperatures in their more than experienced a massive heat wave with record- breaking temperatures and extreme drought. 100-year measurement series. During autumn, temperatures in central Europe This warm episode continued in March, with - many new monthly records set in France, Austria, remained near-normal, while eastern Europe experi Belgium, and Germany. For the first time in station enced above-normal temperature anomalies of up to history (since 1837), a monthly mean temperature +2°C. The Iberian Peninsula was under the inf luence of 9.0°C was measured in Graz, 3.7°C above the of above-average 500-hPa height anomalies, which led - 1981–2010 normal. In Vienna, the high value of to temperatures up to +3°C above normal. Due to fre | S224 AUGUST 2018

245 est, with October (+2.6°C anomaly; 1981–2010 base March monthly means (in 1990) was surpassed by period) fourth warmest since 1901. On 7 November, +0.1°C. Similarly, in France, record-high monthly a station in Freiburg (Germany) recorded 23.2°C, its - mean temperatures of 10.1°C and 10.9°C were mea highest daily maximum temperature for November. sured at the stations in Dunkerque and Le Mans, The year ended with close to or moderately above- respectively. In Germany, at the station Kitzingen, a normal temperatures for the United Kingdom and maximum temperature of 25.6°C was measured on northern France and as much as +2° to +4°C above 30 March, a temperature not generally observed so - early in the year. At Swiss stations La Chaux-de-Fonds normal east of Germany, while Switzerland, particu and Meiringen, records of +4.0°C and +4.1°C above larly the south, and the Mediterranean coast of France were colder than normal in December. their 1981–2010 March normal were reached. After a late-night frost episode in many countries (e.g., (ii) Precipitation Germany, Switzerland, Austria, United Kingdom, the Most of central Europe recorded near- to above- Netherlands) during the second half of April, a low normal precipitation—up to 167% in northeastern pressure system northwest of Ireland in May led to Germany and northern Poland; however, France, had the advection of warm dry air from Africa, leading slightly below-normal precipitation and even expe - to many more daily maximum temperature records rienced exceptional drought in the region Provence- in France (35.1°C at station Biscarrosse), the United - Kingdom (29.4°C at station Lossiemouth), the Neth Alpes-Côte d’Azur, with an average deficit of less than 60% of normal between May and November, erlands (33.5°C at station Volkel), and Austria (35.0°C the lowest cumulative rainfall since 1959. at station Horn). Summer was characterized by exceptional heat, In winter 2016/17, central Europe was much drier than normal with 20%–80% of normal precipita - when high pressure conditions dominated. June was the warmest month since 1901 in the Netherlands tion; except for Poland, which had above-normal precipitation of around 125%. Regionally, Switzerland (together with 1976) as well as the second warmest month in Austria (251 year series), France (after recorded its driest winter for the last 45–55 years. In the mountains in the canton of Ticino, an unprec - June 2003), and Switzerland (after June 2003). In edented 14 cm average of snow depth was the lowest Belgium, June 2017 was also one of the warmest, value since the beginning of measurements. - close to the record of June 2003. In August, sub tropical warm air reached as far north as Bavaria In spring, western Europe continued to be drier than normal on average, with eastern France, the - and Baden-Württemberg (southern Germany), lead ing to new records in those regions. The station in Benelux countries, and western Germany having a Emmendingen-Mundigen, Germany, experienced 11 precipitation deficit as low as 40%–60% of normal; conversely, parts of Poland locally received up to 167% consecutive hot days (daily maximum temperature equal or above 30°C), and the Czech Republic had of normal. In March, a deep low pressure system over the United Kingdom triggered a foehn storm in the daily maximum temperatures as high as 38.3°C. In Figari (Corsica, France), a record high temperature northern Alps, bringing up to 100 mm of precipita - tion within three days in the southern Alps. Austria of 42.7°C was reached. In July and August, Hungary and Switzerland reported intense snowfall in late reported a record 27 days of heat wave conditions and April, with 35 cm accumulation within two days in Budapest experienced a record-breaking 34 tropical nights (minimum temperature ≥ 20°C; Klein Tank et St. Gallen (Switzerland). al. 2009) in the series since 1901. Frequent Atlantic cyclones during the summer brought above-normal precipitation to most of the In autumn, anticyclonic conditions prevailed over northern region. Northern Germany and northern western Europe, which on average made western and Poland received as much as 250% of normal. Often central Europe slightly warmer than normal (around - +1°C) with some exceptions like Austria, where tem accompanied by thunderstorms, the cyclones brought perature anomalies were below average by 1.5°C in heavy rainfall that led to f looding and widespread traffic impact. Highest daily total records were bro - September as well as in France where, although the mean monthly temperature anomaly was +0.9°C, ken, for example, at station Berlin-Tegel where 197 locally daily maximum temperature anomalies mm on 29 June was measured (previous record was - (1981–2010 base period) of −5°C were measured. Swit below 90 mm). Station Shannon Airport (Ireland) zerland reported a September monthly mean anomaly reported its wettest July in its 71 year record with of −1.6°C below its 1981–2010 average. In contrast, 133.6 mm (203% of normal). - autumn in the Netherlands was among its ten warm | S225 AUGUST 2018 STATE OF THE CLIMATE IN 2017

246 In autumn, southern France, under the inf luence up. During Ophelia, an individual wave height record of above-normal 500-Pa heights centered over the of 26.1 m was set at the Kinsale gas platform off the Iberian Peninsula, experienced a rain deficit as low Cork coast (Ireland). Several storms affected central Europe from 17 to as 20% in the Mediterranean region, while eastern −1 , 18 November, with wind gusts of more than 48 m s Europe received up to 250% of normal precipitation impacting traffic and causing major damage to trees due to several low pressure systems. At the end of and buildings. the year, precipitation was distributed unevenly, but with above-normal precipitation for most of central ordic n he 3) t Europe (exceptions are the coast of southern France, ountries c altic b the and eastern Germany, and the border region of Poland This region includes the Nordic countries Iceland, and the Czech Republic, which received below- Norway, Denmark, Sweden, and Finland, and the normal precipitation). Baltic countries Estonia, Latvia, and Lithuania (iii) Notable events and impacts (i) Temperature A severe storm affected France on 6–7 March, with Temperatures across the Nordic and Baltic area in −1 i n Br it ta ny. 2017 were mostly higher than normal, between +1° peak gusts reaching 54 m s - and +2°C. Estonia, Finland, and Denmark had posi On 18–19 May, thunderstorms over Germany (low “Dankmar”), accompanied by hail and heavy rainfall tive anomalies of +1.5°, +1.3° and +1.2°C, respectively. of more than 36.3 mm (Bad Bibra) within 1 hour, led Winter 2016/17 was exceptionally mild due to the inf luence of above-average 500-hPa heights (Fig. to f looding in several cities. Local intense rainstorms occurred in France dur - 7.36a), with +2°C anomalies in the south and up to ing 29–31 May, with 24-hr totals often exceeding 20 +5°C in northern Sweden and Finland. Denmark mm: 53.9 mm in Genouillax, 58.9 mm in Muret, 72.1 recorded its fifth and fourth highest daily minimum and maximum temperatures in December, respec - mm in Castelnau-Magnoac, and 80 mm in Chateau - tively, since 1953. During long-lasting foehn winds ponsac. During the same period, in Switzerland a violent storm with heavy rain and hail the size of golf in January, a daily mean temperature of 13.8°C was balls was reported at the station in Thun; the storm measured at station Sunndalsøra (Norway) on 25 brought the highest daily precipitation amount (59.6 - January, which was the highest daily mean tempera ture ever recorded in January by a weather station in mm) there since the start of measurements in 1875. Norway and is a value commonly measured around In summer, severe hailstorms impacted western the beginning of July. During February, 12 stations in and central Europe. Perhaps the most remarkable was cyclone Zlatan, which developed over England Norway observed new daily maximum temperature records as well as extreme anomalies at the Arctic on 19 July. Moving eastward, it affected the eastern half of France, Switzerland, Austria, Germany, the station of Svalbard Lufthavn, with a monthly mean temperature of +9.3°C. Czech Republic, and Poland with heavy rain and On 26 March, under the inf luence of southwesterly hail causing damage, especially in Germany (North f lows, daily maximum temperatures of 20°C or more Rhine-Westphalia, Station Cologne measured 48.8 −1 −1 and gusts of 26 m s mm h were measured at stations Akershus, Oslo, Hedmark, ) with closed roads and Buskerud, and Telemark in Norway, the first known traffic delays. The airport in Cologne was closed for occurrence of such high temperatures in March. In 90 minutes. In Austria, between 4 and 6 August, intense precipi May, the northeastern part of the North suffered from - a cold wave which led to several record below-average tation in Lungau (state of Salzburg) and Obersteier - mark (state of Styria) led to several landslides, causing anomalies in Latvia (e.g., station Rezekne −4.7°C, Mersrags −6.1°C). Finland was also affected, with an estimated damage of more than 20 million Euros anomalies between −1° and −3°C for May across the ($25 million U.S. dollars) to the local road network. whole country. Temperatures for Lithuania in May In October, three storms (ex-Hurricane Ophelia, also were slightly below normal, and frost days were Storm Brian, and Storm Herwart), with extreme wind −1 - even recorded. In contrast, a new record high maxi or more, brought much damage to speeds of 40 m s - the United Kingdom, Ireland, Germany, the Nether mum temperature of 31.8°C was measured at Sigdal lands, France, Austria, Czech Republic, Poland, and - Nedre Eggedal (Buskerud) in Norway on 27 May. Slovakia, with falling trees killing at least seven people Under the inf luence of below-average 500-hPa in Germany and three in the UK, as well as road and anomalies, summer in the Baltic States on average had railway blockings affecting traffic for days during clean slightly below-normal temperatures with anomalies | S226 AUGUST 2018

247 ranging from −0.5° to −2.5°C in the east of Finland. test). At the end of 2017, precipitation totals continued Lithuania observed an unusually cold July (−1.4°C), to be above normal for almost all of the Nordic and while the rest of the summer was closer to average. Baltic States. Overall, the summer was rather cold without any hot spells, which was apparent in the maximum tempera - (iii) Notable events and impacts On 12 August, widespread thunderstorms left tures; for example, Sweden, with a daily maximum 0 00 households without power in southern temperature of only 28.0°C, had its coolest summer 50 since 1922. Finland. During autumn, temperatures were higher than At the end of September, following heavy precipita - tion and warmth, major damage was caused by f loods normal in all Nordic and Baltic countries, with and landslides in southeastern and eastern Iceland. around +1°C anomaly. The entire north was under the In Lithuania, three microscale extreme heavy rain inf luence of above-average 500-hPa height anomalies, - events were observed in summer. On 12 July, precipi which also led to a new record high sea level air pres - tation totals exceeded 80 mm in 12 hours. sure of 1044.1 hPa at Lycksele and Åsele in northern Between 22 and 24 November, Storm Ylva caused Sweden. November and December were especially −1 wind gusts as high as 47.5 m s mild. It was the warmest autumn in Denmark since at station Narvik- Fagernesfjellet (Nordland) in Norway. Another 1984, with a new maximum temperature record storm, “Birk”, brought heavy precipitation to Horda - of 17.1°C on 2 November at Kjevik (Kristiansand, Vest-Agder, Norway), as well as at station Yngør land and Rogaland counties, with a maximum daily precipitation total of 127.5 mm at Gullfjellet (highest Lighthouse (Tvedestrand, Norway) with a reading of mountain in Bergen, Hordaland) measured on 23 14.4°C. With temperatures up to +5°C above normal in eastern Finland and between +1°C and +4°C for December. most of the Baltic States, 2017 ended rather warm. berian p eninsula 4) i (ii) Precipitation - This region includes Spain and Portugal. Anoma - Except for Iceland and most regions of the north lies refer to a reference period of 1981–2010 for Spain and 1971–2000 for Portugal. ern Baltic States, annual precipitation totals were around normal to above normal. Norway experienced its sixth wettest year since records began in 1900. (i) Temperature Temperatures for 2017 on the Iberian Peninsula During winter 2016/17, all Baltic states and most Nordic countries had a precipitation deficit; only were well above normal, by +1° to +3°C in central Norway received a surplus of up to 167% of normal, Spain. Portugal recorded its second warmest year with northern Norway having its wettest winter on on record with +1.1°C anomaly compared to the 1971–2000 average and a new average annual maxi - record but only the 15th wettest for the country as a whole. This deficit continued in spring, especially in mum record temperature of 22.82°C, +2.32°C above May, with Lithuania observing only 24% of its normal normal, since records began in 1931. With an anomaly precipitation and Latvia having its sixth driest May, of +1.1°C, Spain recorded its warmest year since its with a nationally-averaged total of 21.4 mm. series began in 1965, exceeding the previous record Summer in northern parts of the Baltic was drier of 2011, 2014, and 2015 by +0.2°C. More than thirty than normal; however, Lithuania, Denmark, and individual stations in Spain (almost one-third of - all principal stations) surpassed their annual mean Norway received above-normal precipitation. Lithu ania reported a wet summer, particularly notable for temperature records in 2017. July (almost 150% of normal), due in part to several During winter 2016/17, January had below-average severe storms with heavy rainfall. Norway, with 130% - anomalies of around −0.5°C. This situation was as of normal precipitation, recorded its third wettest sociated with an inf low of cold air from the north summer since 1900. caused by a high pressure system located above the Prevailing westerlies in autumn brought well Canaries ranging as far north as Iceland and a low above-normal precipitation to the Nordic and south - pressure system ranging from Scandinavia to the ern Baltic states. Lithuania and Latvia each reported Mediterranean. Some stations in Portugal measured a record (since 1961) wet season with up to 176% and exceptionally low daily minimum air temperatures 135% of normal precipitation. Latvia, with a seasonal and, on 19 January, some even recorded absolute total of 313.5 mm, reported its second wettest autumn daily minimum records. February, in contrast, was in the last 94 years (for some stations even the wet - | S227 AUGUST 2018 STATE OF THE CLIMATE IN 2017

248 clearly warmer than average, with anomalies as high since 1931 with an average of 23 mm corresponding as +1.6°C in Spain. to 40% of normal, summer in Spain was slightly wet - ter than normal (107% of normal). Notably, much of Above-normal 500-hPa height anomalies situated the precipitation can be attributed to storms during over central Europe led to an extremely warm spring, which new daily maximum precipitation total records with anomalies of +1.7°C, making 2017 the warmest - spring since 1965 in Spain, exceeding the previous re were measured. Autumn was dominated by above-normal 500-hPa cord of 2011 by 0.1°C. In Portugal, April was the fifth and May the third warmest month since the record - height anomalies centered above the Iberian Penin sula. As a result, both Spain and Portugal recorded began in 1931. Additionally, the highest and second below-average precipitation totals, making 2017 the highest average maximum temperatures, respectively, driest autumn on record in Spain and second driest since 1931 were measured for each of these months. autumn since 1931 for Portugal (only 35% of normal The warmth continued into summer, with Spain precipitation). September was the driest such month observing its second hottest since 1965, at +1.6°C of the last 87 years in mainland Portugal, with an above normal, behind only 2003. June was particular - ly warm, with a monthly anomaly of +3.0°C. During average precipitation of 2 mm (5% of the normal). In 12–16 July, the highest daily maximum temperatures many places, no precipitation was measured at all. It of that summer were observed: 46.9°C at Córdoba was followed by the driest October of the last 20 years in Portugal. By the end of October, 25% of Portugal Airport, 45.7°C at Granada Airport, and 45.4°C at Badajoz Airport. Eight stations in the southern half suffered from severe and 75% from extreme drought. This extended period of drought led to widespread of the peninsula observed their highest maximum wildfires in Portugal, with new records in the size of absolute temperature of any summer month. With a area burned. December marked the ninth consecutive mean temperature of 22.70°C, summer in Portugal month with below-normal precipitation for Portugal. was +1.43°C warmer than the 1971–2000 normal, its ninth warmest summer since 1931. (iii) Notable events and impacts While the autumn anomaly for Spain was only A heat wave lasting 17–18 days in the northern +0.8°C above normal, October was the warmest since and central regions of Portugal and 11–12 days in 1965 (+2.6°C). Although it was only the fifth warmest the remainder of the country occurred between 7 - autumn (with respect to mean temperatures) for Por and 24 June. tugal since 2000, the average maximum temperature During summer, high temperatures and severe was 24.40°C (+2.93°C above normal), which is the precipitation deficits in Portugal enhanced extensive highest value since 1931. With slightly below-average wildfires (> 1000 ha), with more than 60 fatalities; temperature anomalies of −0.4°C in December, the the fires were so large they were visible from space. year ended cold for the Iberian Peninsula. Due to ex-Hurricane Ophelia, strong winds prevented the extinguishing of fires in Portugal (ii) Precipitation and Spain (9–21 October). At least 41 people were Overall, the Iberian Peninsula was very dry in killed in wildfires across the region. Additionally, 2017 and was characterized by intensive drought. It was the second driest year since the beginning of the ashes from the fires were transported as far as the series in 1965 for Spain and third driest for Portugal UK, where yellow skies and a red sun were reported, and Switzerland, where ashes were detected at air since 1931. The winter season 2016/17 was drier than normal, monitoring sites in Payerne and on Jungfraujoch (3580 m a.s.l.). According to the Portuguese Institute with 69% of normal precipitation for Portugal. In for Nature Conservation and Forests, burned Spain, a dry January (23% of normal) was followed areas exceeded 440 0 by a wetter-than-normal February (136% of normal). 00 ha, a new record. Central Some regions, including the southern half of Galicia, regions of mainland Portugal were the most affected by very large fires (> 1000 ha) during several west Castilla-León, south Navarra, and extensive periods: 16–21 June, 16–18 July, 23–26 July, 9–19 areas of La Rioja, Central System, Pyrenees, Huesca, and Huelva provinces, as well as the eastern Canary August, 23–27 August, 5–9 September, and 12–15 Islands, had above-normal rainfall of 175%. October. During these periods, the associated For Spain, spring began with a wet March but the meteorological conditions were extremely favorable seasonal average showed a precipitation deficit of to fire propagation and adverse to fire combat; fire about 75% of normal for both Spain and Portugal. weather index values were higher than the 90th While summer in Portugal was its seventh driest percentile in the majority of the regions. | S228 AUGUST 2018

249 5) m and alkan s tates , taly , i editerranean b Croatia categorized its summer as extremely warm, This region includes Italy, Malta, Slovenia, Croatia, with above-average anomalies between +2.7° and +4.5°C. During heat wave events, Bulgaria measured Serbia, Montenegro, Bosnia and Herzegovina, Alba - extreme maximum temperatures of 42.5°C in San - nia, Republic of North Macedonia, Greece, Bulgaria, danski and 43.6°C in Ruse. Macedonia experienced and Turkey. Balkan States include North Macedonia and Bulgaria unless otherwise specified. unusually long-lasting periods of warm weather, with anomalies exceeding +5°C in mid-June. Turkey set a (i) Temperature new all-time high temperature record of 45.4°C in The Mediterranean and the Balkan states showed Antalya on 1 July. Anomalies were up to +2°C in eastern Turkey in average anomalies of mostly between +1° and +2°C autumn. Temperatures mostly f luctuated around in 2017. Based on station Zagreb-Grič, Croatia had its sixth warmest year in the series from 1862. With an normal, and no major extremes were reported. The - anomaly of +1.2°C, Bulgaria experienced its warmest year ended warmer than normal for the region (ex year on record since 1980. cept parts of Italy, Malta, southern Greece) due to Winter 2016/17 was dominated by an unusually prevailing southerly f lows over southeastern Europe cold January due to a low pressure system centered in December. over southeastern Europe leading to an inf low of cold air masses from Siberia. In Greece, monthly (ii) Precipitation minimum and maximum temperatures were 3° to Especially for Italy, 2017 was drier than normal. For most Balkan states, precipitation totals were 5°C below the 1971–2000 normal. Slovenia reported its coldest January of the last 30 years. In west Bul - near-normal but often irregularly distributed in both time and space. For example, while overall Serbia was garia, minimum daily temperatures were close to the near-normal, the station in Zrenjanin recorded its records of the last 50 years, with −26°C in Kyustendil driest year since its record began in 1925. Similarly, and −27°C in Pernik (−1.2°C below normal for the Croatia was slightly wetter than normal for the year whole winter). Serbia observed its fourth coldest January since its record began in 1951. Turkey re but the wider area of the town Split was extremely dry. - ported a negative January anomaly of −1.5°C below Winter 2016/17 was drier than normal, with pre - cipitation of around 60% of normal for the Balkan its 1981–2010 normal. states and most of Italy. Nevertheless, in January, due In spring, due to prevailing southerly f lows, tem - peratures in Italy, Greece, and the other countries to the cold Siberian air masses crossing the warmer of the Balkan Peninsula were above normal with Aegean Sea, high amounts of snowfall were observed in Greece and North Macedonia with adverse impacts anomalies around +1°C. Although anomalies showed - only a slightly warmer-than-normal spring, events on transportation. Turkey reported a winter precipita tion deficit of 19.5%. like the heat wave that occurred between 12 and 13 With the exception of Italy and southern Greece, May in the central and southern parts of Greece led to a new daily maximum temperature record of 40.6°C spring was near to slightly wetter than normal overall. In May, several extreme precipitation events at the station Argos. - were measured during low pressure situations over A prolonged anticyclonic situation over the north Greece and Bulgaria. Thunderstorms accompanied ern Adriatic Sea contributed to a very warm summer for Italy and the Balkan states, with anomalies around by cyclone “Victor” led to f looding and hail that destroyed crops. During 17–18 May, 230 mm of rain +3° to +5°C. Several heat waves contributed to these high anomalies during all three summer months. was measured at station Sitta in Greece, while 139 mm fell in twelve hours at station Semprona at the Some stations in Italy measured new all-time records - in early August, for example, 41°C in Pescara. Slove end of the month. Throughout the summer, several cut-off lows were nia reported its second warmest summer, surpassed - centered south of Greece and supplied Greece and only by 2003. Serbia also reported its second warm - est summer; July and August proved to be extraor western Turkey with well above-normal precipita - tion (up to 250% of the seasonal normal), including dinarily warm. Several stations in Serbia observed heavy rainfall and hailstorms, sometimes leading their third highest August temperature, although on to f looding; conversely, Italy and the Balkan states, average across the country, it was seventh highest. A record-breaking number of days with temperatures under the inf luence of high pressure, experienced above 38°C and new all-time records for the number a drier-than-normal season. The summer for the of tropical nights were set at several stations in Serbia. Emilia-Romagna region in northern Italy was its | S229 AUGUST 2018 STATE OF THE CLIMATE IN 2017

