v4 2 data quality document

Transcript

1 Help Overview JPL D-33509 Rev. D Table Earth Observing System (EOS) BrO S T Aura Microwave Limb Sounder (MLS) CH S 3 Cl T CH Version 4.2x Level 2 data quality and S 3 CN T description document. CH S 3 Equator FWHM / km FWHM / km OH 800 1000 1200 12 -2 0 2 4 6 8 10 200 400 600 0 0.1 T ClO S 1.0 T S CO 10.0 T Pressure / hPa GPH S 100.0 T H S 1000.0 2 O -4 -2 0 2 4 1.2 0.2 0.0 0.4 0.6 0.8 1.0 -0.2 Profile number Kernel, Integrated kernel T 0 N 70 FWHM / km FWHM / km HCl S 0 1200 8 600 800 1000 12 10 400 6 4 2 0 -2 200 0.1 T HCN S 1.0 T HNO S 10.0 Pressure / hPa 3 T 100.0 HO S 2 T 1000.0 HOCl -2 1.2 -0.2 -4 0.0 0 2 4 0.2 0.4 0.6 0.8 1.0 S Kernel, Integrated kernel Profile number T IWC S Nathaniel J. Livesey, William G. Read, Paul A. Wagner, Lucien Froidevaux, Alyn Lambert, T Gloria L. Manney, Luis F. Millán Valle, Hugh C. Pumphrey, Michelle L. Santee, IWP S Michael J. Schwartz, Shuhui Wang, Ryan A. Fuller, Robert F. Jarnot, Brian W. Knosp, T Elmain Martinez, Richard R. Lay N S 2 O T Version 4.2x–3.1 S O 3 January 29, 2018 T OH S T RHI S Jet Propulsion Laboratory California Institute of Technology T SO Pasadena, California, 91109-8099 S 2 T © 2017 California Institute of Technology. Government sponsorship acknowledged. S T T

2 Help Overview Table Navigating this document Clicking on “Help” takes you back to this page. BrO S T CH S 3 Cl An overview of the information in this document is given on the next page. You can T CH return to it by clicking on the “Overview” tab in the navigation bar. S 3 CN T CH S 3 OH First time users of MLS data are advised to read Chapter 1 rst. ¿e discussion includes a table summarizing key aspects of each MLS product. ¿is table can be accessed from T ClO S anywhere in the document by clicking on the “Table” tab in the navigation bar. T S CO T Chapter 3 describes each of the MLS data products individually. You can use the navi- GPH S gation bar to skip immediately to any of the product-specic sections. T H S Each section includes 2 O T • A set of rules for screening out data points not recommended for scientic use. HCl S A table summarizing the precision, accuracy, resolution for each MLS product. • T HCN S Clicking on the product name in the navigation bar takes you to the beginning of that T section; clicking on the “S” and “T” symbols takes you directly to the screening rules HNO S and summary table, respectively, for that product. 3 T HO S 2 T HOCl S T IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S Acknowledgment T ¿is research was carried out at the Jet Propulsion Laboratory, California Institute of SO S Technology, under a contract with the National Aeronautics and Space Administration. 2 T S T Aura Microwave Limb Sounder (MLS) ii Level 2 Version 4.2 x Quality T

3 Help Overview Table Where to nd answers to key questions BrO S Do not use data for any prole where the eld • ¿is document serves two purposes. Firstly, T CH is an odd number. Status to summarize the quality of version 4.x Aura MLS S 3 Cl Level 2 data. Secondly, to convey important infor- • Status Data for proles where the eld is non T mation on how to read and interpret the data to the zero should be approached with caution. See sec- CH S scientic community. 3 tion 1.6 on page 5, and the product by product de- CN ¿e MLS science team strongly encourages scription in chapter 3 for details on how to inter- T users of MLS data to thoroughly read this docu- CH S pret the eld. Status 3 ment. Chapter 1 describes essential general infor- OH Do not use any data for proles where the • mation for all users. Chapter 2 is considered back- T smaller eld is Quality than the threshold given groundmaterialthatmaybeofinteresttodatausers. ClO S in the section of chapter 3 describing your prod- Chapter 3 discusses individual MLS data products T uct of interest. in detail. S CO For convenience, this page provides informa- Do not use any data for proles where the • T tion on how to quickly ascertain answers to antic- Convergence eld is larger than the threshold GPH S ipated key questions. given in the section of chapter 3 describing your T product of interest. H S x MLS Level 2 data? Where do I get v4.2 2 O • Some products require additional screening to re- All the MLS Level 2 data described here can be ob- T move biases or outliers, as described in chapter 3. HCl tained from the NASA Goddard Space Flight Cen- S ter Data and Information Services Center (GSFC- Information on the accuracy of each product is • T HCN DISC, see http://disc.gsfc.nasa.gov/ ). S giveninChapter3. Inthepreviousversionsofthis document, the numbers given were for the previ- T What format are MLS Level 2 data les in? HNO ousv3.3dataset. ¿isdocumenthasnowbeenup- S How do I read them? dated with the specic v4.2 x accuracy estimates. 3 MLS Level 2 data are in HDF-EOS version 5 format. T Data users are strongly encouraged to contact the • HO Details are given in section 1.5 (page 4). S 2 MLS science team to discuss their anticipated us- T Which MLS data points should be avoided? ageofthedata, andarealwayswelcometoaskfur- HOCl S How much should I trust the remainder? ther data quality questions. ¿ese issues are described in section 1.6 (starting on T IWC S page 5), and on a product by product basis in chap- Why do some species abundances show ter 3. ¿e key rules are: T negative values, and how do I interpret IWP S • Only data within the appropriate pressure range these? T (described product by product in chapter 3) are N Some of the MLS measurements have a poor signal S to be used. 2 O to noise ratio for individual proles. Radiance noise T • Always consider the precision of the data, as re- can naturally lead to some negative values for these S O eld. ported in the L2gpPrecision 3 species. It is critical to consider such values in sci- T entic study. Any analysis that involves taking some Do not use any data points where the precision • OH S form of average will exhibit a high bias if the points is zero or a negative number. ¿is indicates poor T with negative mixing ratios are ignored. information yield from MLS. RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x iii Level 2 Version 4.2 Quality T