250 third hottest since 1961; warm temperatures, com - Turkey recorded a severe hailstorm on 27 July, bined with dryness, aggravated drought conditions. with hailstones up to 9 cm in diameter observed in At the end of August, the drought in Italy reached Istanbul. its maximum intensity. Serbia reported a dry to very - In Naples, Italy, a heavy thunderstorm on 5 Sep tember brought hail up to 11.5 cm in diameter and dry summer, and Malta had its driest July since 1951 - weights up to 350 g, injuring several people and ani (56% of normal precipitation). Autumn in the Balkan mals, as well as causing damage to vehicles, houses, states was wetter than normal, especially in Bulgaria trees, and crops. with precipitation up to 167% of normal. In October, - The slow-moving cyclone “Quasimodo”, ap Bulgaria’s average precipitation was 2.5–3 times the normal. Conversely, Italy and most of Turkey suf proaching Italy from the Ligurian Sea, reached the - city of Livorno on 9 September, with heavy precipi fered from precipitation deficits and drought. The - year ended drier than normal for southern Turkey tation causing f looding and damage, along with six and southern Italy, while most of the Balkan states fatalities. After passing over Toscana and the city of Pisa, the cyclone reached the Balkans. The Adriatic received above-normal precipitation. coast and the islands of the Adriatic Sea received more (iii) Notable events and impacts than 500 mm precipitation, causing f loods in Croatia In Bulgaria, for the first time in the past 60 years, that damaged houses and cars. A total of 135 million the coastal waters of the Black Sea were frozen—an Euros (around 160 million US dollars) in damages in occurrence observed only three times since the begin the aftermath of the f lood was estimated just for the - Croatian county Zadar. ning of the 20th century. On 21–22 April, a severe frost event in Slovenia Very heavy precipitation on 1 December led to extensive f looding and landslides in Greece. Between caused catastrophic damage to crops. 8 and 12 December, an exceptional meteorological In Greece, extensive and long-lasting snowfall event occurred in Italy with intense rain at some loca during several days in January caused severe traffic - tions (more than 300 mm in 48 hours). At Cabanne, problems, trapping hundreds of vehicles, disturbing Genoa province (Italy), an overall total of 507.0 mm public transport in Thessaloniki, and suspending −1 were was measured. Strong winds as high as 49.5 m s f lights. After serious power failures, the Aegean islands of Skopelos, Alonnisos, and Evia declared a measured at Loiano. state of emergency. e 6) e urope A severe hailstorm during 7–9 May caused astern This region includes the European part of Russia, heavy damage in the agricultural areas in northern Greece. At the beginning of June, Bulgaria was hit Belarus, Ukraine, Moldova, and Romania. by a series of severe thunderstorms accompanied by heavy rainfall and hail causing f loods and damage to (i) Temperature For most of European Russia and Belarus, 2017 crops. Further local storms with hail were reported in Slovenia and Italy with hourly precipitation totals was warmer than average, with anomalies of +1.45° and +1.7°C, respectively. Moldova and Romania reaching as high as 46.5 mm in Salsomaggiore (Italy). recorded slightly lower positive anomalies of +1.2°C At station Vojsko in Slovenia, on 6 June, a new 24-hr (normal 1961–1990) and +0.7°C (normal 1981–2010), daily precipitation record of 200 mm was set. Two intense hailstorms were registered on 7 and 14 June respectively. Ukraine reported its third hottest year since the beginning of observations in 1961 with an in North of Macedonia. - On 3 July, northwestern and north-central Bulgar anomaly of +1.8°C. ia reported severe convective storms accompanied by The winter season 2016/17 on average was warm for European Russia, Belarus, and Ukraine. Anoma - strong winds and extreme hail, with stones measuring up to 8 cm diameter in Mezdra and Levski. lies were below average only in January, when a cold During the first half of July, an intense heat wave spell in the Volga region dropped temperatures to as low as −40°C and several absolute minimum hit Croatia, drying out the plant cover, which led to - the outbreak of a wildfire on 17 July near Split. Ap temperatures were exceeded in the cities of Arkhan - proximately 4300 ha of forest, brush, olive groves, and gelsk, Kotlas, Naryan-Mar, Kirov, and Tver. Similarly, vineyards were burned. With a 40-km long fire front Romania and Moldova reported severe cold during at its maximum, it was one of the biggest wildfires in this time, with the latter measuring temperatures of Croatian history. 8°–10°C below normal. Winter in Moldova, overall, | S230 AUGUST 2018

251 was slightly colder than normal, with an anomaly of - and 121%, respectively. Nevertheless, it was the sec ond wettest year on record for European Russia. –0.6°C (1961–90 normal) for the season. During winter 2016/17, only Romania, Moldova, At the beginning of spring, under the inf luence of and isolated spots in Russia showed below-average a cyclone located over northwest Europe, new high precipitation of around 60% of normal; the rest of the temperature records in Moscow and St. Petersburg, area was near-normal or slightly above. as well as abnormal warmth at the Arctic coast, were Spring was rather wet for Moldova, where some reported. During the second half of March, tem - perature records in Smolensk, Tambov, Cheboksary, places during April observed values of 85–128 mm, - corresponding to 230%–350% of normal precipita and other Russian cities were measured and, overall, tion, for the first time in the period of record dat - the seasonal weather was climatologically ahead by ing back at least 80 years. Most areas of Russia also a month. A nation-wide heat wave hit Romania in reported a precipitation surplus (of up to 167% and March, with a monthly temperature +3.4°C above its 1961–2010 normal. At the station in Baisoara, the locally even more) while Belarus was near-normal. absolute monthly maximum temperature for March The exception to the wet spring was the central region of Ukraine where deficits of 60% of normal was exceeded. Moldova also reported an abnormally warm March. Then, in late April a wintry spell result - were measured. During summer, this deficit extended over all of ed in additional—this time minimum—temperature - Ukraine, with 60% of the agricultural area affected records in the area, as low as −15°C in Smolensk (Rus sia). The season ended with a cold May, which brought by drought. Belarus reported a dry June but, overall, - additional minimum temperature records. Never summer precipitation was near- or slightly above theless, overall, spring was warmer than normal for normal. Most regions of European Russia, other Moldova, Belarus, and the Ukraine. With a monthly than the south, which received 3%–25% of normal mean air temperature of 10.9°C, Moscow (Russia) monthly precipitation in August, had near-normal experienced its coldest May of the 21st century. precipitation or even a surplus (up to 167% in the northwest). After heavy rain events accompanied by Even with a late frost event on 4 June (temperatures hail and strong winds caused major damage in June down to −1°C) in the Ukraine, overall, summer for the Ukraine, Moldova, and Romania was warmer and July, Moldova experienced a precipitation deficit in August where locally severe drought was reported. than normal, with widespread anomalies up to +2°C With prevailing anticyclonic conditions in Sep - above normal. In contrast, the northern part of Euro - pean Russia was colder than average; a new daily low tember, autumn began with a large precipitation temperature record of 7.8°C was set in Moscow on 15 deficit (mostly below 60% of normal) for the south of European Russia as well as for eastern Ukraine. June, with observations dating to 1949. Autumn was warmer than normal for Russia, October was the wettest month of the year for the eastern countries (Moldova, Belarus, and Romania). Ukraine, Belarus, and Moldova, while temperatures in Romania were near-normal. During the second - At the end of the season, only eastern Ukraine, south half of September and November, anomalies of up ern parts of European Russia, and the Ural region to +2°C were observed, while the rest of the season suffered from precipitation deficits. December was mostly remained close to normal. The year ended wet throughout eastern Europe. with an exceptionally warm December—especially notable in European Russia where anomalies ex - (iii) Notable events and impacts On 20 April, the Kirov region (Russia) reported ceeded +6°C; anomalies for Belarus, Ukraine, and exceptionally heavy snowfalls and freezing rain, lead - Moldova, ranging between +3° and +5°C, were also quite high, and only the southern part of Romania ing to power failures due to damage to transmission lines for 44 settlements and to extensive damage to showed anomalies below +3°C. forests and agriculture. (ii) Precipitation In April, Moldova reported extreme weather Precipitation totals were near- to slightly above conditions, with rain, snow, and sleet depositing on normal in eastern Europe with the exception of wires and trees, as well as strong wind and frost with Ukraine, which reported precipitation 60%–70% of disastrous consequences for the country’s economy. normal in the south and central regions and near- Likewise, reports were made in the Ukraine of unusu - normal for the rest of the country. Russia and Belarus ally high numbers of frost days that damaged fruits, on average had slightly above-average totals of 115% vegetables, and other crops. | S231 AUGUST 2018 STATE OF THE CLIMATE IN 2017

252 A severe thunderstorm (“Falk”) hit Moscow During winter 2016/17, precipitation was unevenly (Russia) and the surrounding area on 30 May. For distributed, but, except for most parts of the South - Caucasus which received slightly above-normal pre the first time since the beginning of instrumental cipitation, totals were below normal, with extreme observations in Moscow for more than 100 years, −1 deficits in some regions. Northeast Israel reported less wind gusts of 30 m s were recorded, resulting in than 15% of its normal monthly average in February, structural damage to buildings. Also in Moscow, 11 which was the driest February since 1958 for this area. people were killed and 70 injured on 29 July during a −1 Spring continued to be dry, with totals between storm where wind gusts reached 29 m s . 20% and 70% of normal for Cyprus, Jordan, Lebanon, Heavy precipitation of 100–120 mm within a 24-hr and Israel. Georgia and some parts of Azerbaijan period in the region of Bucharest (Romania) caused f looding in July. and western Armenia received above-normal pre - cipitation. Summer was extremely dry, with no pre - iddle ast cipitation at all for widespread regions in Lebanon, 7) m e This region includes Israel, Cyprus, Jordan, Leba northern Syria, and Israel. - While autumn was also dry for the region around non, Syria, West Kazakhstan, Armenia, Georgia, and the Mediterranean, locally heavy precipitation events Azerbaijan. provided surpluses of up to 250% of the monthly normal (e.g., the Karmel Region in Israel), sometimes (i) Temperature resulting in f looding. Georgia received near-normal In the Middle East, temperatures in 2017 were precipitation amounts, with Armenia and Azerbaijan above normal by +1° to +2°C. Israel reported its fourth warmest year among the last 67 years. above normal (up to 167%). The year began on a cold note, however. Winter The year ended dry for the Middle East countries on the Mediterranean and near-normal for Georgia, 2016/17 temperatures on average were below normal. Armenia, and Azerbaijan. January and February showed negative anomalies down to −2°C for some regions in Georgia, Armenia, (iii) Notable events and impacts Israel, and Jordan. Israel reported its coolest February Three men died after being swept away by strong daily minimum temperatures since 1999. −1 easterly winds of 14–18 m s accompanied by gusts Spring had slightly above-normal temperatures, −1 of 22–25 m s in northern Israel on 12 April. One with anomalies around +1°C. In May, unusually high temperatures above 40°C were measured in Israel as day later, at the station Neot Smadar in the southern Negev, heavy rainfall of 27 mm, of which 10 mm fell a result of Sharav (heat wave) events, which brought warm sandy air from the Sinai Peninsula. within only 5 minutes, was measured, resulting in The Middle East experienced a hot summer, with f looding and the closing of two main routes to Eilat. On 18 May, severe sandstorms were advected to anomalies of up to 4°C above normal. Multiple heat waves in July were responsible for extreme tempera - southern Israel from the Sinai Peninsula, where they tures in Cyprus, Jordan, and Israel, which tied with were created by downdraft winds related to well- 2012 as the warmest July in Israel since the beginning developed clouds. These Haboob-type sandstorms of records, with average daily temperatures +2° to are uncommon in Israel. As a result, the Eilat Airport was closed for several hours. +2.5°C above normal. - On average, autumn also was warmer than nor On 16 October, a heavy rainfall event in Nahariyya mal, between +1° and +2°C. The year ended with well (northwest coast of Israel) brought more than 70 mm - of precipitation within two hours. During the morn above-normal temperatures in December, with some stations in Israel ranking as high as second warmest ing hours of 30 October more than 50 mm within one hour were measured in Haifa. Both events were since the beginning of measurements. followed by f looding and subsequent road closures. (ii) Precipitation g. Asia With regard to precipitation in the Middle East, 2017 was characterized by widespread deficits of Throughout this section the base periods used 20%–40% of normal and even less for some regions. vary by region. The current standard is the 1981–2010 Israel reported its third lowest annual precipitation average for both temperature and precipitation, but - - earlier base periods are still in use in several coun total; only 1959 and 1999 had less rainfall. Region ally, it was the driest year on record in the coastal tries. All seasons mentioned in this section refer plain of Israel. to those of the Northern Hemisphere, with winter | S232 AUGUST 2018

253 referring to December 2016–February 2017, unless In winter, enhanced convection appeared over the otherwise noted. Maritime Continent and the South China Sea (Fig. 7.41a); to its north, positive anomalies of 500-hPa verview T. Li, Z. Zhu, P. Zhang, T. C. Lee, Y. Mochizuki, 1) o — geopotential height and 850-hPa temperature (Fig. 7.42a) were observed over East Asia. In spring, anti - S.-E. Lee, L. Oyunjargal, and B. Timbal Annual mean surface air temperatures during cyclonic circulation anomalies straddled the equator over the western Pacific in the lower troposphere (Fig. (January–December) 2017 were above normal across most of Asia and much above normal (anomalies 7.41b). In summer, convective activity was suppressed to the east of the Philippines, and the western North >1.5°C) in Siberia (Fig. 7.38). Annual precipitation Pacific subtropical high was shifted westward (Fig. amounts were above normal from western China 7.41c). In autumn, anticyclonic (cyclonic) circulation to northeastern India, from the western part of the anomalies straddled the equatorial western Pacific Indochina Peninsula to the central part of the Malay Peninsula, in the Maritime Continent, and from the (Indian) Ocean in the lower troposphere (Fig. 7.41d). western part of eastern Siberia to western Siberia, and In terms of the summer monsoon in 2017, East Asian summer monsoon rainfall was weaker than they were below normal in the eastern part of eastern Siberia, from the Korean Peninsula across northeast normal, while Indian summer monsoon rainfall was - near normal. Intraseasonal variability of convective ern China to Mongolia, and in central Asia (Fig. 7.39). activity was clearly evident in the monsoon regions. Though annual mean temperature anomalies were virtually all positive, they evolved quite differently O. N. Bulygina, N. N. Korshunova, M. Yu. Bardin, season by season (Fig. 7.40). In winter (Fig. 7.40a), — ussia r 2) and S. G. Davletshin negative temperature anomalies appeared over west - ern Siberia, associated with a large-scale tropospheric This review for Russia and its individual regions, along with estimates of abnormal climate features, barotropic negative geopotential height anomaly - (Figs. 7.41a and 7.42a). In spring, cold anomalies ap are obtained from hydrometeorological observations peared over southwestern China and the Indochina taken at the Roshydromet Observation Network. Peninsula (Fig. 7.40c), corresponding to positive rain - Unless otherwise noted, anomalies are relative to a 1961–90 period, and national rankings ref lect an 82- fall anomalies in the region (Fig. 7.40d). In summer, year (1936–2017) period of record. negative temperature anomalies dominated over the northern part of central to eastern Siberia. Autumn was marked by negative temperature anomalies (i) Temperature - The year 2017 in Russia was warm: the mean annu from central Siberia to Japan. Seasonal precipita - tion amounts were persistently above normal in the al national air temperature was 2.02°С above normal western part of the Tibetan Plateau from winter to (Fig. 7.43). This is the fourth highest such temperature on record. Positive annual mean air temperature summer and in Southeast Asia throughout the year, anomalies were observed in all regions of Russia. The but they were below normal in northeastern China largest anomalies occurred over Asian Russia (east and Korean Peninsula from spring to autumn. - of the Urals, approximately 60°E); the annual tem . 7.39. Annual precipitation (% of normal; 1981–2010 ig F F ig . 7.38. Annual mean surface temperature anomalies base period) over Asia in 2017. (Source: Japan Meteo - (°C; 1981–2010 base period) over Asia in 2017. (Source: rological Agency.) Japan Meteorological Agency.) | S233 AUGUST 2018 STATE OF THE CLIMATE IN 2017

254 decades. March was exceptionally warm (see Fig. 7.44), ranked warmest on record, both for Asian Rus - sia (+6.79°С anomaly) and the whole of Rus - sia (+6.03°С anomaly). Monthly temperature extremes exceeding the 95th percentile were observed at virtually all stations in Asian Russia north of 60°N. - North Atlantic cy clones brought warm, wet air in the north - ern regions as far east as Yakutia. Monthly March anomalies in Yamal-Nenets and Taymyrsky Dolgano- Nenetsky Autonomous Districts exceeded 13°С. At many stations monthly tempera - tures exceeded previ - ous records. In Tiksi, Nar’ jan-Mar, above- - normal daily tempera tures were observed almost throughout the entire month, with even minimum daily temperatures above the normal daily . 7.40. Seasonal mean surface temperature anomalies (°C, left column) and F ig maximum (Fig. 7.44). - seasonal precipitation (% of normal, right column) over Asia in 2017 for (a), (b) win On 21 March a daily ter; (c), (d) spring; (e), (f) summer; and (g), (h) autumn. All relative to 1981–2010. minimum temperature (Source: Japan Meteorological Agency) above the climatologi- perature averaged over this region was 2.27°С above cal maximum daily temperature was recorded in normal, the highest such value on record. Tiksi. European Russia was cooler than its Asian counterpart (+4.14°С anomaly; third highest), but For Russia as a whole, winter was moderately warm, with the mean winter temperature 2.05°С extremes above the 95th percentile were observed at most stations in the central and northeastern parts above normal (14th warmest on record). of this region. Spring 2017 was warm, with an average seasonal Summer 2017 was warm in Asian Russia, with a mean air temperature anomaly of +2.82°С; it was seasonal air temperature anomaly of +1.37°С, the the fourth warmest spring since 1936. However, the - sixth highest since 1936. In European Russia sum temperature averaged over Asian Russia was record- mer was much colder (only +0.46°С, rank 39th, breaking at 3.69°С above normal, while European which is close to the median value in the series) but Russia was only slightly warmer than normal (+0.65°С not unusual, even against the background of the last anomaly; the 27th warmest value of the record) with two decades. June was the coldest summer month, an unusually cold May, especially compared to recent | S234 AUGUST 2018

255 especially in European Russia, where it ranked as the eighth coldest June on record, with an anomaly of −1.44°С. In contrast, August was exceptionally warm, with the second highest anomaly on record of +1.61°С over Asian Rus - sia and the fourth high - est (+1.81°С) for Russia as a whole. Autumn was temper - ate across Russia, with a seasonal anomaly of +1.12°С. December 2017 was very warm in European Russia, with an anomaly of +4.81°С (third warmest). Com - bined with warmth in Asian Russia (+2.35°С), 6 ig . 7.41. Seasonal mean anomalies of 850-hPa stream function (contour, 1 × 10 F Russia as a whole was − 2 − 1 2 m ) using data from the JRA-55 reanalysis and OLR (shading, W m s ) using 3.05°С warmer than data originally provided by NOAA in 2017 for (a) winter, (b) spring, (c) summer, average, the eighth and (d) autumn. Base period: 1981–2010. (Source: Japan Meteorological Agency.) warmest December on record. The location of the Siberian anticy - clone remained stable throughout the month. This enabled advection of warm subtropical air along its western periph - ery and the formation of large positive tempera - ture anomalies in Euro - pean Russia and west - ern Siberia. The largest anomalies (> +10°С) were observed in the Yamal-Nenets Autono - mous District. A large anomaly observed in the northern Yakutia and Chukotka persisted from November through the end of the year. At many stations monthly tem- peratures in December ig F . 7.42. Seasonal mean anomalies of 500-hPa geopotential height (contour, gpm) and 850-hPa temperature (shading, °C) in 2017 for (a) winter, (b) spring, exceeded the 95th per - (c) summer, and (d) autumn, using data from the JRA-55 reanalysis. Base period: centile. On 24, 25, and 27 1981–2010. (Source: Japan Meteorological Agency.) December temperatures | S235 AUGUST 2018 STATE OF THE CLIMATE IN 2017

256 second), while August was drier (87% of normal). Autumn precipitation for the whole of Russia was moderate (108%), with a dry October in the Asian part (87%). Precipitation in De - cember 2017 was much above normal with 124% of normal precipitation, - the third wettest on re cord. This was especially so in European Russia - (128%, tied for second wet test). (iii) Notable events and impacts On 8–9 March, an ex - tremely severe snowstorm raged in the eastern Chu - kotka Autonomous Area, with wind gusts attaining −1 and visibility as 36 m s low as 0.5 m. Schools were closed; traffic and local f lights stopped. . 7.43. Mean annual and seasonal temperature anomalies (°C; base period ig F On 28 April, strong 1961–90) averaged over Russia, 1936–2017. The smoothed annual mean time −1 winds (25 m s ) in the series (11-point binomial filter) is shown as a red bold line. Linear trend ß (°C - Eravninsk region of Bury 1 − decade ) is calculated for the period 1976–2017. atia increased the number above the previous daily maximum were recorded at of wild fires. These burned 21 buildings (including Mys Uelen, and the December mean temperature there 17 homes). Property damage was estimated to exceed exceeded −5°С for the first time on record. 7.5 million rubles ($130 000 U.S. dollars). On 3 May, severe forest fires were recorded in the Irkutsk Region and the Krasnoyarsk Territory (ii) Precipitation In 2017, Russia as a whole received above-normal resulting in the introduction of a federal emergency precipitation, 111% of its 1961–90 normal. This is the regime. In the Krasnoyarsk Territory alone, the fire destroyed about 130 houses, leaving more than 500 second wettest year on record, after 2013, which had 112% of normal precipitation (Fig. 7.45). European people homeless. In the Strelka settlement, where the Russia was relatively wetter (115%, second wettest) Yenisei and Angara rivers merge, streets were burnt away. Two citizens of Kansk were killed. than Asian Russia (109%, fifth wettest). On 29 May, a squall wind event in Moscow, with Winter precipitation was 110% of normal, which −1 gusts as strong as 29 m s , caused the largest number of ranks 15th wettest. Spring in Russia had 119% of normal precipitation, which ranks fourth wettest. victims on record for such an event: officially, 11 people The wettest months of 2017 were April in European were killed. According to the TASS Agency, referring to a source in emergency services, 105 people were Russia (137% of normal) and May in the Asian part (125% of normal). admitted to hospitals. Within just a few minutes, the winds felled thousands of trees, damaged many cars Although summer precipitation in Russia as a and the roofs of 30 houses, and stopped the operation whole was near normal (107%), European Russia of above-ground subway and commuter trains. received much-above-normal precipitation in June and July (135% and 129% respectively, both ranked | S236 AUGUST 2018

257 On 30 June–1 July, a heavy thundershower (65 mm), accompanied by hail as large as 3–8 cm, and strong winds occurred in Moscow. Railroad beds in the Moscow Central Cir - cle and motor roads were inundated, 1200 trees fell, and the roofs of nearly 100 houses and nearly 100 cars were damaged. Two people were killed. Heavy rains (50–250 mm) in the southwestern Maritime Territory on 6–7 August caused flooding in Ussuriisk on 7–8 August. Parked cars were submerged, ground f loors of houses were inundated, and bus services and commuter trains were canceled. On 20–21 August, heavy rain (113 mm) in Krasnoyarsk lasted for more than 24 hours, f looding 40 homes and 43 road sections. In addition, 67 power . 7.44. Air temperature anomalies (°C, shaded) in Mar 2017. Insets show the F ig transforming substations series of mean monthly (from the beginning of the record to 2017) and mean were inundated and cut off daily air temperatures (°C) in Mar 2017 at meteorological stations Igarka, Tiksi, from personnel. Nar’jan-Mar. Plots of daily temperature show observed daily mean (black curve), - minimum (blue) and maximum (red) temperatures along with their climato On 1 September, in the logical normals and absolute maximum temperature; the area between daily upper reaches of the Adyl- mean values above normal and the normal daily mean curve is shaded pink, and Su River in the North where values are above normal daily maximum the shading is red. Caucasus a dam failure occurred, resulting in a 3 water discharge of 400 0 00 m that transformed into a mud f low downstream. A 325-mm gas pipeline was damaged, and a federal highway was blocked. Three people were killed. s outheast a sia ast e 3) and — P. Zhang, T. C. Lee, Y. Mochizuki, S.-E. Lee, L. Oyunjargal, and B. Timbal Countries considered in this section include: China, Hong Kong (China), Japan, Korea, Mongolia, and Singapore. Unless otherwise noted, anomalies . 7.45. Annual precipitation (% of normal; 1961–90 F ig refer to a normal period of 1981–2010. base period) averaged over Russia for 1936–2017. The smoothed time series (11-point binomial filter) is shown as a bold line. | S237 AUGUST 2018 STATE OF THE CLIMATE IN 2017

258 (i) Temperature above normal in the Yellow River (111%), the Yangtze - Annual mean temperatures across East and South River (105%), the Pearl River (105%), and the Huaihe east Asia are visible in Fig. 7.38. The annual mean air River (104%) but below normal in the Liaohe (84%), Songhua River (95%), and Haihe River basins (98%). temperature in 2017 for China was 0.84°C above nor - The annual total rainfall in Hong Kong was 2572.1 mal (9.55°C)—the third warmest year since records began in 1951, behind 2015 and 2007. All seasons mm (107% of normal). In Japan, annual precipitation amounts were were warm throughout the year, especially winter, above normal on the Sea of Japan side of northern with the highest anomaly in the historical record of 2.0°C above normal. Hong Kong had an annual mean and eastern Japan, and on the Pacific side of western temperature of 23.9°C, 0.6°C above normal and the Japan, and they were below normal in Japan’s Oki - nawa/Amami region. Annual total precipitation over third highest since records began in 1884. South Korea was 967.7 mm, 74% of normal (1307.7 Annual mean temperatures were near normal in mm), and the fifth lowest total since 1973. Annual many regions of Japan and were significantly above precipitation over Mongolia was 173.4 mm, 86.3% of normal in the Okinawa/Amami region. The 2017 annual average temperature over South Korea was normal. Although annual total precipitation was near normal, the majority of the growing season, especially 13.1°C, which is 0.6°C above normal, making 2017 the seventh warmest year since national records May, June, and July, were dry (57.1%–66.1%), causing began in 1973. The annual mean temperature over drought conditions over 75% of the whole territory. Relative to normal, March was the wettest month Mongolia for 2017 was 1.9°C, 1.4°C above normal, (225%) while July was the driest (57.1%). January pre which is the second highest annual value since 1961. - cipitation was near normal, with snow cover extent The highest anomaly for Mongolia was April, with a covering more than 70% of the total area. At the end mean temperature anomaly of 2.9°C above normal, of the year, snow covered almost 50% of the area and representing the third warmest April since 1961. The lowest anomaly for Mongolia occurred in October snow depth was 6–45 cm which caused difficulties when the mean temperature anomaly was 1.1°C, for livestock pasturing. −0.3°C below normal. In Singapore, there was a mixture of above- and - In South Korea, May temperatures have signifi below-normal rainfall for the individual months in - cantly increased since the start of the record in 1973. 2017. Overall, the annual total rainfall was approxi The temperature averaged over South Korea in May mately 94% of the normal of 2165.9 mm. In 2017, the overall activity of the East Asian sum - 2017 was 18.7°C, 1.5°C above normal. This marks the mer monsoon was near normal with some area to area fourth consecutive year of a new record high May differences in the region. Intraseasonal variability of temperature. the East Asian summer monsoon was clearly seen After two successive record warm years in 2015 and 2016, the mean annual temperature of Singapore during the season. For example, convective activity (27.7°C) returned closer to the long-term climatologi over and around the Philippines was enhanced from - late June to mid-July and was suppressed from late cal average. This was 0.2°C warmer than normal and tied as the 12th warmest year on record since 1929; July to mid-August. however, it is Singapore’s warmest year on record not inf luenced by an El Niño event. (iii) Notable events and impacts Eight typhoons landed in China, the same number (ii) Precipitation as 2016, and near the average of 7.2. However, the impact period was longer than normal, with the first Figure 7.39 shows 2017 annual precipitation as a percentage of normal over East Asia. The annual typhoon, Merbok, landing in Shenzhen, Guangdong, on 12 June, 13 days earlier than normal, and the last mean precipitation in China was 641.3 mm, 101.8% of normal. Seasonal precipitation was 93% of normal typhoon, Khanun, landing in Zhanjiang, Guangdong, in winter and 108% of normal in summer. Spring on 16 October, 10 days later than normal. Typhoon impacts varied strongly by time and region. For in and autumn were near normal. Regionally, annual - stance, Typhoons Nesat and Haitang landed succes total precipitation was above normal in northwest - China (115% of normal), South China (105%), the sively on the coast of Fuqing city in Fujian Province middle and lower reaches of the Yangtze River (104%), during 30–31 July, and four typhoons hit the Grand north China (104%), and near normal in southwest Bay Area of Guangdong–Hong Kong–Macao in June, China, but below normal in northeast China (90%). July, August, and October. As Tropical Cyclone Hato The annual total precipitation over river basins was headed toward Hong Kong, the subsidence effect | S238 AUGUST 2018

259 ahead of its circulation brought oppressive heat to In northern Japan, on 5–6 July, record-breaking heavy rain associated with the active Baiu front fell the territory on 22 August as the temperature at the −1 Hong Kong Observatory soared to an all-time high of in Kyushu region, with 129.5 mm h and 545.5 mm −1 observed at Asakura in Fukuoka prefecture. 36.6°C. Stormy weather with hurricane-force winds (24-h) battered the city during the passage of Hato on the The heavy rain caused serious damage, including - following morning. With Hato’s approach coincid landslides and river overf lows. In Okinawa/Amami, - ing with the astronomical high tide, its storm surge monthly mean temperatures were record high in Au resulted in serious sea water f looding and damage in gust (+1.4°C above normal) and record-tying (since 1946) high in September (+1.3°C above normal) due to many low-lying areas in Hong Kong. - a stronger-than-normal subtropical high over south In 2017, meteorological disasters caused by rain storms and f loods in China were prominent and of Japan. In western Japan, the monthly precipitation total was record high, at 333% of normal for October. brought major losses, especially in southern China. - In Mongolia, a total of 76 extreme weather events Rainstorms occurred often and frequently in suc were observed, including episodes of heavy snow and cession. Eleven days of persistent heavy rainfall oc - f lash f looding. Together, these events caused about curred over southern China from 22 June to 2 July, $1.9 million (U.S. dollars) in economic loss. associated with a rain belt across the provinces of Hunan, Jiangxi, Guizhou, and Guangxi, where local A. K. Srivastava, J. V. Revadekar, and 4) s outh accumulations exceeded 500 mm. During summer, a sia — the high temperature events hit China earlier in M. Rajeevan northern areas but were more intense in southern Countries in this section include: Bangladesh, India, Pakistan, and Sri Lanka. Climate anomalies areas, which resulted in a record number of days are relative to the 1981–2010 normal. with high temperatures (daily maximum temperature ≥35°C) since the beginning of the record in 1961. (i) Temperature On 21 July, the maximum temperature at Xujiahui, In general, South Asia witnessed significantly in central Shanghai, was 40.9°C, setting a record for above-normal temperatures in 2017. The annual mean its 145-year period of observation (since 1873). In land surface air temperature averaged over India was the west during mid-July, 53 high temperatures were recorded, which tied or broke records in counties (or 0.50°C above the 1981–2010 average, ranking 2017 cities) in Xinjiang, Gansu, Inner Mongolia, Shaanxi as the fourth warmest year on record since nation- (44.7°C in Xunyang), Ningxia, and Shanxi. wide records commenced in 1901 (Fig. 7.46). India’s - South Korea experienced above-normal tem seasonal mean temperatures were above normal peratures and slightly below-normal rainfall during for all four seasons. The country-averaged seasonal mean temperatures during the post monsoon season summer. The summer mean temperature over South Korea was 24.5°C, which was +0.9°C above normal. (October–December, with an anomaly of +0.67°C, the - third highest since 1901) and the winter season (Janu In particular, extreme temperatures were observed from late-June through late-July. During this period, South Korea was strongly inf luenced by the western North Pacific subtropical high that extended more to the northwest compared to its normal position brought hot, moist air by the southwesterlies along its f lank. The summer rainfall (609.7 mm) over South Korea was 84% of normal (723.2 mm). The ratios of monthly rainfall amount to the normal value in June, July, and August were 38%, 103%, and 88%, respectively. The 2017 Changma (early summer rainy period) started on 24 June and ended on 29 July. The Changma rainfall total was below normal (291.7 mm; normal: 356.1mm). The 2017 Changma was notable for the following: 1) onset and retreat were later than normal; 2) heavy rainfall events were concentrated on . 7.46. Annual mean temperature anomalies (°C; F ig the central part of the Korean Peninsula; and 3) large 1981–2010 base period) averaged over India for 1901– spatial differences of rainfall between the southern 2017. The smoothed time series (9-point binomial filter) is shown as a continuous blue line. and central regions of South Korea were observed. | S239 AUGUST 2018 STATE OF THE CLIMATE IN 2017

260 ary–February, anomaly +0.61°C, fourth highest ever - received normal rainfall (100% of LTA). The homo since 1901) mainly accounted for the above-normal geneous regions of central India and east-northeast India received 94% and 96% of LTA of the seasonal temperature for the year. rainfall, respectively. Northwest India received below- normal rainfall (90% of monsoon season LTA). On (ii) Precipitation The summer monsoon season (June–September) monthly scales, rainfall for the country as a whole was normal during June (104% of its LTA value) and July typically contributes about 75% of annual precipita - tion over South Asia. The summer monsoon set in (102%). It was below normal during August (87%) and September (88%). During the monsoon season, out over Kerala (southern parts of peninsular India) on 30 May, 2 days prior to its climatological normal date of 36 meteorological subdivisions, five subdivisions - (1 June) and covered the entire country on 19 July (4 (West Rajasthan; Saurashtra & Kutch; Nagaland, Ma nipur, Mizoram & Tripura; Rayalaseema; and Tamil days later than its normal date, 15 July). For India, the long-term average (LTA) value of the Nadu & Pondicherry) received excess rainfall, 25 - summer monsoon rainfall, calculated using all data received normal rainfall, and the remaining six subdi visions (four subdivisions from the northwest region from 1951 to 2000, is 890 mm. The standard deviation and two subdivisions from central India) received of Indian summer monsoon rainfall (ISMR) is around deficient rainfall. Figure 7.48 shows the standardized 10% of the LTA value. However, over smaller regions rainfall anomaly over the core monsoon region on a natural variability of the monsoon is large (standard daily scale during the season. There was significant deviation around 19%). In view of the above, an ISMR exceeding 110% of the LTA in a year is termed as intraseasonal rainfall variability with marked active and break spells. excess rainfall, while an ISMR that is less than 90% of the LTA in a year is termed as deficient rainfall. During the winter season (January–February), season, rainfall over the country was normal (95% During 2017, the ISMR averaged over the country of LTA); it was normal (98% of LTA) during the as a whole was 95% of its LTA and was characterized by significant spatial and temporal variability (Fig. pre-monsoon season (March–May), while it was - 7.47). The homogeneous region of South Peninsula below normal during the post-monsoon season (Oc tober–December, 89% of - LTA). The northeast mon soon (NEM) normally sets in over southern penin - - sular India during Octo ber and over Sri Lanka in late November. The NEM contributes 30% to 50% of the annual rainfall over southern peninsular India and Sri Lanka as a whole. The 2017 NEM set in over southern peninsular India on 27 October, and the . 7.47. Spatial distribution of monsoon seasonal (Jun–Sep) rainfall over India ig F seasonal rainfall over south in 2017. (a) Actual, (b) normal, and (c) anomalies are in mm. peninsular India was below normal (86% of LTA value). Moderate La Niña condi - tions prevailed during the season, which could be one of the reasons for the below- normal performance of the NEM. Pakistan, at the western edge of the pluvial region of the south Asian mon - ig F . 7.48. Daily standardized rainfall time series averaged over the core mon - soon zone of India (1 Jun–30 Sep 2017). soon, receives 60%–70% | S240 AUGUST 2018

261 of its annual rainfall during the summer monsoon the system did not cause significant weather over the mainland of India, it caused light to moderate rain season (July–September). The summer monsoon over the Andaman and Nicobar islands during its sets in over northeastern parts of Pakistan around 1 formative stage. July with a standard deviation of five days. In 2017, The second severe storm, Mora, formed over the summer monsoon rainfall over Pakistan was below Bay of Bengal during 28–31 May. It made landfall normal (77.5% of its LTA value). Pakistan observed over the Bangladesh coast on 30 May and dissipated below-normal rains during all three months from over the northeastern parts of the country on 31 July to September (79.6%, 74.3%, and 82.7% of its May. This system caused moderate to heavy rain over LTA values). Rainfall was below normal in all the many parts of the northeastern region of India after provinces of Pakistan. Southern parts of Pakistan making landfall. (especially the southwest) received largely deficient The last severe storm, Ockhi, (29 November–5 rainfall. Other areas, including central Pakistan, December) formed over south Bay of Bengal and received normal rainfall during the monsoon season. moved to the Arabian Sea. The storm, while moving Bangladesh received normal rainfall during the 2017 summer monsoon season. across southern parts of India, caused severe damages in Kerala, Tamil Nadu, and Lakshadweep. The storm Sri Lanka received normal rainfall during its sum - also claimed the lives of many fishermen (18 from mer monsoon season (May–September). Northeast monsoon rainfall activity over the island nation, Tamil Nadu and 74 from Kerala). Ockhi ultimately recurved and moved towards the south Gujarat coast. during October–December 2017, was also normal. Heavy rain and f lood-related incidents during the monsoon season claimed around 800 lives from dif - (iii) Notable events and impacts During 2017, three cyclonic storms (one each in - ferent parts of India (see Table 7.2 for 24-hr rainfall re - April, May, and November) formed over the north In cords over India). Around 150 people reportedly died dian Ocean. The first storm, Maarutha, formed over in the state of Assam from 13 June to 11 September the east central Bay of Bengal on 15 April. However, in two spells of f loods. More than a hundred people were reported dead in Uttar Pradesh due to heavy it moved northeastward away from the Indian region rain and f loods of the Ghaghara, Gomati, and Rapti and crossed the Myanmar coast on 16 April. Though le AB t 7.2. Record rainfall (24hr) during the 2017 monsoon season in India. Rainfall Previous Date of Year of During Past 24 Station Date record S. No. record record (mm) Hrs. (mm) Jun 2017 Karnal 140.4 28 80 30 1994 1 Baderwah 30 2 55.8 2 2004 56 147.9 30 90.2 30 2000 Katra 3 Jul 2017 1 Ranchi AP 205.8 26 178.8 23 1958 2 Bhagalpur 2 154.6 27 2009 173. 6 Raisen 208.9 188.4 22 1973 3 28 Jagdalpur 19 180.8 7 1934 4 191.4 Aug 2017 Jalpaiguri 295.2 12 1 10 1987 264.2 2 151.0 20 149.8 22 2010 Osmanabad Adirampattinam 128.4 9 97.9 29 2007 3 4 Thanjavur 110 10 105 30 2016 5 Thanjavur 30 99.0 7 1982 105 S e p 2017 1 373.6 19 308.6 23 1949 Harnai Pamban 113 . 5 9 108.5 27 1901 2 3 Tuticorin 63 1 58.3 6 1979 4 81.6 3 65.3 29 1951 Uthagamandalam 129.3 11 1940 27 5 Mysore 137 | S241 AUGUST 2018 STATE OF THE CLIMATE IN 2017