4 Help Overview Table Document revision history BrO Version 4.x-0.1 S A pre-release version of the document, starting from the most recent version 3 document. T CH S 3 Cl Version 4.x-1.0 T CH S ¿e rst publicly released version of this document. Released when the MLS v4.20 data set was made publicly 3 CN available. T CH S Version 4.x-2.0 3 OH ¿esecondpubliclyreleasedversionofthedocument. ¿isdescribestheminorchangesassociatedwithv4.22 T ClO S (which xes an issue that gave poor temperature proles at the end of many days). In addition, this version also include minor corrections and updates (notably a x to the averaging kernel plots which had the tropical T ○ N cases swapped). and 70 S CO T GPH S Version 4.x-3.0 T ¿e third released version of the document. ¿e changes mainly reect updates of systematic error analysis H S results specic to version 4. 2 O T HCl S Version 4.x-3.1 T Aminorupdatetoreectthechangeinrecommendationof Quality thresholdsforN O(190GHz)andH O, 2 2 HCN S and an additional screening rule for the latter. T HNO S 3 T HO S 2 T HOCl S T IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x iv Level 2 Version 4.2 Quality T

5 Help Overview Table Contents BrO S 1 Essential reading for users of MLS version 4.x data 1 T Scope and background for this document . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 CH S 1 1.2 Overview of v4.2 x and this document . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Cl MLS data validation status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 3 T CH x Dierences between MLS v4.2 1.4 3 data and earlier v3.x data . . . . . . . . . . . . . . . . . . . . . S 3 1.5 Aura MLS le formats, contents, and rst order quality information . . . . . . . . . . . . . . . 4 CN Additional information given in the Quality , Convergence and Status 1.6 5 elds . . . . . . . . T CH S An important note on negative values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.7 3 OH 7 x 1.8 Averaging kernels for MLS v4.2 proles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . T 8 Considerations for comparisons with high vertical resolution datasets . . . . . . . . . . . . . . 1.9 ClO S 9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x A note on the HCl measurements in v4.2 1.10 T . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . O measurements in v4.2 1.11 x 9 A note on N 2 S CO A note on OH measurements in v4.2 1.12 x . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 T GPH 12 Background reading for users of MLS version 4.x data 2 S 12 2.1 Aura MLS radiance observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . T 12 2.2 Brief review of theoretical basis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H S 2 O Core CorePlusRn approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 ¿e , 2.3 T Forward models used in v4.2 x 2.4 16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . HCl S 16 ¿e handling of clouds in v4.2 2.5 x . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . T 19 ¿e quantication of systematic uncertainty in MLS Level 2 data . . . . . . . . . . . . . . . . . 2.6 HCN S A brief note on the Quality eld . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.7 T HNO S 21 Product-specic information 3 3.1 21 Overview of species-specic discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 T 22 Bromine monoxide (BrO) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 HO S 3.3 Methyl chloride (CH Cl) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 2 3 T CN) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methyl cyanide (CH 3.4 32 3 HOCl S OH) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 Methanol (CH 3.5 3 T 42 3.6 Chlorine Monoxide (ClO) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IWC S 50 Carbon monoxide (CO) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7 3.8 56 Geopotential Height (GPH) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . T IWP S Water Vapor (H 3.9 O) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 2 76 Hydrogen Chloride (HCl) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.10 T N 3.11 82 Hydrogen Cyanide (HCN) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S 2 O Nitric Acid (HNO 3.12 ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 3 T 3.13 94 ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Peroxy Radical (HO 2 S O 3 Hypochlorous Acid (HOCl) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.14 98 T 3.15 103 Cloud Ice Water Content (IWC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . OH S 107 3.16 Cloud Ice Water Path (IWP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . T 3.17 111 Nitrous Oxide (N O) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 RHI S 120 Ozone (O 3.18 ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 T 3.19 131 Hydroxyl Radical (OH) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . SO S 2 T S T Aura Microwave Limb Sounder (MLS) x v Level 2 Version 4.2 Quality T

6 Help Overview 3.20 136 Relative Humidity with respect to Ice (RHI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sulfur Dioxide (SO ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 3.21 2 Table 150 Temperature (T) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.22 BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T ClO S T S CO T GPH S T H S 2 O T HCl S T HCN S T HNO S 3 T HO S 2 T HOCl S T IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x vi Level 2 Version 4.2 Quality T

7 Help Overview Table Chapter 1 Essential reading for users of MLS version 4.x data BrO S T CH S 1.1 Scope and background for this document 3 Cl ¿is document describes the quality of the geophysical data products produced by version 4.x of the data T CH processing algorithms for the Earth Observing System (EOS) Microwave Limb Sounder (MLS) instrument S 3 on the Aura spacecra . ¿e intended audience is those wishing to make use of Aura MLS data for scientic CN study. ¿e geophysical products described in this document are all produced by the “Level 2” algorithms, and T CH S briey summarized in Table 1.1.1. 3 OH , MLS data are the fourth “public release” of MLS data, the rst being version 1.5 [ Livesey et al. ¿e v4.2 x T 2005], the second version 2.2, and the third versions 3.3 and 3.4. ¿e v2.2 data are described in a series ClO S Journal of Geophysical Research in 2007/2008. ¿is of validation papers published in a special issue of the x , and gives more general information on the use of document updates ndings from these papers for v4.2 T S CO MLS data. As always, those wishing to use MLS data are advised to consult the MLS science team concerning their intended use. T GPH In addition to describing the data quality, this document gives a brief outline of the algorithms used S to generate these “Level 2” data (geophysical products reported along the instrument track) from the input T “Level 1” data (calibrated microwave radiance observations). H S 2 O An Overview of the EOS MLS More information on the MLS instrument can be found in the document T [ Waters et al. , 2004]. A more general discussion of the microwave limb sounding technique and Experiment HCl S an earlier MLS instrument is given in [1999]. ¿e theoretical basis for the Level 2 so ware is de- Waters et al. T Livesey and Snyder scribed in [2004]. A crucial component of the Level 2 algorithms is the “Forward Model”, HCN S which is described in detail in EOS MLS Re- Read et al. [2004] and Schwartz et al. [2004]. ¿e document T [ trieved Geophysical Parameter Precision Estimates Filipiak et al. , 2004] is a very useful source of information HNO S on the expected precision of the EOS MLS data, and should be regarded as a companion volume to this doc- 3 ument. ¿e impact of clouds on MLS measurements and the use of MLS data to infer cloud properties is T HO S [2004]. All the above documents and papers are available from the MLS web site described in Wu and Jiang 2 http://mls.jpl.nasa.gov/ ). ( T HOCl IEEE Transactions on Geoscience A subset of the information in these documents is also reported in the S [2006], the algorithms that produce the data Waters et al. . An overview of MLS is given in and Remote Sensing T described here are reviewed in [2006], and Wu et al. Livesey et al. [2006], Read et al. [2006], Schwartz et al. IWC S [2006]. Other papers describe the calibration and performance of the various aspects of the MLS instrument T ,2006]. Cuddyetal. ,2006]andtheMLSgrounddatasystem[ CoeldandStek ,2006; Pickett ,2006; [ Jarnotetal. IWP S ¿e detailed validation of the MLS v2.2 dataset is described in a collection of papers in the “Aura Validation” T special issue of JGR-Atmospheres (papers published in 2007 and 2008). ¿ese are cited in Chapter 3 on a N S 2 product-by-product basis. O T S O x and this document 1.2 Overview of v4.2 3 T x essentialreading ¿eremainderofthischapterreviewsissuesthatareconsidered forusersofthev4.2 dataset. OH S Chapter 2 details relevant aspects of the MLS instrument design and operations and the theoretical basis for T . algorithms that are considered x the v4.2 background reading RHI S x Chapter 3 describes the data quality for “Standard” products from the MLS instrument for v4.2 . ¿ese T Cl, CH OH (a new product on are observations of vertical proles of the abundance of BrO, CH CN, CH 3 3 3 SO S ) ClO, CO, H , HOCl, N v4.2 , HO O, O O, HCl, HCN, HNO , and OH and SO , along with temperature, x 2 2 3 2 2 3 2 T S T Aura Microwave Limb Sounder (MLS) x 1 Level 2 Version 4.2 Quality T

8 Help 1.2. Overview of v4.2 x and this document Overview Table 1.1.1: Summary of key information for each MLS standard product. Essential additional information is given in each product section of chapter 3. Table Notes Contact name Product Quality Useful vertical Convergence [2] [1] range / hPa threshold threshold BrO S 1.05 A,D,N 1.3 10 – 3.2 BrO Luis Millán Valle T 1.3 Michelle Santee 147 – 4.6 CH3Cl N 1.05 CH S 3 Cl E,N CH3CN 46 – 1.0 Michelle Santee 1.05 1.4 T CH3OH Not to be used pending further validation Michelle Santee CH S 3 147 – 1.0 Michelle Santee ClO 1.05 1.3 B CN Hugh Pumphrey 100 – 0.0046 CO T 1.5 1.03 C CH S Michael Schwartz 215 – 146 3 OH GPH 83 – 0.001 0.2 – 1.03 Michael Schwartz T 0.9 C 261 – 100 ClO S H2O Alyn Lambert O 83 – 0.002 0.7 2.0 316 – 100 C,O William Read T S CO 1.05 HCl 100 – 0.32 1.2 – Lucien Froidevaux T HCN 0.2 2.0 A,E,N Hugh Pumphrey 21 – 0.1 GPH S See text See text C,O HNO3 Gloria Manney 215 – 1.5 T 22 – 0.046 N/A 1.1 HO2 Luis Millan & Shuhui Wang A,D,N H S 2 O 1.05 N Lucien Froidevaux 10 – 2.2 1.2 HOCl T IWC 215 – 83 N/A N/A B Alyn Lambert HCl S B Alyn Lambert N/A IWP N/A N/A T – 2.0 1.0 68 – 0.46 N2O Alyn Lambert HCN S [3] O3 Lucien Froidevaux 100 – 0.02 T C 1.0 1.03 261 – 121 Michael Schwartz HNO S 1.1 Luis Millan & Shuhui Wang D N/A 32 – 0.0032 OH 3 T [4] See text C,O RHI William Read 316 – 0.002 See text HO S 1.03 William Read E 0.95 SO2 215 – 10 2 T [5] 0.2 – 83 – 0.001 Temperature HOCl S 1.03 Michael Schwartz 261 – 100 0.9 C T Notes: IWC S N This is a “noisy” product requiring signicant averaging A Users should consider using alternative versions of this (e.g., monthly zonal mean). See text for details. T product, produced using dierent algorithms, as de- IWP O This product contains signicant outliers (e.g., spikes or S scribed in the text. oscillations) in some regions (typically related to clouds B This product has signicant biases in certain regions that T in the tropical upper troposphere). These should be may need to be accounted or corrected for in scientic N S screened out as detailed in the text. 2 studies. See text for details. O T C Interference from clouds can aect this product at cer- greater than this value. [1] Only use proles with Quality S O tain altitudes. See text for details. [2] Only use proles with Convergence less than this value. 3 D Biases in this product can be ameliorated (in selected T [3] File also contains a swath giving the column abundance conditions) by taking day/night dierences. See text. OH S above the (MLS-dened) tropopause. E At some altitudes, this product contains biases of a mag- [4] Relative humidity with respect to ice computed from the T nitude that render the product useful only for the study O and Temperature data. MLS H 2 RHI S of “enhancement events” (e.g., volcanic plumes, extreme [5] File also contains a swath giving tropopause pressure re pollution). See text for details. T (WMO denition) inferred from MLS temperatures. SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 2 Level 2 Version 4.2 Quality T