262 Rivers during 4–10 September. About 120 deaths were experienced regional patterns of above- and below- reported from the western industrial state Gujarat average temperatures. There is a sharp contrast during the month of July and 107 from Bihar during between the pattern of temperature anomalies in 13–23 August. On 13 August, 46 deaths were reported winter and summer. Northwestern Iran experienced a colder-than-average winter season but a warmer- due to massive landslides at Kotrupi on the Mandi- Pathankot National Highway near Jogindernagar in than-average summer season. Himachal Pradesh (India). Similarly, 15 people died in Papum Pare, Arunachal Pradesh on 11 July due (ii) Precipitation to a landslide. In 2017, Iran was drier than normal for the year as - Heat wave conditions prevailed mainly over pen a whole, receiving 205 mm precipitation, with totals insular parts of India during the second fortnight of of 117.4, 53.1, 6.2, and 28.4 mm for winter, spring, May, which claimed the lives of about 100 people in summer, and autumn, respectively. The country the state of Telangana. However, the loss of lives in experienced drier-than-normal conditions in spring, 2017 was much less than in the previous years due summer, and autumn. The only relatively wet season was winter; however, several provinces, especially to timely heat wave warnings and heat wave action plans initiated by government. In April 2017, Larkana, in southwestern, northwestern, and parts of central Iran, observed precipitation deficits of up to 50% of a city in the southern province of Sindh of Pakistan, experienced a record maximum temperature of normal during winter. The winter seasonal rainfall 51.0°C on 20 April. On 28 May, Turbat, in western largely occurred during the second half of the season. MODIS snow data indicate that snow coverage over Pakistan, recorded a temperature of 53.5°C, tying the all-time highest temperature recorded in Moen the country was greatest in February and declined in March. Snow coverage was observed in December Jo Daro, Pakistan, on 26 May 2010. In August, Bangladesh suffered one of its worst 2017, the first month of winter 2017/18. - f loods in the past four decades, which affected ap According to spatial analysis of the standardized proximately one-third of the country, primarily in the precipitation index (SPI), areas with mild to severe drought conditions were encountered especially in northern, northeastern, and central regions. Rangpur district in the northeast experienced a month’s worth the northwest, west, and southwest in winter, the northwest, east, and northeast in the spring, and of rain—360 mm—in just two days (11–12 August). the northwest and northern parts in the summer. Around 140 deaths from the f loods were reported, - over fifty thousand people were displaced, and ap The SPI is a tool that was developed primarily for defining and monitoring drought. Mathematically, proximately six million were affected. it is based on the cumulative probability of a given - Fifteen districts of Sri Lanka were affected by se vere f loods during the last ten days of May. Parts of rainfall event occurring at a station. According to this indicator, most regions of the country were in mild Sri Lanka received 300–500 mm of heavy monsoon rain in a 24-hour period around 25 May, resulting in to very severe drought during autumn. The central widespread f looding. The highest recorded rainfall and eastern areas were in extreme drought conditions during autumn. was 533 mm in Kukuleganga. Galle, a coastal city, received 223 mm and Ratnapura experienced 453 (iii) Notable events and impacts mm of rainfall during 27–30 May, leading to severe A maximum temperature of 53.7°С was observed inland f looding. Around 150 people were killed and in Ahvaz in summer. In spring, an extreme warm around 450 000 were affected. temperature of 51.7°С was recorded in Sistan and outhwest 5) s Balochestan. While the entire country experienced M. Khoshkam and A. Kazemi — sia a This subsection covers only Iran. Turkey is in - dry conditions during 2017, an extreme rainfall event, corporated in the Europe subsection, 7f. Climate with 264 mm in 24 hours, was observed in summer at Station Lahijan. Table 7.3 lists measured extreme anomalies are relative to a 1996–2015 base period. events in each season over Iran. The frequency and duration of dust storms in 2017 became higher in (i) Temperature In general, the year was warmer than average some parts of the country, especially in the southwest for Iran. The mean annual air temperature was and southeast during winter and spring. 0.5°–1.5°С above the 20-year average. Spring and autumn temperatures were above average for the entire country, while the winter and summer seasons | S242 AUGUST 2018

263 t AB 7.3. Measured values of 2017 record temperatures and 24-hr precipitation in Iran. Location le includes station (province). Minimum Maximum Maximum 24-hr Season precipitation (mm) temperature (°C) temperature (°C) 37 Rask 164.6 Jem −30 Hezar Kanian (Kordestan) Winter (Sistan and Balochestan) (Bushehr) 51.7 Rask 99.7 Ghir Kazeroun 8.6 Zarineh Obatoo (Kordestan) − Spring (Sistan and Balochestan) (Fars) 0.8 Ardebil 264 Lahijan 53.7 Ahvaz Summer (Ardebil) (Gilan) (Khozestan) 200.1 Ramsar 45.2 Abadan (Khozestan) − Autumn 20.2 Khayerabad (Zanjan) (Gilan) S I D E B A R 7. 3: ABNORMAL WEST CHINA AUTUMN RAINFALL IN 2017 AND PERSISTENCE OF THE PACIFIC–JAPAN PATTERN IN AUGUST 2017 — Z. ZHU, T. LI, AND H. TOGAWA Climatologically, a precipitation peak over West China (27°–35°N, 105°–114°E) during September and October is re - ferred to as West China Autumn Rainfall (WCAR). It is the final stage of the rainy season in mainland China. WCAR can have severe impacts on agricultural produc - tion—including the harvesting and sowing of winter crops—and reservoir levels. Due to the fragile ecological environment of West China, above-normal WCAR often results in landslides and debris flows, which threaten lives and economic development in the region. WCAR in 2017 was 170% of normal (Fig. SB7.4a), the second highest total since records began in 1979 (Fig. SB7.4b). The event greatly affected 6 million people over seven Chinese provinces. Over 480 00 0 hectares of crops were damaged, and the total economic loss was over $20 billion (U.S. dollars). 1 − F ig . SB7.4. (a) Precipitation (shading; mm day ) and 850-hPa wind - In the lower troposphere during Sep 1 − (vector; m s ) anomalies in Sep and Oct 2017. Only winds with speeds tember and October 2017, an anticyclonic 1 − >1 m s are shown. (b) Year-to-year time series of the WCAR (red bars; anomaly at 850 hPa covered a large domain 1 − mm day ). 2017 is blue bar. Gray dashed line is the 1981–2010 climatologi - − 1 of southern China, with southwesterly cal mean. (c) Wind (vectors; m s ), geopotential height (contours; gpm) - anomalies at its western flank. Another an at 850-hPa, and SST (°C) fields regressed onto the WCAR index. (d) as − 2 in (c) but for 200-hPa wind, geopotential height, and OLR (W m ) fields. ticyclonic anomaly (high pressure system) Only winds significant at the 95% confidence level are shown, and the SST existed to the north of the anticyclonic and OLR significant at 95% confidence level are marked by red dots. “A anomaly. These two south–north anticy - (C)” denotes the anticyclonic (cyclonic) anomaly. Red box indicates the clonic anomalies led to a horizontal trough key region of WCAR, dashed purple line marks the horizontal trough. over West China. The southwesterly into West China while the horizontal trough kept the moisture anomalies in the southern portions of the trough transported stationary in the region, resulting in enhanced WCAR. moisture from the northern Indian Ocean and South China Sea | S243 AUGUST 2018 STATE OF THE CLIMATE IN 2017

264 CONT ’ ABNORMAL WEST CHINA AUTUMN RAINFALL IN 2017 S I D E B A R 7. 3: AND PERSISTENCE OF THE PACIFIC–JAPAN PATTERN IN — Z. ZHU, T. LI, AND H. TOGAWA AUGUST 2017 The unusualness of the rainy season was the result of the combination of various atmospheric patterns. Typically, the enhanced WCAR is associated with positive tropical Indian Ocean SST anomalies (Fig. SB7.4c). Positive tropical Indian Ocean SST anomalies could induce a Kelvin wave response in terms of easterly anomalies in the lower troposphere. The Kelvin wave easterlies generated anticyclonic shear and resulted in anticyclonic anomalies over the western North Pacific, lead - ing to anomalous moisture transport as seen in the 2017 event. Divergence at 200 hPa (Fig. SB7.4d was predominant over West China and appeared to be part of the circumglobal wave train over midlatitudes. Divergence aloft and convergence (associ - ated with the horizontal trough) at lower levels over the region provided a favorable dynamical condition to the enhanced rainfall in the region. - From early to mid-August, convective activity was particu larly inactive over and around the Philippines. During the same period, the North Pacific subtropical high (NPSH) did not extend to mainland Japan as usual but shifted southward from its normal position, corresponding to the Pacific–Japan (PJ) pattern (Nitta 1987; Kosaka and Nakamura 2010; Fig. SB7.5a), - with suppressed convective activity over and around the Philip pines. Furthermore, the Tibetan high in the upper troposphere extended southward to cover Okinawa/Amami. Meanwhile, the Okhotsk high, which brought cool wet northeasterly flows to the Pacific side of northern and eastern Japan, had persisted since late July. The persistence of the Okhotsk high was pre - sumed to be mainly due to blocking-high development over the Sea of Okhotsk, in association with the meandering westerly jet stream over northern Eurasia (Fig. SB7.5b). Corresponding to the PJ pattern with suppressed convec - tive activity over and around the Philippines, this anomalous atmospheric circulation in the lower troposphere brought F ig . SB7.5. (a) Stream function anomalies at 850-hPa − 1 2 - longer-than-normal sunshine durations, adiabatic heating associ s (contours) and OLR anomalies (m ; color shadings) averaged over 1–20 Aug 2017. Thick and thin contours ated with stronger-than-normal subsidence, and westerly warm 6 − 1 2 6 are intervals of 10 × 10 s - , respective m and 2.5 × 10 air inflow over Okinawa/Aamami. These factors contributed ly. For the NH, solid and dashed lines indicate anticy - to significantly warm conditions over Okinawa/Amami, and clonic and cyclonic circulation anomalies, respectively. monthly mean temperature over Okinawa/Amami in August For the SH, vice versa. (b) 500-hPa height anomaly in - 2017 was the highest on record for August since 1946. A south the NH averaged over 1–20 Aug 2017. Contours show ward extension of the Tibetan high presumably contributed to 500-hPa height at intervals of 60 m. Shading indicates the warm condition over the area. At the same time, the low- anomalies. Base period is 1981–2010. level anticyclonic circulation anomalies brought considerable The condition of enhanced NPSH over the south of Japan moisture to the middle and lower Yangtze River basin where persisted from August to October. It is considered to have been above-normal precipitation was observed. Meanwhile, due to caused by active convection over the Maritime Continent due both the Okhotsk high and the PJ pattern, the Pacific side of to positive SST anomalies in the western tropical Pacific, which northern and eastern Japan experienced significantly below- were related to the development of weak La Niña conditions normal sunshine duration. in the eastern equatorial Pacific (see Fig. 3.2). | S244 AUGUST 2018

265 h. Oceania and maximum temperatures for selected locations — verview 1) o C. Ganter across Micronesia are summarized in Table 7.4. The climate of Oceania experienced a neutral (ii) Precipitation ENSO state for most of 2017, which then transitioned to weak La Niña conditions late in the year. The Annual rainfall totals during 2017 were mostly Indian Ocean dipole (IOD) was neutral during its higher than average throughout Micronesia, with typical active period (May–November), while several below-average seasonal and annual rainfall amounts countries were inf luenced by a positive southern restricted to the northern Mariana Islands (Rota, Tinian, and Saipan), a few of the northern atolls of annular mode during the austral winter. Persistent strong stationary high pressure systems in the Tas - Chuuk State, and the northern atolls of the Republic of the Marshall Islands (RMI). The 6-month and - man Sea during November and December contrib annual rainfall values for selected locations across uted to warmer weather across southeast Australia Micronesia are summarized in Table 7.4. and New Zealand, and they also warmed surface waters of the Tasman Sea, causing a notable marine (iii) Notable events and impacts heatwave which persisted into 2018 (see Sidebar 7.4). Late in December 2016, and again in October 2017, orthwest acific and m icronesia p M. A. Lander 2) n a landfalling waterspout caused damage on an atoll — and C. P. Guard of Micronesia. In the former case, a large and intense - waterspout swept across Falalop (one of the islets of This assessment covers the area from the inter national dateline west to 130°E, between the equator the Ulithi Atoll). Eyewitnesses described a surge of and 20°N. It includes the U.S.-affiliated Islands of high wind that blasted across the islet, filling the air with lofted debris that appeared to be rotating. Micronesia but excludes the western islands of Kiri - bati and nearby northeastern islands of Indonesia. The Ulithi waterspout/tornado occurred in asso - ciation with deep convection in a near-core rainband Temperature and precipitation anomalies in this section are relative to a 1981–2010 period. of Tropical Storm Nock-ten. As reported in the 23 January 2017 issue of the Khaselehlia Press : Weather conditions across Micronesia during 2017 were mostly unremarkable. Annual rainfall was near to above average at most locations, and tropical - “On December 22, 2016, a water spout turned tor cyclone activity was much lower than average. The nado ripped through the island, tearing apart over western North Pacific summer monsoon system 20 newly repaired homes and cook houses along its path. “It sounded like a jet was f lying low over the was displaced to the west and north of Micronesia, island. Luckily, we had been warned that Typhoon accompanying a similar westward and northward displacement of the basin’s tropical cyclones. These Nock-ten could be headed in our direction so we were prepared for a potential disaster. If we hadn’t patterns of rainfall, wind, and typhoon distribution received warning about Nock-ten, this tornado were typical for an ongoing La Niña. The regional would have claimed lives on Falalop,” said local oceanic response to La Niña climate conditions (e.g., increased trade wind strength) was sustained higher- resident Jon Rumal Jr.” ... than-average mean sea level. It was not officially verified that this event was a (i) Temperature tornado, but eyewitness accounts are convincing. Temperatures across Micronesia in 2017 were The next incident of a landfalling waterspout oc - curred on 14 October when waterspouts were observed mostly above average. The warmth was persistent, at Nomwin Atoll in the Hall Islands of Chuuk State. with above-average readings occurring during most One of the waterspouts went ashore on Nomwin where or all the months of the year. Only Yap Island and - Pohnpei Island reported moderate negative tem “it was strong enough to topple banana trees, and weak infrastructure houses were down and damaged” as - perature departures for any of the time periods sum reported to the Chuuk Weather Service Office. A boat marized in Table 7.4. Saipan reported extraordinary warmth with daytime highs 3.42°C above average was found capsized in Nomwin waters on 15 October. during the second half of the year. The reason for It is thought by islanders that the boat was capsized by a waterspout. The Nomwin incident of waterspout Saipan’s excessive warmth (with many records set for highest daily maximum and highest minimum formation occurred in association with a large area of temperatures) is uncertain. The 6-month minimum heavy convective showers comprising the monsoon | S245 AUGUST 2018 STATE OF THE CLIMATE IN 2017

266 7.4. Temperature (°C) and rainfall (mm) anomalies for selected Micronesia locations during 2017. le AB t - Average (AVG) values are for the 1981–2010 base period. Latitudes and longitudes are approximate. “Kap inga” stands for Kapingamarangi Atoll in Pohnpei State, Federated States of Micronesia. Shading of the boxes indicates: red for above-average temperature and blue for below average; green for above-average rainfall and yellow for below average. M a x Te m p Rainfall (mm) Location Min Temp Jul–Dec Jul–Dec Jan–Jun Jan–Jun Jan–Jun Jul–Dec Jan–Jun Jan–Dec Jan–Dec Jul–Dec °C °C % 2017 % AVG 2017 % 2017 AVG +3.42 +2.63 Saipan, 65.1 118 . 9 1322 .8 861.3 534.2 1395. 5 78.8 449.1 15° N , 146° E +2 .61 +1.60 +0.80 Guam, 13°N, +0.70 130. 3 178 8 .4 1561.6 87. 5 2462.5 99.4 900.9 691.6 +0.82 145° E +0.64 − 0.70 − 0.43 Yap, 104.4 1902 .0 132.4 1548.4 116 9. 7 1985.0 115 . 0 3533.4 9°N, 138°E +0.01 +0.08 +0.97 +0.77 Palau, 2037. 8 1717. 6 4490.2 119. 7 120.7 2452.4 2032.5 118 . 6 +0.20 7°N, 134°E +0.18 +1.10 +0. 51 Chuuk, 1544.3 1584.2 101.5 1923.8 3468.1 104.9 1833.1 97. 6 7°N, 152°E +1.28 +1. 31 0.54 − 0.54 − Pohnpel, 2613.7 2266.4 4844.8 95.5 2231.1 2336.5 115 . 3 105.3 7°N, 158°E +1. 81 +1.92 Kapinga, 1750. 8 137. 5 130.7 1855.0 1510. 5 122 .8 4261.9 2406.9 — — 1°N, 155°E 0.04 − Kosrae, +0.09 3062.5 2342.9 5844.5 119. 0 119. 3 2782.1 118 . 7 2567.9 +2.03 5°N, 163°E +2.39 Majuro, +0.07 +0.09 2414.3 1368. 3 1739.1 127.1 1868.2 128.3 4153.4 129.2 +0.81 +0.65 7 ° N , 171° E +0.98 +0.65 Kwajalein, 1503.6 95.2 100.9 112 .1 1579.1 801.4 891.8 2395.5 +0.65 +0.53 9°N, 168°E in 2017 only resulted in mostly nuisance inundations depression that would become Tropical Storm Lan two days later. - at times of unusually high astronomical tides. How ever, two incidents of moderate inundation (both The sea level across Micronesia exhibits large f luctuations related to ENSO. During the 2015 El related to brief episodes of high wind and waves) Niño, the sea level dramatically fell across the region. occurred on the lagoon side of Majuro and on the During 2016, with the demise of El Niño, the sea level northeast coast of Kosrae. began a dramatic climb to become 6–8 cm above aver - age by the end of the year. The sea level remained well above average throughout Micronesia for all of 2017, as La Niña became established in the second half of the year. Interannual variations of sea level across Micronesia are F . 7.49. Time series of sea level at Guam (gray line) and Kwajalein (black ig almost entirely a result of line) since 1945. Blue line is a time series of NOAA’s trade wind index for the forcing by the Pacific trade region. Note the rapid rise in all times series at the end of the 1990s through wind system (blue line in the mid-2000s coinciding with La Niña, and also the sharp minima during Fig. 7.49). Fortunately, due strong El Niño events (e.g., 1983, 1997, and 2015). There is a sharp rise of sea to a general lack of high surf level in 2016 that remained high during 2017. A 12-month moving average has and swell, the high sea levels been applied to the raw monthly values of each time series. | S246 AUGUST 2018

267 Low rainfall from late 2016 through September 2017 resulted in drought emergencies in most of the atolls in the northern RMI. Loss of potable water sources required the emergency shipment of bottled water, reverse osmosis systems, emergency food supplies, and public health and hygiene assistance. During this period the Guam Weather Forecast Of - fice issued 18 bi-monthly specially tailored drought information statements for the RMI Government and U.S. State Department relief agencies. A. Peltier — p acific 3) s outhwest . 7.51. Monthly average surface air temperature ig F Countries considered in this section include: anomalies (°C, 1981–2010 base period) for the entire American Samoa, the Cook Islands, Fiji, French southwest Pacific area (25°–10°S and 156°–226°E) from Polynesia, Kiribati, Nauru, New Caledonia, Niue, 1981–2017. (Source: NOAA NCEP CPC CAMS.) Papua New Guinea (PNG), Samoa, the Solomon - Islands, Tokelau, Tonga, Tuvalu, Vanuatu, and Wal Apart from the equatorial area, which was average lis and Futuna. Air temperature and precipitation to below average, the region experienced relatively anomalies are relative to a 1981–2010 period unless large positive temperature anomalies until April. otherwise indicated. - While steadily declining in magnitude, tempera ture anomalies remained positive until the end of (i) Temperature the austral winter. In October, surface temperature anomalies leaned toward negative values over French The southwest Pacific experienced warmer-than- average conditions throughout most of 2017. The only Polynesia and the Cook Islands. By the end of Decem - ber, near to below-average temperatures covered most exceptions were islands close to the equator such as of the Pacific island countries east of the dateline. - Nauru and Kiribati, where near or below-average tem peratures were observed, as ENSO-neutral conditions (ii) Precipitation prevailed for most of the year (Fig. 7.50). - In Nauru and the Kiribati Gilbert Islands, which The year started with exceptionally warm condi - are located near the equator west of the dateline, tions across the region. In January, monthly anoma drier-than-average conditions prevailed during the lies exceeded two standard deviations over a wide first half of 2017, as cloudiness was reduced in both area east of the dateline encompassing Tuvalu, Tonga, the intertropical convergence zone and the north - Wallis and Futuna, Samoa, American Samoa, and western sector of the southern Pacific convergence - the Cook Islands. This resulted in the second high est monthly anomaly of +0.96°C, behind the record zone (SPCZ). After two months of respite in July and set in December 1998 (+0.99°C) for the temperature August with above-average precipitation, relatively anomaly averaged over the southwest Pacific area dry weather conditions resumed during the last four 25°–10°S and 156°E–226°E (Fig. 7.51). months of the year with ENSO trending towards La Niña. Precipitation totals for 2017 were as low as 50% of average in Nauru and the Gilbert Islands. These dry conditions in 2017 were an extension of a drought event that began during the second half of 2016. Farther south, the transi - tion from a neutral ENSO state to a weak La Niña re - sulted in a rainfall pattern close to normal across most . 7.50. Annual average surface air temperature anomalies (°C, 1981–2010 ig F of the South Pacific from base period; Source: NOAA NCEP CPC CAMS.) January to October. Yet, two | S247 AUGUST 2018 STATE OF THE CLIMATE IN 2017