9 Help 1.3. MLS data validation status Overview O and temperature data), cloud ice water content geopotential height, relative humidity (deduced from the H 2 these proles are mostly output on and cloud ice water path, all described as functions of pressure. In v4.2 x Table 2.5km), thinning out to ~ a grid that has a vertical spacing of six surfaces per decade change in pressure ( three surfaces per decade above 0.1hPa. Exceptions to this are water vapor, temperature, ozone and relative humidity which are on a ner 12 per decade grid from 1000hPa to 1hPa. Cloud ice water content is also BrO S reported on this ne grid, though these proles do not extend to the stratosphere and mesosphere. ¿e OH T productmaintainsa6perdecadegridspacingintotheuppermesosphere. Horizontallytheprolesarespaced CH S ○ 3 great-circle angle along the orbit, which corresponds to about 160km. ¿e true vertical and along- by 1.5 Cl T track horizontal resolution of the products is typically somewhat coarser than the reporting grid described CH S here. For some of the products, the signal to noise ratio is too low to yield scientically useful data from a 3 CN single MLS prole observation. In these cases, some form of averaging (e.g., weekly maps, monthly zonal T means etc.) will be required to obtain more useful results. CH S In addition to these standard products, the algorithms also produce data for many “diagnostic” products. 3 OH ¿e bulk of these are similar to the standard products, in that they represent vertical proles of retrieved T species abundances. However, the information on these diagnostic products has typically been obtained ClO S from a dierent spectral region than that used for the standard products. ¿ese diagnostic products are not T discussed in this document. Further information on these is available from the MLS science team. S CO At the time of writing, the current version of the data processing so ware is version 4.20, producing les T . Any minor updates for bug xes, or to reect changes in input data such as the analysis v04-20 labeled GPH S a priori information for temperature will be referred to as v4.21, v4.22, etc. ¿is version of the elds used as T document is intended to be applicable to any v4.2 x data les. More substantial changes at a later date (e.g., H S due to a change in the MLS instrument performance) may necessitate a larger change in the data processing 2 O T etc., and will be accompanied by an updated x so ware and/or its conguration. ¿ese will be numbered v4.3 HCl S version of this document (though not necessarily immediately, depending on the circumstances dictating the update). T HCN S 1.3 MLS data validation status T HNO S As discussed above, a complete set of MLS validation papers describe the validation state of the earlier v2.2 1 data. ¿e majority of the v2.2 MLS data products have, accordingly, completed “Stage 3 Validation” dened 3 T as: HO S 2 Product accuracy has been assessed. Uncertainties in the product and its associated structure are T HOCl well quantied from comparison with reference in situ or other suitable reference data. Uncertain- S ties are characterized in a statistically robust way over multiple locations and time periods repre- T sentingglobalconditions. Spatialandtemporalconsistencyoftheproductandwithsimilarproducts IWC S has been evaluated over globally representative locations and periods. Results are published in the T peer-reviewed literature. IWP S x Work, including that described in this document, has re-validated the v4.2 data, with the level of scrutiny T N for some (notably ozone and water vapor) establishing them as “Stage 4” validated, dened as: S 2 O T Validation results for stage 3 are systematically updated when new product versions are released S O and as the time-series expands. 3 T OH S data and earlier v3.x data 1.4 Dierences between MLS v4.2 x T data processing so ware includes a wide range of updates and changes, leading to dierences x ¿e MLS v4.2 RHI S ranging from signicant to minor in all the MLS data products. Some of the most important are detailed below. T SO 1 S See http://science.nasa.gov/earth-science/earth-science-data/data-maturity-levels/ 2 T S T Aura Microwave Limb Sounder (MLS) x 3 Level 2 Version 4.2 Quality T

10 Help 1.5. Aura MLS le formats, contents, and rst order quality information Overview x and v3.4 datasets, several products Improved composition proles in cloudy regions: x For the earlier v3.3 ) showed poor behavior in regions of thick clouds in the upper troposphere , CO and HNO (notably O 3 3 Table (mainly in the tropics). Complex screening approaches enabled users to remove many, but not all of these spurious proles from consideration. Improved performance in this regard was a key goal for v4.2 , and the manner in which contamination from cloud signals is handled in the MLS gas- x BrO S phase retrievals was substantially modied. ¿is has resulted not only in a signicant reduction in the T number of spurious MLS proles reported in cloudy regions, but also in an appreciable simplication CH S 3 of the steps users need to take to screen out the most cloud-contaminated measurements, and a more Cl T eective screening. CH S through ad- x ¿e MLS O and HCN: Improvements to O and HCN products have been improved in v4.2 3 3 3 CN dition of specic retrieval “phases” dedicated to obtaining more accurate estimates for these products, T CH and reducing the impact of contaminating signals from other species. S 3 OH Improvements to H O product is retrieved in two “phases” with a preliminary phase ob- O: ¿e MLS H 2 2 T taining a coarse-vertical-resolution initial guess and informing the choice of smoothing constraint for ClO S a subsequent more detailed high-resolution retrieval. ¿e use of more channels and more accurate T forward model calculations in this initial retrieval have led to improvements in the convergence etc. of S CO the second phase, and thus to improvements in the H O product. 2 T ¿isproduct,obtainedfrommeasurementsinthe190-and640-GHzspectralregions New CH OH product: 3 GPH S . ¿is product is scientically x (also from a dedicated retrieval “phase”), has been introduced in v4.2 T useful in the tropical upper troposphere (see section 3.5). H S 2 O As of v4.2 x More suitable reference surface for geopotential height: , the MLS geopotential height (GPH) T product has been redened to be GPH from a reference geoid (a surface of constant geopotential) HCl S rather than from an ellipsoid, as was used in previous versions. T Asdiscussedinsection1.12, MLSobservationsofOHhavebeen More useful OH retrievals for recent years: HCN S made only sparingly in recent years in light of increased aging in the MLS “THz” subsystem making T algorithms have enabled a greater yield of useful OH proles x that measurement. Updates in the v4.2 HNO S x in the more recent years of observations aected by this aging than was the case in v3.3 . and v3.4 x 3 T In addition to these specic changes, changes in all products, including those not listed above, have re- HO S 2 sulted from updates to spectroscopy and instrument calibration knowledge, and in indirect response to the T Quality larger changes detailed above. Note that the threshold values of “ ” (see below) Convergence ” and “ HOCl S to be used in data screening have been updated for all products. All the standard product les (apart from T a priori information used in the retrieval (as an additional “swath”, see IWC, see below) now also include the IWC S below). T IWP S 1.5 Aura MLS le formats, contents, and rst order quality information T All the MLS Level 2 data les described here are available from the NASA Goddard Space Flight Center N S 2 ). ¿estandardand DataandInformationServicesCenter(GSFC-DISC,see http://disc.gsfc.nasa.gov/ O T (L2GP) les. ¿ese are standard Level 2 Geophysical Product diagnostic products are stored in the EOS MLS S O HDF-EOS (version 5) les containing swaths in the Aura-wide standard format. For more information on 3 Craig et al. [2003]. A sample reading function for the Interactive Data Language (IDL, version this format see T OH S 6.1 or later required), known as readl2gp.pro may have been supplied with the data and is available from . A reader for MATLAB http://www.openchannelsoftware.org/ Open Channel So ware Foundation the T ) is also available at the same site, and one for python is planned to be added shortly. readL2GP.m ( RHI S ¿e standard products are stored in les named according to the convention T SO S MLS-Aura_L2GP-_v04-20-c01_d.he5 2 T S T Aura Microwave Limb Sounder (MLS) x 4 Level 2 Version 4.2 Quality T

11 Help 1.6. Additional information given in the Status elds , Quality Convergence and Overview Table 1.5.1: Additional swaths in specic standard product les. Table Additional swaths Notes Product HNO3 HNO3-190, HNO3-240 Nitric acid from the 190 and 240GHz radiometers BrO S IWC IWP partial Ice Water Path. (¿is le has no a priori swath) Ozone column above the tropopause (see below) O3 O3 column T RHI UTRHI, UTRHI-APriori “Single layer” relative humidity (see Section 3.20) CH S Temperature WMOTPPressure Tropopause (WMO denition, based MLS temperature) 3 Cl T CH S 3 CN is O3 Temperature BrO where , etc. ¿e les are produced on a one-day granularity (midnight , , T CH to midnight, universal time), and named according to the observation date where is the four digit S 3 = 1 January). calendar year and is the day number in that year ( 001 OH ¿ese les contain the corresponding standard product in an HDF-EOS swath given the same name as T ClO S a priori ”. With the exception of IWC, each le also contains the H2O the product, e.g., “ values for that product inside a second swath whose name ends with the substring “ , IWC, O3, RHI, and Temper- ”. HNO -APriori 3 T ature les are special in that they carry extra standard or non-standard products, as detailed in Table 1.5.1. S CO ¿e les contain an HDF-EOS swath given the same name as the product. In addition, the standard O 3 T product les also contain swaths describing column abundances, and the standard Temperature le contains GPH S additionalswathsdescribingtropopausepressure. AssomeL2GPlescontainmultipleswaths, itisimportant T les is requested from the le. In the case where the “default” to ensure that the correct swath in the L2GP H S 2 O swath is requested (i.e., no swath name is supplied) most HDF so ware will access the one whose name T falls earliest in ASCII order. ¿is generally results in the desired result for all products. For example, for HCl S ” swath temperature, the standard “ Temperature ” product will be read in preference to the “ WMOTPPressure T that gives tropopause pressures. HCN S Each swath contains data elds L2gpValue and L2gpPrecision , which describe the value and precision T (reported at 1 should is set negative or zero L2gpPrecision ) of the data, respectively. Data points for which σ HNO S not be used a priori precision, indicating that , as this ags that the resulting precision is worse than 50% of the 3 instrumentand/orthealgorithmshavefailedtoprovideenoughusefulinformationforthatpoint. Inaddition T to these elds, elds such as latitude etc. describe geolocation information. ¿e eld time describes time, HO S 2 in the EOS standard manner, as the number of seconds elapsed (including from ve to nine subsequent leap T seconds to date) since midnight universal time on 1 January 1993. HOCl S ¿e (“Diagnostic products on a Geophysical Grid”) le contains a large number of additional L2GP-DGG T swath quantities. ¿e vast majority of these are MLS diagnostic products that will be of little interest to most IWC S data users. A few which may be of interest to some users are detailed below. T IWP S Quality 1.6 Additional information given in the Status and Convergence , T elds N S 2 O In addition to the data and their estimated precisions, three quality metrics are output for every prole of each T product. ¿e rst, called Quality , gives a measure of the quality of the product based on the t achieved by S O 3 the Level 2 algorithms to the relevant radiances. Larger values of Quality generally indicate better radiance T closer to zero indicate poorer radiance ts and Quality ts and therefore more trustworthy data. Values of OH S Quality thereforelesstrustworthydata. ¿evalueof tobeusedasa“threshold”forrejectingdatainscientic T studies varies from product to product, and is given later in this document. RHI S Status . ¿is is a 32 bit integer that acts as a bit eld containing several ¿e second quality metric is called T “ags”. Figure 1.6.1 describes the interpretation of these ags in more detail. ¿e rst two bits (bits 0 and 1) SO S are “agging” bits. If the rst bit is set it indicates that the prole . should not be used in any scientic study 2 T S T Aura Microwave Limb Sounder (MLS) x 5 Level 2 Version 4.2 Quality T