268 in the city. Major roads were submerged and f looding was reported in the main hospital and in many resi - dences. All of the Guadalcanal plains were covered with water, and farmers who supply vegetables to Honiara lost around 70%–80% of their production. New Caledonia experienced its driest winter on record. Cold fronts that usually bring precipitation to the island passed farther south than usual from mid-June to mid-July as a consequence of a positive southern annular mode. While the southern annu - lar mode returned to neutral by August and frontal systems returned to their near-average position, most frontal systems rapidly dissipated while approaching New Caledonia. In Noumea, the capital city, the win - ter precipitation total was 33% of normal, the lowest on record (since 1951). Spring (SON) is usually the driest season in New Caledonia, and 2017 was no exception. The rain showers in early December ended the longest sequence of daily rainfall totaling 5 mm or less: 139 days for Noumea. As a result, New Caledonia experienced agricultural and hydrological droughts . 7.52. Precipitation anomalies (% of normal wrt F ig in the second half of 2017. The main consequences 1951–2000): (a) Feb 2017, (b) Jun–Aug 2017. (Source: GPCC Monitoring Product version 5.) were restrictions on drinking water and fires that destroyed vast areas of vegetation, including primary notable periods stood out. The first of these saw forests, despite a territory-wide fire ban. the SPCZ become remarkably enhanced in early February in response to an active Madden–Julian oscillation (MJO) phase over the western equatorial S. Tobin and S. J. Jacobs — ustralia 4) a The base period for this section is 1981–2010. Pacific. This resulted in well above-average Febru - Nationwide monthly average temperatures are based ary rainfall over many Pacific nations (Fig. 7.52a): on the ACORN-SAT dataset (Trewin 2013), which the Solomon Islands (170% of average), Fiji (160%), extends to 1911. The rainfall and daily temperatures Wallis and Futuna (190%), Tonga (250%), and Niue are based on the AWAP dataset (Jones et al. 2009), (350%). The second notable period saw relatively dry which extends to 1910. conditions prevail between June and September west of the dateline, resulting in precipitation as low as 60% of average in Fiji, 40% in Vanuatu, and 30% in New (i) Temperature Caledonia for the austral winter season (Fig. 7.52b). The 2017 annual mean temperature for Australia was 0.64°C above the 1981–2010 average, its third As oceanic indicators leaned towards La Niña in the warmest year on record. The decade ending 2017 was tropical Pacific, cloudiness patterns across the region 0.30°C higher than average and the warmest 10-year also showed a clear La Niña signal. In November period in Australian records. and December, the SPCZ was slightly enhanced and Australian mean maximum temperatures (Fig. displaced towards the southwest. This resulted in above-average rainfall across the Solomon Islands, 7.53) were 0.97°C above average, the second highest on record. Mean minimum temperatures (Fig. 7.54) Vanuatu, Fiji, Tonga, and Niue. were 0.31°C above average, the 11th highest on record. Annual mean temperatures were above average for (iii) Notable events and impacts On 7 February, a trough over the Solomon Islands almost all of Australia and record high for much of and a slow-moving tropical low pressure system locat - the southern half of Queensland, northwestern New South Wales, and an area on the central coast of New ed in the Coral Sea caused heavy rain in and around Honiara, the capital city, situated on the northwestern South Wales between Sydney and Port Macquarie. - Maxima were above average for nearly all of Aus coast of Guadalcanal. The rainfall total for that day tralia and in the highest 10% of observations for nearly was 208 mm, the sixth highest since records began in 1949 (Source: Global Historical Climate Network - all of eastern Australia, South Australia, and most of the Northern Territory. Maxima were the highest on Daily). The heavy downpour triggered f lash f looding | S248 AUGUST 2018

269 F i g . 7.54. Minimum temperature anomalies (°C) ig . 7.53. Maximum temperature anomalies (°C) for F for Australia, averaged over 2017, relative to a Australia, averaged over 2017, relative to a 1981–2010 1981–2010 base period. (Source: Australia Bureau of base period. (Source: Australia Bureau of Meteorology.) Meteorology.) record for much of the southern half of Queensland In contrast, daytime temperatures during July were and some parts of northern New South Wales. particularly warm, with maxima highest on record Minima were in the highest 10% of observations for much of northern Australia. Exceptional warmth for much of Queensland, northern and eastern New - was present during September–December, largely as South Wales, southwest Victoria, parts of coastal sociated with blocking highs over the Tasman Sea and South Australia, western Tasmania, and parts of particularly affecting eastern Australia (see Notable events and impacts and Sidebar 7.4). the Top End in the Northern Territory. Minima were highest on record for an area of southwest These prolonged warm spells on land also af - Queensland. Minima were above average for much fected the surrounding oceans. For the Tasman Sea of the rest of Australia but cooler than average for an region, October, November, and December were area of inland northwestern Australia spanning the each warmest on record for their respective months. border between Western Australia and the Northern This occurred without a southward extension of the Territory. warm East Australian Current and in the absence of El Niño—both of which contributed to exceptional Persistent warmth was featured throughout 2017. Daytime temperatures were especially warm, with sea surface temperatures (SSTs) in the region during the first half of 2016. monthly mean maxima ranking among the ten Annual SSTs for the Australian region were the warmest on record for March, each month from May eighth highest on record, based on ERSSTv5 data. to September, and December. Above-average annual SSTs have been observed January to March was much warmer than average for eastern Australia, while above-average rainfall each year from 1995 to 2017 (inclusive), with a range of negative effects on the marine environment. Pro - kept days cooler than average in the northwest. Clear - skies associated with a persistent strong high pres longed high SSTs led to significant coral bleaching on sure ridge across the country contributed to warm, the Great Barrier Reef during early 2017, following record bleaching during summer 2015/16. This is sunny days and cooler-than-average nights during the first time mass bleaching events have occurred in late autumn and winter. June nights were much cooler consecutive years and in the absence of El Niño (see than average for much of the southern mainland. Southeast Australia experienced cooler-than-average Sidebar 3.1 for more details). nights for a longer period, extending into September. | S249 AUGUST 2018 STATE OF THE CLIMATE IN 2017

270 (ii) Precipitation pressure systems and cold fronts crossing southern Australia. A climate change signal has been identified 2017 was a year of contrasts for rainfall, with a in the observed increase in the strength of the sub- wet start, a dry middle, and a wet end. Averaged across Australia, rainfall for 2017 was 504 mm, 4% tropical ridge and reduction of cool season rainfall in southern Australia over recent decades (Timbal and above the 1981–2010 average, the 30th wettest in the Drosdowsk y 2012). 118-year record. Annual rainfall was above average for the southeast, interior, and far north of Western June was the second driest on record nationally, Australia, for most of the Northern Territory, and for and the driest on record for southeastern Australia as a whole (land area south of 33°S, and east of 135°E), the west of South Australia. Large parts of Western while September was the driest on record for the Australia had annual rainfall in the highest 10% of their records (Fig. 7.55). Rainfall was below average Murray–Darling Basin. for most of inland Queensland, most of New South October rainfall was above average for much of Australia, with f looding on the east coast of Wales, eastern to central Victoria, all of Tasmania, Queensland around Bundaberg and Tully. For and pockets of the west coast of Western Australia. Queensland, it was the third wettest October on January and February rainfall was above average across the western half of Australia, while February record. November rainfall was generally average to rainfall was below average for large parts of eastern above average, and while December was drier than average for Queensland and the Northern Territory, Australia. heavy rainfall events in southeast Australia and In March, heavy rainfall events in New South Tropical Cyclone Hilda in Western Australia resulted Wales and Victoria, and Severe Tropical Cyclone in above-average monthly rainfall across large areas. Debbie in Queensland and northern New South The main natural climate drivers for Australia— Wales, brought above-average monthly rainfall along ENSO and the IOD—were in a neutral phase for most the east coast. From April to September rainfall was generally of the year. However, cooler-than-average waters to below average, particularly over southeastern Aus - Australia’s west and warmer-than-average waters to tralia. A positive southern annular mode (SAM) and the east of Africa to the south of the IOD regions cre - strong subtropical ridge contributed to below-average ated a strong temperature gradient across the Indian winter rainfall by shifting the belt of westerly winds Ocean during the year, exerting a drying inf luence on Australia. southward, resulting in fewer rain-bearing low (iii) Notable events and impacts Exceptional warmth affected large parts of eastern Australia from late December 2016 into February 2017. Records were set in southeastern Australia and southern Queensland for consecutive warm days or nights, or for total number of warm days or nights during January. Five separate locations in Queensland broke previous state records for hottest February day on the 12th. The McArthur Forest Fire Danger Index (FFDI) reached catastrophic levels across much of New South Wales on 12 February. A fire in the Warrumbungle Shire destroyed most of the small township of Uarbry. Slow-moving tropical lows brought heavy rain over much of northern and western Australia between late January and early February. Cumulative rainfall resulted in f looding in the Kimberley and in parts of southwest Western Australia, the latter of which typically has low summer rainfall. Flooding affected large areas of the east coast during March, resulting from thunderstorms in New . 7.55. Rainfall deciles for Australia for 2017, based ig F on the 1900–2017 distribution. (Source: Australia Bu - South Wales around mid-month, thunderstorms in reau of Meteorology.) Victoria on 20 and 21 March, and Severe Tropical | S250 AUGUST 2018

271 — 5) n ew z ealand B. E. Noll Cyclone Debbie at the end of the month. Debbie In the following discussion, the base period is caused f looding and widespread wind damage in 1981–2010, unless otherwise noted. The nationwide Queensland and northeastern New South Wales, average temperature is based upon the National In - with f looding continuing into April in some rivers. An exceptional period of warm weather during stitute of Water and Atmospheric Research (NIWA) seven-station temperature series that began in the last week of September saw many records for high temperatures or early season warmth set in eastern 1909 (see www.niwa.co.nz/our-science/climate /information-and-resources/nz-temp-record Australia (see Sidebar 7.4). /seven-station-series-temperature-data). All statistics In early October, heavy rainfall associated with are based on data available as of 9 January 2018. surface and upper-level troughs affected southeast - ern Queensland and northeastern New South Wales, with f looding around Bundaberg. Further heavy (i) Temperature rain midmonth affected the same region, as well as According to NIWA’s seven-station temperature areas of Queensland’s tropical coast, with f looding series, 2017 was New Zealand’s fifth warmest year around Tully. since records began in 1909. The nationwide average temperature for 2017 was 13.15°C, 0.54°C above the - After a cool and frosty start to November an ex annual average. Annual mean temperatures were tended period of very warm weather affected Victoria near average (within 0.5°C of the annual average) and Tasmania, driven by long-lived blocking highs or above average (0.51°–1.20°C above the annual over the Tasman Sea during both November and December (see also Sidebar 7.4). November monthly average) throughout the country (Fig. 7.56a). Only January observed a below-average nationwide tem - mean temperatures were the highest on record for perature (0.7°C below average). The three months in Tasmania and second highest for Victoria. Warmth was more widespread in December, affecting all states 2017 with the largest national temperature anomalies and the Northern Territory. These high pressure - were December (+2.4°C), August (+1.3°C), and No systems also contributed to record high sea surface vember (+1.1°C). These marked New Zealand’s second warmest December, third warmest August, and sixth temperatures for Bass Strait and the Tasman Sea as clear skies allowed more solar radiation absorption, warmest November on record. The warmth seen in November and December were likely inf luenced by and light winds limited mixing of surface waters. synoptic patterns which also contributed to the ex An inf lux of tropical moisture between 1 and 3 - December brought two to three times the month - ceptional warmth also experienced across southeast ly average rainfall to - parts of northern Vic toria and southern New South Wales. Flooding resulted in central to northeastern Victoria, with some f lash f lood - ing around Melbourne. For further detail on these and other sig - nificant events please see Monthly Weather Reviews, Special Cli - mate Statements, and the Annual Climate Statement—all available from www.bom.gov.au /climate/current/. F ig . 7.56. 2017 annual (a) mean temperature anomaly (°C) and (b) total rainfall (%), relative to 1981–2010. (Source: NIWA.) | S251 AUGUST 2018 STATE OF THE CLIMATE IN 2017

272 Australia and the Tasman Sea (see Sidebar 7.4). The Waikato, Taranaki, Manawatu-Whanganui, Wel - highest recorded air temperature for 2017 was 35.5°C, lington-Wairarapa, Hawke’s Bay, Marlborough, Tas - at Wairoa (Hawke’s Bay) and Ashburton (Canterbury) man, the West Coast, and Southland. In December, on 6 February (see Fig. 7.57 for localities). The lowest the Ministry for Primary Industries classified the drought as a medium-scale adverse event in Taranaki, recorded air temperature for 2017 (excluding high western parts of the Manawatu-Whanganui region, altitude alpine sites) was −14.6°C, observed at Lake and around Wellington. The drought conditions Tekapo (Canterbury) on 29 July. occurred following a wet start to the year, which fea - (ii) Precipitation tured significant rain impacts, especially across the North Island. These included two ex-tropical cyclones Annual rainfall totals for 2017 were above average (Cook and Debbie), which affected the country during (120%–149% of the annual average) in Auckland, April, following a heavy rainstorm between 7 and 12 Waikato, Bay of Plenty, coastal Canterbury, and north coastal Otago (Fig. 7.56b). On the other hand, rainfall March. The impact from ex-Tropical Cyclone Debbie (4 April) led to several one-day rainfall records for the was below average (50%–79% of the annual average) - month of April across the Bay of Plenty, which con - across much of Southland and interior Otago. Else tributed to severe f looding in parts of the region, and where, 2017 annual rainfall totals were near average (within 20% of the annual average). Five locations was particularly notable for the town of Edgecumbe. Oamaru (Otago) had its second wettest year on observed near-record high annual rainfall totals while three locations observed record or near-record low record (813 mm of rain). On 21 July, 161 mm of rain rainfall totals. fell, leading to f looding and making it the wettest day Of the regularly reporting rainfall gauges, the in the town since records began in 1950; thereafter, wettest location in 2017 was Cropp River, in the Oamaru recorded just 163 mm during the remainder Hokitika River catchment (West Coast, South Island, of the year (August–December 2017). 975 m above sea level), with an annual rainfall total of 8662 mm (76% of the long-term average). The driest of the regu - larly reporting rainfall sites in 2017 was Clyde (Central Otago), which recorded 278 mm of rainfall (67% of the long-term average). Milford Sound (South - land) experienced the - highest one-day rain fall total in 2017: 309 mm on 31 January. (iii) Notable events and impacts Figure 7.57 provides a schematic of notable events. By the end of 2017, parts of eleven of New Zealand’s sixteen geographical regions had experienced me- teorological drought following a dry No - vember and December. These regions included ig . 7.57. Notable weather events and climate extremes for New Zealand in 2017. F (Source: NIWA.) Northland, Auckland, | S252 AUGUST 2018

273 S I D E B A R 7.4 : SUMMER ARRIVES EARLY IN AUSTRALIA AS THE AUSTRAL SPRING BREAKS RECORDS —S. TOBIN AND S. J. JACOBS Two significant heat events occurred during the Australian spring of 2017. The first event in September broke temperature records across eastern - Australia while the second broke dura tion records in Victoria and Tasmania. A high pressure system moved over New South Wales on 20 September. As a result, large parts of eastern and northern Australia had sunny, cloud-free - days. With low rainfall through Septem ber and below-average soil moisture, these sunny days led to rapid heating of the land surface and overlying air in Central Australia, Queensland, and New South Wales. By 22 September, the high pressure system became slow moving over the - northern Tasman Sea while a low pres sure system developed to the south of Australia. The two weather systems directed hot, dry air from the desert 0 12 −12 −8 −4 84 interior into eastern Australia, causing Anomaly (°C) unprecedented hot September weather. F . SB7.6. Maximum temperature difference (°C) from the 1981–2010 ig - The 22nd was Australia’s warmest Sep average for 23 Sep 2017. (Source: Australia Bureau of Meteorology.) tember day since national area-averaged The heat returned in mid-to-late November when analyses commenced in 1911, although the highest local - the southeastern states of Victoria and Tasmania expe temperatures were observed on the 23rd. Maximum rienced an unusually long run of warm days and nights. temperatures were more than 12°C above the 1981–2010 A long-lived blocking high was again responsible for the average across much of the mainland southeast on the high temperatures, but this time the center was over 23rd (Fig. SB7.6). A number of sites in New South Wales the southern Tasman Sea, directing the hot, dry air into reached 40°C, the first such occurrences in the state southern states from Central Australia. during September, while in Victoria, Mildura set a state The extended warm spell lasted from 10 November record for September reaching 37.7°C. until the start of December when a strong trough accom - A new high pressure system crossed the southeast into panied by heavy rain crossed the states. This heat wave - the Tasman Sea between the 26th and 27th, bringing re was notable for its duration rather than its intensity, with newed heat to eastern Australia. During this period some many records set for consecutive days with maximum or New South Wales sites broke the records they had set minimum temperatures above thresholds and only a few only days earlier, while in Queensland, Birdsville reached records set for individual days. Events of this duration - 42.8°C on the 27th, setting a new state record for Sep are unusual in spring when weather systems are normally tember. By 29 September more than 20% of Australia (by more mobile than in late summer and autumn. area) had recorded its hottest September day on record. | S253 AUGUST 2018 STATE OF THE CLIMATE IN 2017

274 CON'T ’ S I D E B A R 7.4 : SUMMER ARRIVES EARLY IN AUSTRALIA —S. TOBIN AS THE AUSTRAL SPRING BREAKS RECORDS AND S. J. JACOBS During this event Melbourne experienced six consecu - tive days with maxima of at least 30°C and nine consecu - tive days of at least 28°C—the latter breaking the previous record of six days set in 2009. Melbourne also experienced 14 consecutive nights above 15°C, solidly surpassing the previous spring record of nine days in 2009. A warm spell of this length had not occurred in Melbourne before mid- summer (January) in at least the 108-year record. In Tasmania, the length of the late spring warm spell was unprecedented for any time of year, particularly in the south and west. Many locations set November records for consecutive days above 25°C. Strahan, on the west coast, had 18 consecutive days (from 13 to 30 November) of maximum temperatures 21°C or above, including seven consecutive days of 27°C or above, both records for any time of year. Hobart’s six consecutive days of 25°C or above equaled the record for any time of year, while its five consecutive nights above 15°C was a November record. The prolonged heat event also caused a marine heat - wave to form around Tasmania due to the clear skies and light winds associated with the blocking high over the Tasman Sea. The surface waters increased to 3°C above the 1971–2000 November average and ranked among the highest values on record in that region. The marine heatwave persisted into the austral summer with November, December, and January 2018 monthly sea surface temperatures highest on record for large areas around Tasmania and extending to the western coast of New Zealand (Fig. SB7.7). For further details on both events, see Special Climate Statement 62 and 63: www.bom.gov.au/climate/current /statements/scs62.pdf and www.bom.gov.au/climate/cur F . SB7.7. SST deciles for (a) Nov and (b) Dec 2017 ig and (c) Jan 2018, based on the 1900–2017 distribution using the NOAA ERSST v 5 dataset. (Source: Australia Bureau of Meteorology.) | S254 AUGUST 2018

275 APPENDIX 1: RELEVANT DATASETS AND SOURCES General Variable or Specific Dataset or Source Section Phenomenon Variable CAMS Reanalysis 2g3 Aerosols https://atmosphere.copernicus.eu/ CERES Energy Balanced and https://ceres.larc.nasa.gov/ 3e1 Filled https://eosweb.larc.nasa.gov/project/ceres/ebaf CERES FLASHflux 3e1, 3e4 Air-sea fluxes _surface_table Woods Hole Oceanographic 3e1, 3e3, http://oaflux.whoi.edu Institute OAFlux 3e4 MODIS http://ladsweb.nascom.nasa.gov 2h1, 5e3 Albedo GFAS v1.4 ftp://ftp.mpic.de/GFAS/sc17 2h3 https://daac.ornl.gov/VEGETATION/guides/fire Biomass, Greenness GFEDv4 2h3 _emissions_v4.html or Burning MODIS NDVI https://ecocast.arc.nasa.gov/data/pub/gimms/ 5h https://ladsweb.modaps.eosdis.nasa.gov/api/v1 2d5 Aqua MODIS C6 /productPage/product=MYD06_L2 https://eosweb.larc.nasa.gov/project/calipso 2d5 CALIPSO /calipso_table https://ceres.larc.nasa.gov/science_information CERES MODIS 2d5 .php?page=ModisCloudRetr https://climatedataguide.ucar.edu/climate-data /clara-a1-cloud-properties-surface-albedo-and CLARA-A2 2d5 -surface-radiation-products-based-avhrr Clouds, Cloudiness www.esa-cloud-cci.org 2d5 CLOUD_CCI HIRS 2d5 www.ssec.wisc.edu/~donw/PAGE/CLIMATE.HTM MISR https://l0dup05.larc.nasa.gov/L3Web/ 2d5 www.ncdc.noaa.gov/cdr/atmospheric/avhrr-cloud PATMOS -x /AVH RR 2d5 -properties-patmos-x PATMOS-x/MODIS C6 http://ladsweb.nascom.nasa.gov 2d5 SatCORPS No public archive 2d5 Coral Bleaching Heat Coral Reef Watch https://coralreefwatch.noaa.gov/product/5km/v3.1/ SB3.1 Stress 2d9 scPSDI https://crudata.uea.ac.uk/cru/data/drought/ Drought https://crudata.uea.ac.uk/cru/data/hrg/ 2d9 CRU TS 3.26 Evaporation, Interception, GLEAM www.gleam.eu/ 2d10 Transpiration, Sublimation MERIS http://earth.esa.int/level3/meris-level3/ 2h2 FAPAR 2h2 http://ladsweb.nascom.nasa.gov/ MODIS -TIP SeaWiFS v 2010.0 http://fapar.jrc.ec.europa.eu/ 2h2 www.ecmwf.int/en/research/climate-reanalysis 6b ERA-Interim /era-interim Geopotential Height NCEP/NCAR Reanalysis 1: www.esrl.noaa.gov/psd/data/gridded/data.ncep 4f2, 5b .reanalysis.pressure.html Pressure S255 | STATE OF THE CLIMATE IN 2017 AUGUST 2018

276 General Variable or Specific Dataset or Source Section Phenomenon Variable https://earth.esa.int/web/guest/missions/esa Cryosat-2 5f -operational-eo-missions/cryosat Norwegian Water 5f Resources and Energy http://glacier.nve.no/Glacier/viewer/CI/en/nve Glacier Mass or Directorate Volume Randolph Glacier Inventory www.glims.org/RGI/ 5f v3.2 World Glacier Monitoring https://wgms.ch/latest-glacier-mass-balance-data/ 2c3, 5f Service Groundwater and GRACE https://gracefo.jpl.nasa.gov/data/grace-data/ 2d7 terrestrial water storage 2d1 by email to [email protected] Dai www.ecmwf.int/en/forecasts/datasets/archive ERA-Interim 2d1 -datasets/reanalysis-datasets/era-interim HadCRUH www.metoffice.gov.uk/hadobs/hadcruh 2d1 HadISDH www.metoffice.gov.uk/hadobs/hadisdh 2d1 Humidity, [Near] https://doi.org/10.5676/EUM_SAF_CM/HOAPS HOAPS 2d1 Surface /V001 JRA-55 Atmospheric http://jra.kishou.go.jp/JRA-55/index_en.html 2d1 Reanalysis MERRA-2 https://gmao.gsfc.nasa.gov/reanalysis/MERRA-2/ 2d1 http://artefacts.ceda.ac.uk/badc_datadocs/nocs 2d1 NOCS 2.0 _flux/nocs_flux.html www.ncdc.noaa.gov/cdr/fundamental/hirs-ch12 HIRS 2d3 -brightness-temperature Humidity, Upper www.esrl.noaa.gov/psd/data/gridded/data.ncep 4f4 NCEP/NCAR Reanalysis Atmosphere .reanalysis.html 2d3 UTH by email to [email protected] DMSP-SSMIS http://nsidc.org/data/nsidc-0032 5e1, 6d http://podaac.jpl.nasa.gov/datasetlist?ids=Platform Ice Sheet GRACE 5e4 &values=GRACE Characteristics www.promice.dk/home.html 5e2 PROMICE (Greenland) Advanced Along-Track https://earth.esa.int/web/guest/missions/esa Scanning Radiometer 5e2 -operational-eo-missions/envisat/instruments/aatsr (AATSR) https://earth.esa.int/web/guest/missions/esa 5e2 AT S R -operational-eo-missions/ers/instruments/atsr AVHRR https://lta.cr.usgs.gov/AVHRR 5e2 Globolakes www.globolakes.ac.uk 2b2 Lake Vättern (Sweden) Vättern Water Protection Association 2b2 Lake Temperature City of Zurich Water Supply and Amt für Abfall, 2b2 Lake Zurich (Switzerland) Wasser, Energie und Luft of the Canton of Zurich Mondsee (Austria) http://hydro.ooe.gv.at/#Startseite 2b2 http://wasser.bgld.gv.at/hydrographie/online Neusiedler See (Austria) 2b2 -daten.html Polish Lakes www.imgw.pl 2b2 https://info.ktn.gv.at/asp/hydro/daten/hydroportal Wörther See (Austria) 2b2 /see_wt.asp S256 | AUGUST 2018

277 General Variable or Specific Dataset or Section Source Variable Phenomenon www.cpc.ncep.noaa.gov/products/precip/CWlink Arctic Oscillation (AO) 2e1 /daily_ao_index/teleconnections.shtml Madden-Julian Oscillation www.bom.gov.au/climate/mjo/graphics (MJO) - Real-time 4c /rmm.74toRealtime.txt Multivariate MJO Multivariate ENSO index www.esrl.noaa.gov/psd/enso/mei/ 2b4, 3i (MEI) www.cpc.ncep.noaa.gov/products/analysis 3c Niño3.4 (detrended) _monitoring/ensostuff/detrend.nino34.ascii.txt https://climatedataguide.ucar.edu/climate-data North Atlantic Oscillation /hurrell-north-atlantic-oscillation-nao-index 2e1 (NAO) -station-based North Atlantic Oscillation - 2e1 by request from Folland Summer (SNAO) Modes of Variability www.cpc.ncep.noaa.gov/products/analysis Oceanic Niño Index (ONI) 2d6, 4b _monitoring/ensostuff/ensoyears.shtml Pacific Decadal Oscillation www.cpc.ncep.noaa.gov/products/GODAS/ 2d6 (PDO) Quasi-biennial Oscillation multiple inputs 2b5 (QBO) Southern Annular Mode 6c www.antarctica.ac.uk/met/gjma/sam.html (SAM) Southern Annular Mode www.cpc.ncep.noaa.gov/products/precip/CWlink 2e1, 6b (SAM, AAO) /daily_ao_index/aao/aao.shtml Southern Oscillation Index ftp://ftp.bom.gov.au/anon/home/ncc/www/sco/soi 2e1 /soiplaintext.html (SOI) Southern Oscillation Index 6c www.cpc.ncep.noaa.gov/data/indices (EQ-SOI) pCO2 www.socat.info 3j1 http://hahana.soest.hawaii.edu/hot/products/HOT 3j2 pCO2 _surface_CO2.txt http://hahana.soest.hawaii.edu/hot/products/HOT 3j2 pH _surface_CO2.txt Ocean Carbon Global Ocean Ship-Based Hydrographic Investigations www.go-ship.org 3j3 Program GLODAPv2 https://odv.awi.de/data/ocean/glodap-v2-bottle-data/ 6f https://soccom.princeton.edu/content/float-data SOCCOM 6f, SB6.1 Atlantic Meridional 3h www.rapid.ac.uk/rapidmoc/rapid_data/datadl.php Overturning Circulation Ocean Circulation www.oceansites.org/data/ 3h MOVE XBT Data www.aoml.noaa.gov/phod/hdenxbt/index.php 3h www.cmar.csiro.au/sealevel/thermal_expansion CSIRO/ACE CRC/IMAS- 3c UTAS estimate _ocean_heat_timeseries.html https://climatedataguide.ucar.edu/climate-data IAP/CAS /ocean-temperature-analysis-and-heat-content 3c -estimate-institute-atmospheric-physics Ocean Heat Content http://oceans.pmel.noaa.gov 3c PMEL/JPL/JIMAR www.data.jma.go.jp/gmd/kaiyou/english/ohc/ohc 3c MRI/JMA _global_en.html http://apdrc.soest.hawaii.edu/datadoc/godas NCEP Ocean Reanalysis 4h _pentad.php S257 | STATE OF THE CLIMATE IN 2017 AUGUST 2018

278 General Variable or Specific Dataset or Source Section Phenomenon Variable NCEI www.nodc.noaa.gov/OC5/3M_HEAT_CONTENT/ 3c Ocean Heat Content www.metoffice.gov.uk/hadobs/en4/download (continued) 3c UK Met Office EN4.0.2 -en4-0-2-l09.html NASA Gravity Recovery and https://grace.jpl.nasa.gov/data/get-data/monthly 3f Ocean Mass Climate Experiment -mass-grids-ocean/ Aquarius V3.0 3d1 http://podaac.jpl.nasa.gov/aquarius www.argo.ucsd.edu, http://argo.jcommops.org Argo 3c, 3d1, 6f Blended Analysis for Surface ftp://ftp.cpc.ncep.noaa.gov/precip/BASS 3d1, 3d2 Ocean Salinity Salinity CARICOOS Underwater www.aoml.noaa.gov/phod/goos/gliders/index.php 4h Gliders 3d1, 3d2 www.nodc.noaa.gov/OC5/woa13/ World Ocean Atlas 2013 CERES FLASHFlux Project http://flashflux.larc.nasa.gov 3e1, 3e4 Outgoing Longwave 4b2, 4c, www.ncdc.noaa.gov/cdr/atmospheric/outgoing Radiation Daily OLR -longwave-radiation-daily 4f3, 4f6 www.bodekerscientific.com/data/total-column Bodeker Scientific 5j -ozone CCMI REFC2 www2.acom.ucar.edu/gcm/ccmi-output 2g4 GOME/SCIAMACHY/ www.iup.uni-bremen.de/gome/wfdoas/ GOME2 (GSG) Merged 2g4 Total Ozone GOME/SCIAMACHY/ GOME2 (GTO) Merged www.esa-ozone-cci.org 2g4 Total Ozone GOZCARDS ozone profiles https://gozcards.jpl.nasa.gov/info.php 2g4 https://disc.sci.gsfc.nasa.gov/Aura/data-holdings /MLS ftp://toms.gsfc.nasa.gov/pub/omi/data/ozone/ (for years 2013-2017) Aura OMI/MLS 5j, 6g https://disc.gsfc.nasa.gov/datasets/OMTO3d_003 Ozone, Total Column /summary and Stratospheric https://disc.gsfc.nasa.gov/datasets/OMUVBd_003 /summary Multi Sensor Reanalysis www.temis.nl 2g4 (MSR-2) of total ozone NASA BUV/SBUV v8.6 http://acdb-ext.gsfc.nasa.gov/Data_services/merged 2g4 (MOD v8.6) Merged Ozone NOAA BUV/SBUV v8.6 ftp://ftp.cpc.ncep.noaa.gov/SBUV_CDR 2g4 (MOD v8.6) Merged Ozone Ozone Mapping & Profiler 6g https://ozoneaq.gsfc.nasa.gov/omps/ Suite (OMPS) Ozonesonde www.esrl.noaa.gov/gmd/dv/spo_oz 6g SAGE II/OSIRIS dataset linked to Bourassa et al. (2014) 2g4 www.esrl.noaa.gov/csd/groups/csd8/swoosh/ SWOOSH 2g4 https://woudc.org/archive/Projects-Campaigns WOUDC Ground-based 2g4 /ZonalMeans/ Ozone http://acd-ext.gsfc.nasa.gov/Data_services/cloud Aura OMI/MLS 2g6 _slice/new_data.html Ozone, Tropospheric 2g6 NOAA Observatory Data ftp://aftp.cmdl.noaa.gov/data/ozwv/SurfaceOzone/ Tropospheric Ozone https://doi.pangaea.de/10.1594/PANGAEA.876108 SB2.3 Assessment Report S258 | AUGUST 2018

279 General Variable or Specific Dataset or Source Section Variable Phenomenon Active Layer Thickness 5g2 www2.gwu.edu/~calm/ Global Terrestrial Network http://gtnp.arcticportal.org 2c1 for Permafrost (GTN-P) Permafrost Temperature 5g1 http://permafrost.gi.alaska.edu/sites_map Permafrost Temperature at permafrance.osug.fr 2c1 French sites Permafrost Permafrost Temperature at www.tspnorway.com 2c1, 5g1 www.met.no Norwegian sites Permafrost Temperature at https://bolin.su.se/ 5g1 Swedish sites Permafrost Temperature at www.permos.ch 2c1 Swiss sites MODIS-Aqua R2018.0 http://oceancolor.gsfc.nasa.gov/cms/reprocessing/ 3i Phytoplankton, 3i SeaWiFS R2018.0 http://oceancolor.gsfc.nasa.gov/cms/reprocessing/ Ocean Color http://oceancolor.gsfc.nasa.gov/cms/reprocessing/ VIIRS R2018.0 3i www.cpc.ncep.noaa.gov/products/janowiak CMORPH 4d1, 4d2 /cmorph_description.html www.ncdc.noaa.gov/temp-and-precip/ghcn-gridded GHCN 2d4 -products/precipitation SB2.1 GHCNDEX www.climdex.org/datasets.html Precipitation 2d4, 4e, GPCP v2.3 http://precip.gsfc.nasa.gov 4h 2d4, www.dwd.de/EN/ourservices/gpcc/gpcc.html GPCC SB2.1 JRA-55 Atmospheric http://jra.kishou.go.jp/JRA-55/index_en.html 6c Precipitation (net), Reanalysis Freshwater Flux http://precip.gsfc.nasa.gov, http://oaflux.whoi.edu 3e2, 3e4 GPCPv23, OAFlux Antarctic Meteorological Research Center (AMRC) 6b http://amrc.ssec.wisc.edu/data AWS www.ecmwf.int/en/research/climate-reanalysis 6b ERA-Interim /era-interim Pressure, Sea Level HadSLP2r www.metoffice.gov.uk/hadobs/hadslp2/ 2e1 or Near-Surface JRA-55 Atmospheric 6c http://jra.kishou.go.jp/JRA-55/index_en.html Reanalysis www.esrl.noaa.gov/psd/data/gridded/data.ncep 5b NCEP/NCAR Reanalysis .reanalysis.html https://legacy.bas.ac.uk/met/READER/ 6c READER ELSE No public archive 2d6 River Discharge 5d2 EASE-Grid v3 http://nsidc.org/data/nsidc-0611/ Sea Ice Age AMSR2 Daily https://seaice.uni-bremen.de SB6.1 Near-Real-Time DMSP 6e http://nsidc.org/data/nsidc-0081.html SSM/I-SSMIS Daily Polar Sea Ice Gridded Concentration Nimbus-7 SMMR and DMSP http://nsidc.org/data/nsidc-0079.html 6e SSM/I (Bootstrap) Nimbus-7 SMMR and DMST 6e http://nsidc.org/data/NSIDC-0051 S S M /I (N a s aTe am) Nimbus-7 SMMR and DMSP http://nsidc.org/data/seaice_index/ 5c,5d1 Sea Ice Extent SSM/I (Bootstrap) S259 | STATE OF THE CLIMATE IN 2017 AUGUST 2018

280 Specific Dataset or General Variable or Source Section Phenomenon Variable ESA CryoSat-2 5d3 http://data.meereisportal.de/ Sea Ice Freeboard/ https://epic.awi.de/45409/1/CampaignReport Thickness TIFAX 5d3 _TI FA X 2017. pd f www.star.nesdis.noaa.gov/sod/lsa/SeaLevelRise NOAA/NESDIS/STAR 3f /LSA_SLR_timeseries.php http://marine.copernicus.eu/services Ssalto/Duacs Multimission -portfolio/access-to-products/?option=com 3f, SB3.2 Altimeter Products _csw&view=details&product_id=SEALEVEL_GLO Sea Level / Sea _PHY_L4_NRT_OBSERVATIONS_008_046 Surface Height 3f Tide Gauge http://uhslc.soest.hawaii.edu/ https://tidesandcurrents.noaa.gov/ Tide Gauge 3h noaatidepredictions/NOAATidesFacade .jsp?Stationid=8722670 www.ncdc.noaa.gov/data-access/marineocean-data E R S S Tv4 4f2, 4f4 /extended-reconstructed-sea-surface-temperature -ersst-v4 ERSST v5 https://doi.org/10.7289/V5T72FNM 3b HadSST3 www.metoffice.gov.uk/hadobs/hadsst3 2b1, 3b NOAA Optimum http://apdrc.soest.hawaii.edu/dods/public_data Interpolation SST (OISST) 4f Sea Surface /NOA A _ SST/OISST/monthly v2 Temperature 3b, 4b1, NOAA Optimum www.ncei.noaa.gov/data/sea-surface-temperature 4d2, 4f1, Interpolation SST (OISST) -optimum-interpolation/access/ 4f3, 4f6, v2 4h, 6e NOAA Optimum http://nsidc.org/data/g02135 Interpolation SST (OISST) 5c v2 http://sio-argo.ucsd.edu/RG_Climatology.html, Argo 3c,6f ttp://argo.jcommops.org CARICOOS Underwater Sea Subsurface www.aoml.noaa.gov/phod/goos/gliders/index.php 4h Gliders Temperature 4b1, 4c, NCEP Ocean Reanalysis www.cpc.ncep.noaa.gov/products/GODAS/ 4h NOAA Interactive Multi- sensor Snow and Ice www.natice.noaa.gov/ims/index.html 5i Mapping System (Snow Cover Duration) Snow Cover NOAA Snow Chart Data www.snowcover.org 2c2, 5i Record (Snow Cover Extent) NH Snow Water Equivalent http://nsidc.org/data/NSIDC-0668 5i Ensemble Canadian Meteorological 5i Centre Snow Depth Analysis IceBridge Flights ftp://ftp.star.nesdis.noaa.gov/pub/socd/lsa/ 5d3 Snow Depth https://data.npolar.no/dataset/3099ea95-c3cd-4a8b 5d3 in situ snow stakes -af5d-73750e46d791 https://data.npolar.no/dataset/3d72756d-788b-4c49 Magnaprobe 5d3 -b0cc-8a345c091020 2d8 ESA CCl SM www.esa-soilmoisture-cci.org/index.php Soil Moisture S260 | AUGUST 2018

281 General Variable or Specific Dataset or General Variable or Specific Dataset or Section Source Section Source Phenomenon Variable Variable Phenomenon Brazil-Malvina Region www.aoml.noaa.gov/phod/altimetry/cvar/mal ESA CryoSat-2 http://data.meereisportal.de/ 5d3 3g Sea Ice Freeboard/ Confluence Region /BM_anm.php https://epic.awi.de/45409/1/CampaignReport Thickness 5d3 TIFAX Long Term Time Series of _TI FA X 2017. pd f 3g Surface Currents: Agulhas www.aoml.noaa.gov/phod/altimetry/cvar/agu/ www.star.nesdis.noaa.gov/sod/lsa/SeaLevelRise NOAA/NESDIS/STAR 3f Current /LSA_SLR_timeseries.php Long Term Time Series of http://marine.copernicus.eu/services 3g Surface Currents: North www.aoml.noaa.gov/phod/altimetry/cvar/nbc Surface Current -portfolio/access-to-products/?option=com Ssalto/Duacs Multimission Brazil Current 3f, SB3.2 _csw&view=details&product_id=SEALEVEL_GLO Altimeter Products Sea Level / Sea Ocean Surface Current _PHY_L4_NRT_OBSERVATIONS_008_046 Surface Height Analysis - Real time 3g www.oscar.noaa.gov http://uhslc.soest.hawaii.edu/ Tide Gauge 3f (OSCAR) https://tidesandcurrents.noaa.gov/ www.aoml.noaa.gov/phod/altimetry/cvar/yuc 3g Yucatan Current Tide Gauge noaatidepredictions/NOAATidesFacade 3h /transport.php .jsp?Stationid=8722670 Antarctic Meteorological www.ncdc.noaa.gov/data-access/marineocean-data Research Center (AMRC) http://amrc.ssec.wisc.edu/data 6b E R S S Tv4 4f2, 4f4 /extended-reconstructed-sea-surface-temperature AWS -ersst-v4 www.metoffice.gov.uk/hadobs/crutem4 2b1, 5b CRUTEM4 http://ww.cru.uea.ac.uk/cru/data/temperature https://doi.org/10.7289/V5T72FNM ERSST v5 3b Danish Meteorological http://research.dmi.dk/research/research-topics www.metoffice.gov.uk/hadobs/hadsst3 2b1, 3b HadSST3 5e6 /climate/ Institute NOAA Optimum http://apdrc.soest.hawaii.edu/dods/public_data Interpolation SST (OISST) 4f www.ecmwf.int/en/forecasts/datasets/archive Sea Surface /NOA A _ SST/OISST/monthly ERA-Interim 2b1, 6b v2 -datasets/reanalysis-datasets/era-interim Temperature 3b, 4b1, 2b3 www.climdex.org/datasets.html GHCNDEX NOAA Optimum www.ncei.noaa.gov/data/sea-surface-temperature 4d2, 4f1, HadCRUT4 Global Interpolation SST (OISST) 2b1 www.metoffice.gov.uk/hadobs/hadcrut4/ -optimum-interpolation/access/ 4f3, 4f6, Temperature v2 Temperature, [Near] 4h, 6e https://ds.data.jma.go.jp/tcc/tcc/products/gwp/temp Surface JMA Global Temperature 2b1 NOAA Optimum /map/download.html Interpolation SST (OISST) http://nsidc.org/data/g02135 5c JRA-55 Atmospheric v2 2b1 http://jra.kishou.go.jp/JRA-55/index_en.html Reanalysis http://sio-argo.ucsd.edu/RG_Climatology.html, Argo 3c,6f MERRA-2 http://gmao.gsfc.nasa.gov/reanalysis/MERRA-2/ 2b1 ttp://argo.jcommops.org NASA/GISS Global CARICOOS Underwater Sea Subsurface 2b1, 2b2 https://data.giss.nasa.gov/gistemp/ 4h www.aoml.noaa.gov/phod/goos/gliders/index.php Temperature Gliders Temperature www.esrl.noaa.gov/psd/data/gridded/data.ncep 4b1, 4c, 5b NCEP/NCAR Reanalysis NCEP Ocean Reanalysis www.cpc.ncep.noaa.gov/products/GODAS/ .reanalysis.html 4h www.ncdc.noaa.gov/data-access/marineocean-data NOAA Interactive Multi- 2b1 NOAAGlobalTemp /noaa-global-surface-temperature-noaaglobaltemp sensor Snow and Ice 5i www.natice.noaa.gov/ims/index.html 6c READER https://legacy.bas.ac.uk/met/READER/ Mapping System (Snow Cover Duration) Snow Cover NOAA Snow Chart Data 2c2, 5i www.snowcover.org Record (Snow Cover Extent) NH Snow Water Equivalent http://nsidc.org/data/NSIDC-0668 5i Ensemble Canadian Meteorological 5i Centre Snow Depth Analysis IceBridge Flights 5d3 ftp://ftp.star.nesdis.noaa.gov/pub/socd/lsa/ Snow Depth https://data.npolar.no/dataset/3099ea95-c3cd-4a8b 5d3 in situ snow stakes -af5d-73750e46d791 https://data.npolar.no/dataset/3d72756d-788b-4c49 5d3 Magnaprobe -b0cc-8a345c091020 2d8 www.esa-soilmoisture-cci.org/index.php ESA CCl SM Soil Moisture S261 | STATE OF THE CLIMATE IN 2017 AUGUST 2018