12 Help 1.6. Additional information given in the Status elds Quality , Convergence and Overview 10 5 8 3 9 0 6 1 2 4 7 Bit Table 256 64 16 128 8 1024 512 32 4 2 1 Value BrO S Flag – Bad Prole: Do not use this T prole (see “information” bits for CH S an explanation). 3 Cl T This prole is Flag – Warning: CH S questionable (see “information” 3 CN bits for an explanation). T See Flag – Comment CH S “information” bits for additional 3 OH comments concerning this prole. T ClO S (Warning) This Information: prole may have been aected by T high altitude clouds. S CO T (Warning) This Information: GPH S prole may have been aected by low altitude clouds. T H S (Comment) GEOS-5 Information: 2 O a priori temperature data were T unavailable for this prole. HCl S T (Bad prole) The Information: HCN S retrieval for this phase encountered a numerical error. T HNO S (Bad prole) Too few Information: radiances were available for good 3 T retrieval of this prole. HO S 2 Information: (Bad prole) The T task retrieving this prole crashed HOCl S (typically a computer failure). T IWC Status eld. The bits not labeled are not used in v4.2x. Later The meaning of the various bits in the Figure 1.6.1: S versions may implement specic meanings for these bits. Note that bit 6 (GEOS-5 data) was not used in v1.5, and T that the information in bits 7 and 8 were combined into bit 8 in versions 1.5 and 2.2. IWP S T is an odd number should not be used. ¿e second bit indicates that Status Accordingly, any prole for which N S 2 O data are considered questionable for some reason. Higher bits give more information on the reasons behind T 18 the setting of the rst two bits. So, for example, a value of Status of (2+16) indicates that the data are S O 3 ≡ 2 questionable ( bit 4). 16 bit 2) because of the possible presence of high altitude clouds ( ≡ T ¿e most commonly set information bits are the “high altitude cloud” and “low altitude cloud” bits. ¿ese OH S indicate that the data have been marked as questionable because the Level 2 so ware believed that the mea- T surements may have been aected by the presence of clouds (clouds alone will never cause a prole to be RHI S ). ¿e implications of this vary from product to product and, Status marked as not to be used, i.e., with odd T more importantly, height to height. For example, situations of either “low cloud” or “high cloud” typically SO S have very little impact on the quality of stratospheric data. Further details of the implications of these ags 2 T S T Aura Microwave Limb Sounder (MLS) x 6 Level 2 Version 4.2 Quality T