282 General Variable or Specific Dataset or Source Section Variable Phenomenon CMIP5 https://esgf-node.llnl.gov/search/cmip5/ 2b5 2b4, 2b5, www.ecmwf.int/en/research/climate-reanalysis ERA-Interim /era-interim 6b JRA-55 Atmospheric http://jra.kishou.go.jp/JRA-55/index_en.html 2b4, 2b5 Reanalysis 2b4, 2b5, http://gmao.gsfc.nasa.gov/reanalysis/MERRA-2/ MERRA-2 2g5 RAOBCORE, RICH www.univie.ac.at/theoret-met/research/raobcore 2b4, 2b5 www.ncdc.noaa.gov/data-access/weather-balloon Temperature, Upper R ATPAC A 2 /radiosonde-atmospheric-temperature-products 2b4, 2b5 Atmosphere -accessing-climate/ratpac-a RSS 2b4, 2b5 www.remss.com www.esrl.noaa.gov/psd/data/gridded/data.ncep NCEP/NCAR Reanalysis 5b .reanalysis.html NOAA/NESDIS/STAR www.star.nesdis.noaa.gov/smcd/emb/mscat/ 2b5 Ozonesonde 6g www.esrl.noaa.gov/gmd/dv/spo_oz SSU-3 2b5 UAH MSU http://vortex.nsstc.uah.edu/public/msu 2b4, 2b5 University of New South http://web.science.unsw.edu.au/~stevensherwood 2b4, 2b5 Wales /radproj/index.html https://ceres-tool.larc.nasa.gov/ord-tool/jsp 2f1 CERES FLASHFlux /EBAF4Selection.jsp TOA Earth Radiation Budget https://ceres-tool.larc.nasa.gov/ord-tool/jsp 2f1 CERES EBAF Ed4.0 /EBAF4Selection.jsp SORCE/TIM http://science.nasa.gov/missions/sorce/ 2f1 Total Solar Irradiance Atmospheric Greenhouse www.esrl.noaa.gov/gmd/aggi 2g1 Gas Index (AGGI) Carbon Dioxide (CO2) www.esrl.noaa.gov/gmd/dv/iadv 2g1 ftp://aftp.cmdl.noaa.gov/products/trends/co2/co2 Carbon Dioxide (CO2) 3j2 _mm_mlo.txt 2g7 Carbon Monoxide (CO) https://atmosphere.copernicus.eu/ Chlorine Monoxide (ClO) - http://mls.jpl.nasa.gov/products/clo_product.php 6g Aura MLS Halocarbons (CFCs, HFCs, www.esrl.noaa.gov/gmd/hats/data.html 2g1 Trace Gases HCFCs) Hydrogen Chloride (HCl) - http://disc.sci.gsfc.nasa.gov/datacollection/ML2HCL 2g1 _V004.html Aura MLS Methane www.esrl.noaa.gov/gmd/dv/iadv 2g1 Nitrous Oxide www.esrl.noaa.gov/gmd/hats/combined/N2O.html 2g1 Ozone-Depleting Gas Index www.esrl.noaa.gov/gmd/odgi 2g2 (ODGI) Perfluorocarbons http://agage.eas.gatech.edu 2g1 Sulfur Hexafluoride www.esrl.noaa.gov/gmd/hats/combined/SF6.html 2g1 S262 | AUGUST 2018

283 General Variable or Specific Dataset or Source Section Phenomenon Variable H U RDAT 2 www.aoml.noaa.gov/hrd/hurdat/Data_Storm.html 4f2 International Best Track 4f1, 4f3, www.ncdc.noaa.gov/ibtracs/ Archive for Climate 4f6, 4f7 Stewardship (IBTrACS) JTWC Best-track Dataset www.usno.navy.mil/NOOC/nmfc-ph/RSS/jtwc 4f4, 4f5 Tropical Cyclone (2011 preliminary) /best_tracks Data www.jma.go.jp/jma/jma-eng/jma-center/rsmc-hp RSMC-Tokyo, JMA best- 4f4 track data -pub-eg/besttrack.html Southwest Pacific Enhanced Archive of Tropical Cyclones http://apdrc.soest.hawaii.edu/projects/speartc 4f8 (S PE A rTC) ftp://exp-studies.tor.ec.gc.ca/pub/uvdata Canadian sites 5j /Preliminary/MSC http://uv.biospherical.com/Version2/data.asp Greenland site 5j UV Radation Data http://litdb.fmi.fi/soundingst_uvradiation.php, Finnish sites 5j http://uv.fmi.fi/uvdb/ Norwegian sites https://github.com/uvnrpa/Minute_Data 5j Frost Point Hygrometer 2g5 ftp://aftp.cmdl.noaa.gov/data/ozwv/WaterVapor Data (Boulder, Hilo, Lauder) Frost Point Hygrometer Water Vapor, http://physics.valpo.edu/ozone/ticosonde.html 2g5 Data (San Jose) Stratosphere NASA Aura Microwave Limb https://mls.jpl.nasa.gov/products/h2o_product.php 2g5 Sounder http://cdaac-www.cosmic.ucar.edu/cdaac/products COSMIC GPS-RO 2d2 .html www.ecmwf.int/en/research/climate-reanalysis 2d2 ERA-Interim /era-interim GNSS Ground-Based Total Water Vapor, Total https://rda.ucar.edu/datasets/ds721.1/ 2d2 Column Water Vapor Column JRA-55 Atmospheric 2d2 http://jra.kishou.go.jp/JRA-55/index_en.html Reanalysis 2d2 MERRA-2 http://gmao.gsfc.nasa.gov/reanalysis/MERRA-2/ RSS SSM/I -AMSR-E Ocean www.remss.com 2d2 Total Column Water Vapor Australian (McVicar) http://doi.org/10.4225/08/56A85491DDED2 2e2 www.ecmwf.int/en/research/climate-reanalysis 2e2 ERA-Interim /era-interim HadISD2 www.metoffice.gov.uk/hadobs/hadisd/ 2e2 Wind, [Near] Surface JRA-55 Atmospheric http://jra.kishou.go.jp/JRA-55/index_en.html 2e2, 4h Reanalysis 2e2 MERRA-2 http://gmao.gsfc.nasa.gov/reanalysis/MERRA-2/ RSS SSM/I Ocean Winds 2e2 www.remss.com/measurements/wind S263 | STATE OF THE CLIMATE IN 2017 AUGUST 2018

284 Specific Dataset or General Variable or Source Section Phenomenon Variable www.ecmwf.int/en/forecasts/datasets/archive 2e3 CERA-20C -datasets/reanalysis-datasets/cera-20c www.ncdc.noaa.gov/data-access/model-data 4c, 4f3, Climate Forecast System /model-datasets/climate-forecast-system-version2 4f6 Reanalysis -cfsv2 www.ecmwf.int/en/research/climate-reanalysis 2e3, 6b, ERA-Interim /era-interim 4e Wind, Upper Atmosphere 2e3 https://doi.pangaea.de/10.1594/PANGAEA.823617 GRASP JRA-55 Atmospheric http://jra.kishou.go.jp/JRA-55/index_en.html 2e3, 4h Reanalysis MERRA-2 2e3 http://gmao.gsfc.nasa.gov/reanalysis/MERRA-2/ 4b2, 4c, www.esrl.noaa.gov/psd/data/gridded/data.ncep 4f2, 4f4, NCEP/NCAR Reanalysis .reanalysis.html 5g S264 | AUGUST 2018

285 ACKNOWLEDGMENTS In addition to the specific acknowledgments below Section 2b3 (Temperature Extremes): Appendix 1: Relevant Datas - SEPK is supported by ARC grant number • readers are directed to FT170100106. ets and Sources for information on specific data and products availability and relevant climate variables Section 2b5 (Lower Stratospheric Temperature): or phenomena. Work performed by Stephen Po-Chedley at LLNL • The editors thank the AMS Journals editorial staff, was performed under the auspices of the U.S. in particular Andrea Herbst, and the NCEI graphics Department of Energy under Contract DE-AC52- team for facilitating the construction of the report, 07NA27344 and under LDRD 18-ERD-054. and executing the countless number of technical edits Section 2d6 (River Discharge): needed. We also express our gratitude to Dr. Rick • Hyungjun Kim was supported by the Japan So - Rosen, who again served as the AMS special editor for ciety for the Promotion of Science KAKENHI this report. His handling of the reviews was thought - (16H06291) for this contribution. ful and timely. We also thank NOAA’s Rotational Assignments Program and the Raleigh, NC National Section 2d7 (Terrestrial Water Storage): Weather Service Forecast Office for bringing Gail • Hartfield to the document this year; her fresh set of This research was supported by grants from NASA’s GRACE and GRACE-FO Science Team. eyes and professional perspective were quite valuable. GRACE land data processing is supported by the Chapter 2 NASA MEaSUREs Program. : Editors The chapter editors thank David Parker, John • Section 2d8 (Soil Moisture): • - The ESA CCI SM datasets and the authors were Kennedy, and Elizabeth Good for providing de supported by ESA’s Climate Change Initiative for tails and comprehensive internal reviews of this chapter. The editors also thank Paul Berrisford Soil Moisture (Contract No. 4000104814/11/I- NB and 4000112226/14/I-NB) and the European (ECMWF), Mike Bosilovich (NASA), and Shinya Kobayashi (JMA) for timely provision of reanalysis - Union’s FP7 EartH2Observe “Global Earth Obser vation for Integrated Water Resource Assessment” data used herein. Robert Dunn, Kate Willett, Colin Morice, and Rob Allan were supported by the Joint project (grant agreement number 331 603608). We would also like to thank support from the UK BEIS/Defra Met Office Hadley Centre Climate Programme (GA01101). Copernicus Climate Change Service implemented by ECMWF. Section 2b2 (Lake Surface Temperature): Section 2d9 (Soil Moisture): RIW was funded by EUSTACE (EU Surface Tem - • perature for All Corners of Earth) which received Tim Osborn received funding from UK NERC • funding from the European Union’s Horizon - (NE/P006809/1). Jonathan Barichivich re ceived funding from (CR)2 Chile (CONICYT 2020 Programme for Research and Innovation, - /FONDAP/15110009). Ian Harris received fund under Grant Agreement no 640171. Satellite data ing from UK National Centre for Atmospheric processed under UK Natural Environment Re - search Council project Globolakes (grant reference Science (NCAS). NE/J023345/2). This work was partly supported by Russian Ministry of Education and Science Section 2g (Atmospheric Composition): Historical ozone observations from the 1950s at (project #6.1387.2017), by the grants of Russian • Mauna Loa Observatory and from the 1960s at Science Foundation МА - 17-14-01063 and of South Pole were digitized by Samuel J. Oltmans, the Lake Baikal Foundation for support of ap - NOAA, retired. The authors would like to ac plied ecological studies (https://baikalfoundation - .ru/project/tochka-1/). SGS and SW funded by the - knowledge a NASA Upper Atmosphere Composi Tahoe Regional Planning Agency. Lake surface tion grant for support that helps sustain the frost point hygrometer soundings at Boulder, Lauder water temperatures in Poland were provided by the Institute of Meteorology and Water Manage - and Hilo. Henry Selkirk and Holger Vömel would ment in Warsaw. like to acknowledge support from the NASA | AUGUST 2018 STATE OF THE CLIMATE IN 2016 S265

286 Upper Atmosphere Composition Observations Chapter 3 program. The CAMS interim Reanalysis has been • The editor thanks NOAA/PMEL’s Sandra Bigley - produced by the Copernicus Atmosphere Moni for her invaluable work in assembling and copy editing Chapter 3. toring Services (CAMS). The European Centre for Medium-Range Weather Forecast (ECMWF) - runs CAMS on behalf of the European Commis Sidebar 3.1 (Coral bleaching): NOAA Coral Reef Watch work is supported pri • sions. Antje Inness and Johannes Flemming are - marily by the NOAA Coral Reef Conservation funded by CAMS. Melanie Coldewey-Egbers, Program and the NOAA National Environmental Daan Hubert, Diego Loyola, Victoria Sofieva, and Mark Weber are grateful to ESA’s Climate Change Satellite, Data, and Information Service’s Center Initiative Ozone project and to the EU Copernicus for Satellite Applications and Research. The scien - tific results and conclusions, as well as any views Climate Change Service 312a Lot4 Ozone project or opinions expressed herein, are those of the for supporting the generation and extension of the author(s) and do not necessarily ref lect the views GTO-ECV total ozone and SAGE-CCI-OMPS data of NOAA or the U.S. Department of Commerce. records. Stacey M. Frith is supported by the NASA Long Term Measurement of Ozone program WBS Section 3e (Global ocean heat, freshwater, and mo - 479717. Lucien Froidevaux’s contribution, with the mentum f luxes): assistance of Ryan Fuller, was performed at the • Jet Propulsion Laboratory, California Institute The GPCP SG combined precipitation data were developed and computed at the NASA/Goddard of Technology, under contract with NASA. Daan Hubert acknowledges the partial support by the - Space Flight Center’s Mesoscale Atmospheric Pro cesses Laboratory–Atmospheres as a contribution EU/ERC Horizon 2020 project GAIA-CLIM. to the GEWEX Global Precipitation Climatology Section 2h2 (Vegetation Dynamics): Project. • The authors would like to thank the SeaWiFS Section 3h (AMOC and AMHT): Project (Code 970.2) and the Goddard Earth Data from the RAPID-WATCH MOC monitoring Sciences Data and Information Services Center/ • Distributed Active Archive Center (Code 902) at project are funded by the Natural Environment the Goddard Space Flight Center, Greenbelt, MD Research Council. • 20771, for the production and distribution of these MOVE contributions were made under award data, respectively. These activities are sponsored NA15OAR4320071 from the Climate Observa - by NASA’s Earth Science Enterprise. tions Division, National Oceanic and Atmospheric Administration, U.S. Department of Commerce. Section 2h3 (Biomass Burning): Previously, MOVE was funded by the German The GFASv1.4 dataset was provided the by GFAS- Bundesministerium für Bildung und Forschung • CLIM project, which is funded by the German (Grants 03F0246A and 03F0377B). Bundesministerium für Wirtschaft und Energie • The Florida Current cable and section data are funded by the DOC-NOAA Climate Program Of - (BMWi FKZ 50EE1543), and the Copernicus fice - Ocean Observing and Monitoring Division. CAMS_44 project. The XBT data are funded by the NOAA Office of • Sidebar 2SB.1 (Extreme Precipitation): Climate Observations. - • • NCAR is sponsored by the National Science Argo data are collected and made freely avail able by the International Argo Program and the Foundation. national programs that contribute to it; the Argo Sidebar 2SB.2 (Phenology): Program is part of the Global Ocean Observing • The authors thank reviewers, T. Rutishäuser System. (Swiss Academies of Arts and Science, Berne, Switzerland) and T. Sparks (Coventry University, Coventry, UK), for their internal review. | AUGUST 2018 S266

287 Chapter 4 - George Washington University), Natural Resourc The editors would like extend their thanks and ap es Canada and the Norwegian Meteorological In - - stitute. Support was also provided by the Russian preciation to Joe Pelissier, the retired Meteorologist- Science Foundation (project RNF 16-17-00102) in-Charge of the NOAA/NWS Weather Forecast and by the government of the Russian Federation. Office in Greenville-Spartanburg, SC, for his inputs regarding the passing of his good friend and colleague Section 5j (Ozone and UV radiation): Charlie Neumann. Carl Schreck was supported by • NOAA through the Cooperative Institute for Climate G. Bernhard and coauthors acknowledge the U.S. and Satellites - North Carolina under Cooperative National Science Foundation for supporting UV Agreement NA14NES432003. measurements at Barrow and Summit, a Research Council of Norway Centres of Excellence award (Project 223268/F50) to the Norwegian Radiation Chapter 5 Protection Authority, and the Academy of Fin - Editor Team: land for supporting UV measurements through • Martin Jeffries thanks the Office of Naval the FARPOCC, SAARA, and ILMA pilot proj Research (Code 32) for supporting his role as - - Associate Editor of the State of the Climate Re ects. We also acknowledge Tove Svendby from port and co-editor of the Arctic section. Jackie the Norwegian Institute for Air Research and Arne Dahlback from the University of Oslo for Richter-Menge acknowledges that this publication overseeing UV measurements at Oslo, Andøya is the result in part of research sponsored by the Cooperative Institute for Alaska Research with Ålesund, and thank Juha M. Karhu, and Ny- Tomi Karppinen and Markku Ahponen from the funds from the National Oceanic and Atmospheric Finnish Meteorological Institute for operating the Administration under cooperative agreement Brewer UV spectroradiometer at Sodankylä. NA13OAR4320056 with the University of Alaska. Emily Osborne acknowledges the NOAA Arctic Research Program for supporting her role as co- Chapter 6 The editors wish to acknowledge and thank the au editor of the State of the Climate Report Arctic - thors for their timely contributions, with additional section and coauthor of the paleoclimate sidebar. special thanks to the internal and external reviewers and Document Editors for their thoughtful and con Section 5e (Greenland): - • Marco Tedesco would like to acknowledge the structive comments. The editors also wish to thank NASA Cryosphere Program (NNX17AH04G, Dr. Sam Batzli of the Space Science and Engineering Center at the University of Wisconsin-Madison for NNX16AH38G), the NASA IDS program (NNX14AD98G) and the Office of Polar Programs his generation of the map in Figure 6.1. at the National Science Foundation (OPP 1643187, Ted Scambos was supported under NASA grant • PLR-1603331). PROMICE stations are funded by the Danish Energy Agency. KAN stations are NNX16AN60G and NSF PLR 1565576, and thanks funded by SKB. IS acknowledges funding by the the National Snow and Ice Data Center. Helmholtz Climate Initiative REKLIM (Regional • Sharon Stammerjohn was supported under NSF PLR 1440435; she also thanks the Institute of Climate Change), a joint research project of the Arctic and Alpine Research and the National Helmholtz Association of German Research Cen - tres (HGF) and the German Research Foundation Snow and Ice Data Center, both at the University through grant SA 1734/4-1. The Netherlands Polar of Colorado Boulder, for institutional and data Programme (NPP) of the NWO (Netherlands support. • Institute for Scientific Research) is acknowledged Linda Keller and Matthew Lazzara were supported for the K-transect support. - by the Automatic Weather Station Program, Na tional Science Foundation, PLR-1543305. Section 5g (Terrestrial permafrost): Sebastiaan Swart was supported by a Wallenberg • • Academy Fellowship (WAF 2015.0186). Vladimir Romanovsky and coauthors acknowl - - edge the support of the State of Alaska, the Na tional Science Foundation (grants PLR-0856864 and PLR-1304271 to the University of Alaska Fairbanks; PLR-1002119 and PLR-1304555 to the | AUGUST 2018 STATE OF THE CLIMATE IN 2016 S267

288 • Chapter 7 The work of Rob Massom, Phil Reid, Jan Lieser, and Steve Rintoul was supported by the Austra The editors acknowledge and thank the numerous - lian Government’s Cooperative Research Centre national meteorological and hydrological services program through the Antarctic Climate & Eco - for collecting and providing data for this report. systems CRC, and contributes to AAS Project Data centers like NCEP/NCAR, ECMWF-ERA, and 4116. PR was also supported through the Bureau CHIRPS are also acknowledged for making their data of Meteorology. freely available. Acknowledgments for specific regions are as follows: Australian Research Council’s Special Research • Initiative for Antarctic Gateway Partnership (Project ID SR140300001). Africa: - The editor expresses particular thanks to the follow • Jean-Baptiste Sallée was supported by the Euro - ing national meteorology and hydrology services for pean Research Council (ERC) under the European their contributions to our report: Union›s Horizon 2020 research and innovation - North Africa: National Meteorological and Hydro • program (Grant Agreement no 637770); Mike Meredith received funding from the Natural logical Services of Morocco and Egypt West Africa: National Meteorological and Hydro Environment Research Council via award NE/ • - logical Services of The Gambia and Nigeria N018095/1. East Africa: National Meteorological Agency of • • Work at the Jet Propulsion Laboratory, California Institute of Technology, was done under contract Ethiopia with the National Aeronautics and Space Adminis - • South Africa: National Meteorological Services tration (NASA). Work at NASA GSFC is supported of South Africa Indian Ocean: Meteorological and Hydrological through NASA’s Atmospheric Composition Mod • - eling and Analysis Program. Services of Seychelles, Comoros, Mayotte, La Reunion, Mauritius, and Rodriguez. Europe: Valuable climate information was provided by Na - tional Meteorological and Hydrological Services (NMHSs) of the WMO RA VI Region, either by direct submission to the authors or via the web. Asia: The Editor wishes to thank all the authors for their timely contributions, and the internal and external reviewers and document editors for their thoughtful and constructive comments. The following grants (China National Key R&D Program 2017YFA0603802 and 2015CB453201, NSFC grant 41630423, NSF grant AGS-1643297) are acknowledged. | AUGUST 2018 S268