13 Help 1.7. An important note on negative values Overview are given later in this document on a product by product basis. ¿e third diagnostic eld Convergence describes how the t to the radiances achieved by the retrieval Table 2 algorithms compared to the degree of t to be expected. ¿is is quantied as a ratio of an aggregate value to χ that predicted based on the assumption of a linear system [ Livesey et al. , 2006]. Values around unity indicate good convergence, the threshold values above which proles should not be used are given on a product by BrO S product basis later in this document. T CH S 3 Cl 1.7 An important note on negative values T Some of the MLS observations are “noisy” in nature. A consequence of this is that negative values may o en CH S 3 not be ignored or masked be reported for species mixing ratios. It is important that such values . Ignoring CN such values will automatically introduce a positive bias into any averages made of the data as part of scientic T CH analysis. Water vapor is retrieved using a logarithmic basis (both vertically and horizontally, as discussed in S 3 OH section 1.9). Accordingly, no negative water vapor abundances are produced by v4.2 x . T ClO S proles 1.8 Averaging kernels for MLS v4.2 x T Asiscommonforremotesoundinginstruments, considerationofthe“AveragingKernel”[e.g., ,2000] Rodgers S CO can be important in some scientic studies. However, the relatively high vertical resolution of the MLS obser- T vations (compared, for example, to nadir sounding composition instruments) allows for many scientically GPH S useful studies to be undertaken without reference to the averaging kernels. ¿is section reviews the role aver- T aging kernels play in comparing MLS proles to other observations and/or model proles and describes how H S data. x to obtain representative kernels for the v4.2 2 O T ¿e averaging kernel matrix A relates the retrieved MLS proles (given by the vector x ) to the “true” ˆ HCl S x atmospheric state (the vector ) according to T x ˆ ∂ HCN S A = . (1.1) ∂ x T HNO S A matrix accordingly describe the contributions of the true atmospheric prole to the given level Rows of the 3 in the retrieved prole. ¿e gures later in this document show these rows as individual colored lines. T x , the averaging kernels, Given an independent observation or model estimate of an atmospheric prole HO S prole a priori , can be used to compute the proles that MLS would observe, in combination with the MLS x 2 a T were the true prole to be in the state given by , according to x HOCl S x − (1.2) A + x = x ˆ ] x [ T a a IWC S prole for each MLS observation is available within each of the stan- x , the As of MLS version v4.2 a priori T ” (note the hyphen). dard product les, in swaths named according to the product, with the sux “ -APriori IWP S ”. ¿is information can also be obtained from the O3-APriori ” and “ Temperature-APriori Examples are “ T les (the only source for this information in v3.3 x and v3.4 x and earlier versions). MLS L2GP-DGG N S 2 O Note that in the case of water vapor where (as described below) a logarithmic interpolation is used for T x x and containing the prole, the calculations in equation 1.2 should be performed in log space, i.e., with a S O A logarithm of the given H matrix as supplied). O mixing ratio (leaving the 3 2 T ¿e full MLS averaging kernels are complicated entities, reecting the two dimensional “tomographic” OH S nature of the MLS retrievals (see section 2.2). We anticipate that few, if any, users will need to apply these full T two dimensional kernels, whose interpretation is complex (please contact the MLS team for further infor- RHI S mation on these). ¿e full kernels can be “collapsed” in the horizontal, to provide a single vertical averaging kernel for each product (as is done for many nadir sounding instruments). Such kernels are shown for each T SO S product (along with “horizontal” averaging kernels) in chapter 3. ¿e MLS averaging kernels typically change 2 T S T Aura Microwave Limb Sounder (MLS) x 7 Level 2 Version 4.2 Quality T

14 Help 1.9. Considerations for comparisons with high vertical resolution datasets Overview little with latitude / season / atmospheric state. Accordingly, two representative kernels are shown for each product, one for the tropics and one for polar winter conditions. ¿ese representative kernels are available Table proles ob- to users as described below. If variability in the averaging kernels is a concern, comparison of ˆ x tained using the two kernels (likely to represent two extreme cases) can provide a quantitative estimate of the magnitude of dierences introduced by kernel variations. BrO S ¿e two averaging kernels for each product are distributed as text les, named according to T CH S MLS-Aura_L2AK--_v04-2x_0000d000.txt 3 Cl T ○ 70N for the equator and 70 N, respectively (or Day and Night for OH, see section 3.19). is Eq or where CH S . ¿ese les are available from the MLS web site at http://mls.jpl.nasa.gov/data/ak/ 3 CN ¿ese les contain comment lines (prexed with a semicolon) describing their format. ¿e rst non- T comment line gives the name of the product and the number of levels in the vertical prole. A list of the CH S 3 pressure levels in the prole (matching those in the L2GP les) is then given, followed by all the values of the OH averaging kernel matrix, with the row index (the level in the retrieved prole) varying most rapidly. T ClO Typically, of course, the MLS prole pressures are not those of the observation or model dataset to which S the comparison is being made. In many cases, particularly where the resolution of the other dataset is com- T parable to that of the MLS proles, a simple linear interpolation is the most practical manner in which to S CO x transform the other dataset into the prole space. However, we note that more formal approaches have been T , 2003] for the case where the comparison dataset is also remotely sounded Rodgers and Connor described [ GPH S and has an averaging kernel. In cases where the comparison dataset has high vertical resolution (e.g., sonde T or Lidar observations), an additional consideration is described in the following section. H S 2 O T 1.9 Considerations for comparisons with high vertical resolution datasets HCl S ¿e MLS Level 2 data describe a piecewise linear representation of vertical proles of mixing ratio (or tem- T HCN L2GP perature, GPH, etc.) as a function of pressure, with the tie points given in the les (in the case of water S vapor, the representation is piecewise linear in log mixing ratio). ¿is contrasts with some other instruments, T which report proles in the form of discrete layer means. ¿is interpretation has important implications that HNO S may need to be considered when comparing proles from MLS to those from other instruments or models, 3 T particularly those with higher vertical resolution. HO S It is clearly not ideal to compare MLS retrieved proles with ner resolution data by simply “sampling” 2 the ner prole at the MLS retrieval surfaces. One might expect that instead one should do some linear T HOCl S interpolation or layer averaging to convert the other dataset to the MLS grid. However, in the MLS case where the state vector describes a prole at innite resolution obtained by linearly interpolating from the T xed surfaces, it can be shown that the appropriate thing to do is to compare to a least squares t of the IWC S non-MLS prole to the lower resolution MLS retrieval grid. T , and a lower resolution MLS retrieved prole Consider a high resolution prole described by the vector z h IWP S described by the vector x . We can construct a linear interpolation in log pressure that interpolates the low l T x z resolution information in . We describe that operation by the (typically to the high resolution grid of h l N S 2 O highly sparse) n matrix n × such that H h l T = (1.3) x Hx h l S O 3 where . It can be shown that the best estimate is the high resolution interpolation of the low resolution x x h l T , prole that an idealized MLS instrument could obtain, were the true atmosphere in the state described by z h OH S is given by T z Wz = (1.4) l h RHI S where T − 1 T T SO S   H = H W H (1.5) 2 T S T Aura Microwave Limb Sounder (MLS) x 8 Level 2 Version 4.2 Quality T