289 ACRONYMS AND ABBREVIATIONS MLO: auna Loa Observatory (Hawaii, M B lended Analysis of Surface Salinity BASS: US) (NOA A) eteorological Research Institute/ M MRI/JMA: CCI: C limate Change Initiative Japan Meteorological Agency C louds and the Earth’s Radiant CERES: N NASA: ational Aeronautics and Space Energy Systems Administration (US) CFSR: C limate Forecast System Reanalysis N NCAR: ational Center for Atmospheric (NCEP) Research (US) CIIFEN: entro Internacional para la C N ational Centers for Environmental NCEI: Investigación del Fenómeno El Niño Information (NOAA) (Ecuador) N NCEP: ational Centers for Environmental CMAP: PC Merged Analysis of C Prediction (NOAA) Precipitation NOAA: ational Oceanic and Atmospheric N C CMEMS: opernicus Marine and Administration (US) Environment Monitoring Service NSIDC: ational Snow and Ice Data Center N CPC: C limate Prediction Center (NOAA) (US) CSIRO/ACE CRC/IMAS-UTAS: OLR: ou tgoing longwave radiation (Australia) Commonwealth PMEL/JPL/JIMAR: (US) Scientific and Industrial Research P acific Marine Environmental Organisation Laboratory/Jet Propulsion ntarctic Climate & Ecosystems A Laboratory/ Cooperative Research Centre J oint Institute for Marine and I nstitute for Marine and Antarctic Atmospheric Research Studies - University of Tasmania (International, RAPID-MOC/MOCHA/WBTS: DWD: D eutscher Wetterdienst U K-led) E ECV: ssential Climate Variable APID Climate Change Programme- R ESA: E uropean Space Agency Meridional Overturning Circulation arth System Research Laboratory E ESRL: M eridional Overturning Circulation (NOA A) and Heatf lux Array ast Longwave And Shortwave F FLASHFlux: W estern Boundary Time Series Radiative Fluxes RSW: r ef lected shortwave lobal Ocean Ship-based G GO-SHIP: t TOA: op of atmosphere Hydrographic Investigations TRMM: ropical Rainfall Measuring Mission T Program TSI: otal solar irradiance t GODAS: G lobal Ocean Data Assimilation orld Ocean Atlas W WOA: System G GPCP: lobal Precipitation Climatology Project - Additional acronyms and abbrevia ravity Recovery and Climate G GRACE: tions can be found at this AMS website Experiment www.ametsoc.org/ams/index.cfm/publications nstituto Nacional de Meteorologia I /authors/journal-and-bams-authors/and-manuscript INMET: (Brazil) -components/list-of-acronyms-and-abbreviations/ | AUGUST 2018 STATE OF THE CLIMATE IN 2016 S269

290 El Niño Southern ——, J. A. Lindesay, and D. E. Parker, 1996: REFERENCES CSIRO Publications, Oscillation and Climatic Variability. - Abernethy, R., J. Garforth, D. Hemming, M. Kendon, M. Mc 405 pp. Carthy, and T. Sparks, 2017: State of the UK Climate 2016: - Amador, J. A., 1998: A climatic feature of the tropical Ameri Phenology supplement. 39 pp., www.metoffice.gov.uk cas: The trade wind easterly jet. Top. Meteor. Oceanog r. , 5 , /binaries/content/assets/mohippo/pdf/uk-climate/state 91–102, http://doi.org/10.1196/annals.1446.012. -of-the-uk-climate/mo-phenology-supplement-v4.pdf. ——, E.J. Alfaro, H.G. Hidalgo, and B. Calder Ón, 2011: Abraham, J. P., and Coauthors, 2013: A review of global ocean State of the Climate in 2010”]. Central America [in “ temperature observations: Implications for ocean heat Bull. Amer. Meteor. Soc. (6), S182-S183, http://doi 92 , , , Rev. Geophys. content estimates and climate change. 51 .org/10.1175/1520-0477-92.6.s1. 450–483, http://doi.org/10.1002/rog.20022. ——, H. G. Hidalgo, E. J. Alfaro, B. Calderón, N. P. Mora, and Ackerman, S. A., R. E. Holz, R. Frey, E. W. Eloranta, B. C. - I. L. Rivera, 2012: Central America [in “State of the Cli 93 (7), S169–S170, Maddux, and M. McGill, 2008: Cloud detection with mate in 2011”]. , Bull. Amer. Meteor. Soc. http://doi.org/10.1175/2012BAMSStateoftheClimate.1. MODIS. Part II: Validation. J. Atmos. Oceanic Technol. , An, R., and Coauthors, 2016: Validation of the ESA CCI soil , 1073–1086, http://doi.org/10.1175/2007jtecha1053.1. 25 moisture product in China. Int. J. Appl. Earth Obs. Geoinf. , Acosta Navarro, J. C., and Coauthors, 2016: Amplifica - , 28–36, http://doi.org/10.1016/j.jag.2015.09.009. 48 tion of Arctic warming by past air pollution reductions Anagnostou, E., and Coauthors, 2016: Changing atmospheric , in Europe. Nat. Geosci. , 277–281, http://doi.org/10.1038 9 concentration was the primary driver of early Ceno - CO /ngeo2673. 2 , 380–384, http://doi.org/10.1038 , Nature zoic climate. 533 - Adler, R. F., and Coauthors, 2003: The version-2 Global Pre /nat u re174 23. - cipitation Climatology Project (GPCP) monthly precipita Anderson, M. C., and Coauthors, 2011: Mapping daily J. Hydrometeor. tion analysis (1979–present). , 4 , 1147–1167, evapotranspiration at field to continental scales using ht t p://doi.org /10.1175/1525-7541(20 03)0 04<1147:t vg pcp geostationary and polar orbiting satellite imagery. - Hy >2.0.co;2. , drol. Earth Syst. Sci. 15 , 223–239, http://doi.org/10.5194 Agence France-Presse, 2016: Thailand closes dive sites /hess-15-223-2011. The Guardian , May 26. over coral bleaching crisis. Ansorge, I. J., and Coauthors, 2014: Basin-wide www.theguardian.com/environment/2016/may/26 oceanographic array bridges the South Atlan - /thailand-closes-dive-sites-over-coral-bleaching-crisis. 95 Eos, Trans. Amer. Geophys. Union , 53 –5 4 , tic. , AICC Predictive Services, 2015: Alaska Fire Numbers http://doi.org/10.1002/2014EO060001. 2015. 38 pp., https://fire.ak.blm.gov/content/aicc/stats Aono, Y., and K. Kazui, 2008: Phenological data series /Alaska%20Fire%20Numbers.pdf. of cherry tree f lowering in Kyoto, Japan, and its ap - Aiyyer, A., and J. Molinari, 2008: MJO and tropical cyclo - plication to reconstruction of springtime temperatures genesis in the Gulf of Mexico and eastern Pacific: Case , 9 05 –914, 28 , Int. J. Climatol. since the 9th century. , 65 study and idealized numerical modeling. J. Atmos. Sci. , http://doi.org/10.1002/joc.1594. 2691–2704, http://doi.org/10.1175/2007jas2348.1. Aquila, V., W. H. Swartz, D. W. Waugh, P. R. Colarco, S. Alexander, L. V., and Coauthors, 2006: Global observed Pawson, L. M. Polvani, and R. S. Stolarski, 2016: Isolating changes in daily climate extremes of temperature the roles of different forcing agents in global stratospheric and precipitation. , D 0510 9, 111 , J. Geophys. Res. temperature changes using model integrations with incre - http://doi.org/10.1029/2005JD006290. , J. Geophys. Res. Atmos. mentally added single forcings. Allan, R. J., and R. D. D’Arrigo, 1999: ‘Persistent’ , 8067–8082, http://doi.org/10.1002/2015JD023841. 121 ENSO sequences: How unusual was the 1990- Armstrong, R., K. Knowles, M. J. Brodzik, and M. A. The Holocene , 101–118, http://doi 9 , 1995 El Niño? Hardman, 1994, updated 2016: DMSP SSM/I-SSMIS .org /10.1191/095968399669125102. Pathfinder Daily EASE-Grid Brightness Temperatures. ——, and T. Ansell, 2006: A new globally complete monthly Version 2. National Snow and Ice Data Center, http://doi - historical gridded mean sea level pressure dataset (Had .org/10.5067/3EX2U1DV3434. Accessed February 2018. 19 J. Climate , 5816–5842, http://doi SLP2): 1850 –2004. , - Arnone, E., D. Pumo, F. Viola, L. V. Noto, and G. La Log .org/10.1175/jcli3937.1. Hydrol. gia, 2013: Rainfall statistics changes in Sicily. ——, and C. K. Folland, 2017: Mean sea level pressure and Earth Syst. Sci. , 2449–2458, http://doi.org/10.5194 17 , related modes of variability [in “State of the Climate /hess-17-2449-2013. 98 , Bull. Amer. Meteor. Soc. in 2016”]. (8), S35–S37, http://doi.org/10.1175/2017BAMSStateoftheClimate.1. | AUGUST 2018 S270

291 Ashok, K., Z. Guan, and T. Yamagata, 2003: A look at the rela - Ball, W. T., and Coauthors, 2018: Evidence for a continuous tionship between the ENSO and the Indian Ocean dipole. decline in lower stratospheric ozone offsetting ozone 81 , J. Meteor. Soc. Japan , 41–56, http://doi.org/10.2151 18 Atmos. Chem. Phys. layer recovery. , 1379–1394, , /jmsj.81.41. ht t p://doi.org /10.5194/acp-18-1379-2018. Annual climate Australian Bureau of Meteorology, 2017: Banzon, V. F., and R. W. Reynolds, 2013: Use of WindSat statement 2017. Bureau of Meteorology Australia. - to extend a microwave-based daily optimum interpola w w w.bom.gov.au/climate/current/a nnua l/aus/. J. Climate , tion sea surface temperature time series. 26 , Azorin-Molina, C., and Coauthors, 2014: Homogenization 2557–2562, http://doi.org/10.1175/jcli-d-12-00628.1. and assessment of observed near-surface wind speed Barbero, R., H. J. Fowler, G. Lenderink, and S. Blenkinsop, J. Climate trends over Spain and Portugal, 1961–2011. , 2017: Is the intensification of precipitation extremes with global warming better detected at hourly than , 3692–3712, http://doi.org/10.1175/jcli-d-13-00652.1. 27 44 daily resolutions? , 974 –983, , Geophys. Res. Lett. ——, J.-A. Guijarro, T. R. McVicar, S. M. Vicente-Serrano, http://doi.org/10.1002/2016GL071917. D. Chen, S. Jerez, and F. Espírito-Santo, 2016: Trends of Baringer, M. O., and Coauthors, 2017: Meridional overturn - daily peak wind gusts in Spain and Portugal, 1961–2014. ing and oceanic heat transport circulation observations J. Geophys. Res. Atmos. 121 , 1059–1078, http://doi , in the North Atlantic Ocean [in “State of the Climate .org/10.1002/2015JD024485. in 2016”]. Bull. Amer. Meteor. Soc. , 98 (8), S84–S87, - ——, S. M. Vicente-Serrano, T. R. McVicar, J. Revuelto, S. Jer http://doi.org/10.1175/2017BAMSStateoftheClimate.1. ez, and J.-I. López-Moreno, 2017a: Assessing the impact of Barnes, E. A., A. M. Fiore, and L. W. Horowitz, 2016: Detec - measurement time interval when calculating wind speed tion of trends in surface ozone in the presence of climate means and trends under the stilling phenomenon. Int. J. , 121 , 6112– 6129, variability. J. Geophys. Res. Atmos. , 480–492, http://doi.org/10.1002/joc.4720. 37 Climatol. , http://doi.org/10.1002/2015JD024397. ——, R. J. H. Dunn, C. A. Mears, P. Berrisford, and T. R. Barrett, K., T. Loboda, A. D. McGuire, H. Genet, E. Hoy, McVicar, 2017b: Surface winds [in “State of the Climate and E. Kasischke, 2016: Static and dynamic controls in 2016”]. (8), S37–S39, 98 , Bull. Amer. Meteor. Soc. on fire activity at moderate spatial and temporal scales http://doi.org/10.1175/2017BAMSStateoftheClimate.1. , e 01572 , 7 , Ecosphere in the Alaskan boreal forest. ——, and Coauthors, 2017c: Wind speed variability over the http://doi.org/10.1002/ecs2.1572. Canary Islands, 1948–2014: Focusing on trend differences Bauer-Marschallinger, B., W. A. Dorigo, W. Wagner, and at the land–ocean interface and below–above the trade- A. I. J. M. van Dijk, 2013: How oceanic oscillation drives , http://doi.org/10.1007 wind inversion layer. Climate Dyn. soil moisture variations over mainland Australia: An /s00382-017-3861-0. J. Climate , analysis of 32 years of satellite observations. ——, S. Rehman, J. A. Guijarro, T. R. McVicar, L. Minola, D. 26 , 10,159–10,173, http://doi.org/10.1175/jcli-d-13-00149.1. Chen, and S. M. Vicente-Serrano, 2018a: Recent trends Baxter, S., S. Weaver, J. Gottschalck, and Y. Xue, 2014: Pentad in wind speed across Saudi Arabia, 1978–2013: A break evolution of wintertime impacts of the Madden–Julian 38 (S1), e966 –e984, , Int. J. Climatol. in the stilling. oscillation over the contiguous United States. J. Climate , http://doi.org/10.1002/joc.5423. 27 , 7356–7367, http://doi.org/10.1175/jcli-d-14-00105.1. ——, J. Asin, T. R. McVicar, L. Minola, J. I. Lopez-Moreno, ——, C. J. Schreck, and G. D. Bell, 2017: Tropical intraseasonal S. M. Vicente-Serrano, and D. Chen, 2018b: Evaluating activity [in “State of the Climate in 2016”]. Bull. Amer. anemometer drift: A statistical approach to correct biases Meteor. Soc. , 98 (8), S98–S101, http://doi.org/10.1175 in wind speed measurement. Atmos. Res. , 203 , 175–188, /2017BAMSStateoftheClimate.1. http://doi.org/10.1016/j.atmosres.2017.12.010. BCRP (Banco Central de Reserva del Perú), 2017: Reporte de Bahr, K. D., P. L. Jokiel, and K. S. Rodgers, 2015: The 2014 Inf lación marzo 2017: Panorama actual y proyecciones ʻ ohe coral bleaching and freshwater f lood events in Kāne macroeconómicas 2017-2018 [Inf lation Report March Bay, Hawa i ʻ PeerJ , 3 , e1136, http://doi.org/10.7717 i. 2017: Current Panorama and Macroeconomic Projections /peer j.1136. 2017-2018]. 116 pp., www.bcrp.gob.pe/docs/Publicaciones Bakker, D. C. E., and Coauthors, 2016: A multi-decade - data in version 3 of the Sur /Reporte-Inf lacion/2017/marzo/reporte-de-inf lacion record of high-quality CO f 2 Atlas (SOCAT). Earth Syst. Sci. Data , 8 , face Ocean CO -marzo-2017.pdf. 2 383–413, http://doi.org/10.5194/essd-8-383-2016. Beck, F., A. Bárdossy, J. Seidel, T. Müller, E. Fernandez San - - Baldwin, M. P., and Coauthors, 2001: The qua chis, and A. Hauser, 2015: Statistical analysis of sub-daily si-biennial oscillation. Rev. Geophys. , 39 , 179–229, precipitation extremes in Singapore. J. Hydrol.: Reg. Stud. , http://doi.org/10.1029/1999RG000073. 3 , 337–358, http://doi.org/10.1016/j.ejrh.2015.02.001. | AUGUST 2018 STATE OF THE CLIMATE IN 2016 S271

292 Becker, A., P. Finger, A. Meyer-Christoffer, B. Rudolf, K. ——, ——, S. B. Goldenberg, T. Kimberlain, C. W. Landsea, - Schamm, U. Schneider, and M. Ziese, 2013: A descrip - R. Pasch, and J. Schemm, 2009: Tropical cyclones: Atlan - tion of the global land-surface precipitation data prod tic basin [in “State of the Climate in 2008”]. Bull. Amer. ucts of the Global Precipitation Climatology Centre (8), S79–S83. 90 , Meteor. Soc. with sample applications including centennial (trend) ——, ——, T. B. Kimberlain, C. W. Landsea, J. Schemm, Earth Syst. Sci. Data analysis from 1901–present. , 71–99, 5 , R. J. Pasch, and S. B. Goldenberg, 2011: Tropical cy - http://doi.org/10.5194/essd-5-71-2013. clones: Atlantic basin [in “State of the Climate in 2010”]. Beckley, B. D., P. S. Callahan, D. W. Hancock, G. T. Mitchum, , 92 (6), S115–S121, http://doi Bull. Amer. Meteor. Soc. and R. D. Ray, 2017: On the “Cal-Mode” correction to .org/10.1175/1520-0477-92.6.s1. ——, ——, C. W. Landsea, T. B. Kimberlain, S. B. Goldenberg, TOPEX satellite altimetry and its effect on the global J. Schemm, and R. J. Pasch, 2012: Tropical cyclones: At - , 122 , J. Geophys. Res. Oceans mean sea level time series. Bull. Amer. lantic basin [in “State of the Climate in 2011”]. 8371–8384, http://doi.org/10.1002/2017JC013090. (7), S99–S105, http://doi.org/10.1175 , Meteor. Soc. 93 Beerling, D. J., and D. L. Royer, 2011: Convergent Ce - history. Nat. Geosci. , 4 , 4 18 – 4 2 0 , /2012BAMSStateoftheClimate.1. nozoic CO 2 http://doi.org/10.1038/ngeo1186. ——, C. W. Landsea, S. B. Goldenberg, R. J. Pasch, E. S. Blake, Behera, S. K., J. J. Luo, S. Masson, S. A. Rao, H. Sakuma, J. Schemm, and T. B. Kimberlain, 2014: Tropical cyclones: - and T. Yamagata, 2006: A CGCM study on the iterac Bull. Atlantic basin [in “State of the Climate in 2013”]. tion between IOD and ENSO. , 19 , 1688–1705, J. Climate , Amer. Meteor. Soc . 95 (7), S 86 –S9 0, http://doi.org/10.1175 http://doi.org/10.1175/jcli3797.1. /2014BAMSStateoftheClimate.1. Behrenfeld, M. J., and Coauthors, 2006: Climate-driven ——, ——, E. S. Blake, J. Schemm, S. B. Goldenberg, T. B. trends in contemporary ocean productivity. , Nature , 444 Kimberlain, and R. J. Pasch, 2015: Tropical cyclones: At - 752–755, http://doi.org/10.1038/nature05317. Bull. Amer. lantic basin [in “State of the Climate in 2014”]. - ——, and Coauthors, 2016: Revaluating ocean warming im Meteor. Soc. , 96 (7), S105–S107, http://doi.org/10.1175 , 6 Nat. Climate Change , pacts on global phytoplankton. /2015BAMSStateoftheClimate.1. 323–330, http://doi.org/10.1038/nclimate2838. ——, ——, ——, ——, ——, ——, and ——, 2016: Tropical cy - Behringer, D. W., M. Ji, and A. Leetmaa, 1998: An improved clones: Atlantic basin [in “State of the Climate in 2015”]. coupled model for ENSO prediction and implications 97 (8), S105 –S10 6 , http://doi.org Bull. Amer. Meteor. Soc. , - for ocean initialization. Part I: The ocean data as /10.1175/2016BAMSStateoftheClimate.1. , similation system. 126 , 1013–1021, Mon. Wea. Rev. ——, M. L’Heureux, and M. S. Halpert, 2017a: ENSO and http://doi.org/10.1175/1520-0493(1998)126<1013:aicmfe>2.0.co;2. the tropical Pacific [in “State of the Climate in 2016”]. Bell, G. D., and M. Chelliah, 2006: Leading tropical modes (8), S93–S98, http://doi.org 98 , Bull. Amer. Meteor. Soc. - associated with interannual and multidecadal f luctua /10.1175/2017BAMSStateoftheClimate.1. 19 , J. Climate tions in North Atlantic hurricane activity. , ——, and Coauthors, 2017b: Tropical cyclones: Atlantic 590–612, http://doi.org/10.1175/jcli3659.1. Bull. Amer. basin [in “State of the Climate in 2016”]. ——, M. S. Halpert, C. F. Ropelewski, V. E. Kousky, A. V. Meteor. Soc. (8), S108–S112, http://doi.org/10.1175 98 , Douglas, R. C. Schnell, and M. E. Gelman, 1999: The /2017BAMSStateoftheClimate.1. 1998 North Atlantic and eastern North Pacific hurricane Bellouin, N., J. Quaas, J. J. Morcrette, and O. Boucher, season [in “Climate Assessment for 1998”]. Bull. Amer. 2013: Estimates of aerosol radiative forcing from the Meteor. Soc. , 80 (5), S24–S28. MACC re-analysis. Atmos. Chem. Phys. , 13 , 2045–2062, - ——, and Coauthors, 2000: The 1999 North Atlantic and east http://doi.org/10.5194/acp-13-2045-2013. - ern North Pacific hurricane season [in “Climate Assess Bernhard, G., and Coauthors, 2015: Comparison of OMI UV , Bull. Amer. Meteor. Soc. (6), S19–S22, ment for 1999”]. 81 observations with ground-based measurements at high http://doi.org/10.1175/1520-0477(2000)081<1328:caf>2.3.co;2. northern la