15 Help 1.10. A note on the HCl measurements in v4.2 x Overview In other words, . ¿is operation is illustrated by an example in represents a least squares linear t to z z h l Figure 1.9.1. Precision uncertainty on high resolution measurements may be similarly converted to the MLS Table grid by applying T (1.6) S W WS = l h BrO S where S is the covariance of the original high resolution data (typically diagonal) and S is its low resolution h l T representation on the MLS pressure grid. Following this transfer of the high-resolution prole onto the state CH S vector vertical grid, the prole can be multiplied by the averaging kernels, as described above, according to 3 Cl equation 1.2. T In some cases, the application of this least-squares “smoothing” is as important, if not more important, CH S 3 than the application of the averaging kernels described above. ¿is is particularly true when the averaging CN kernelsareclosetodeltafunctions, indicatingthattheverticalresolutioniscomparabletotheretrievedprole T CH level spacing. S 3 OH In the case of water vapor, where a logarithmic vertical basis is used, the vectors should describe z and x the logarithm of the mixing ratio. T ClO S T x 1.10 A note on the HCl measurements in v4.2 S CO 35 isotopologue) Starting in February 2006, the primary MLS band for measuring HCl (specically the HCl T or “band 13”) began to exhibit symptoms of aging and was deactivated to conserve life. ( R4:640.B13F:HCl GPH S ¿is is likely to be due to a radiation susceptibility issue for a batch of transistors identied shortly before T R4:640.B14F:O3 or “band 14”) launch. Useful observations of HCl are still made with the adjacent band ( H 37 35 S line (and a smaller line for the HCl which, as can be seen from Figures 2.1.1 and 2.1.2 also observe the HCl 2 O isotopologue). T HCl S In order to avoid undesirable discontinuities in the v4.2 x HCl dataset, the band 13 radiances are not considered in the retrieval of the standard HCl product, even on days for which it was active. For days prior T HCN S to the 16 February 2006 deactivation of band 13, and the few subsequent days when band 13 has been (or algorithms also produce a second HCl product (in the swath may be) reactivated, the v4.2 x HCl-640-B13 T HNO in the L2GP-DGG ) le which includes the band 13 radiances, giving a product with improved precision and S resolution in the upper stratosphere and mesosphere. See section 3.10 for more information. 3 T As discussed in section 3.10, while the band 14 and band 13 data show very good agreement in the lower HO S stratosphere, theydisagreeonthemagnitudeofthedecliningtrendinupperstratosphericHCl(reectingcuts 2 T inemissionsofozonedepletingsubstances). AtthesehighaltitudestheHCllineissignicantlynarrowerthan HOCl S the single channel in band 14 in which it resides, whereas the band 13 channels (by design, as this band was targeted to HCl) resolve the line shape. Accordingly, the band 13 trend is judged to be the more accurate one. T IWC S x O measurements in v4.2 1.11 A note on N T 2 IWP S ) began to show signs ¿e preferred MLS band for measuring N O (band 12 in the 640GHz region, N2O-640 2 T of aging in June 2013 (although more slowly than the rapid decline observed in the HCl band 13 in 2006). N S O Band 12 was subsequently turned o on August 6, 2013. However, the level-2 processing stream for the N 2 2 O ) retrieval on 7 June 2013 and standard product in v3.x was switched to output the band 3 (190GHz, N2O-190 T S O later data for are not recommended for scientic use. N2O-640 3 For v4.x the decision was taken to assign the band 3 (190GHz, N2O-190 ) retrieval as the standard product T ) retrieval is now only available in the N2O-640 swath from launch, wherease the band 12 (640GHz, N2O-640 OH S in the le. L2GP-DGG T As discussed in section 3.17, the band 3 and band 12 data show good agreement except at 100 hPa where RHI S there is a 30% high bias in the N O values from band 3. > 2 T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 9 Level 2 Version 4.2 Quality T

16 Help 1.12. A note on OH measurements in v4.2 x Overview 1 Table BrO S T CH S 3 Cl 10 T CH S 3 CN T CH S 3 OH Pressure / hPa T ClO S 100 T S CO T GPH S T H S 2 O 1000 T HCl 300 100 200 400 0 S N O / ppbv 2 T HCN S Comparisons of MLS (v1.5) N O observations with in-situ balloon data (courtesy of J. Elkins). The raw Figure 1.9.1: 2 T balloon data ( z ) are shown as the grey shaded region (shading indicates precision). A coincident MLS prole ( ) x h l HNO S is shown in red with the red error bars indicating precision. The red dots show the MLS data linearly interpolated to matrix (i.e., the balloon pressures using the H x from equation 1.3). The black line shows the “least squares” inter- 3 h T z polation of the balloon data onto the MLS grid using the from equation 1.4). matrix as described in the text (i.e., W l HO S The black line therefore represents the closest possible match at this resolution to the original grey line, and is the 2 appropriate quantity to compare to the red MLS prole, and to be multiplied by the averaging kernels for formal T HOCl comparison. S T IWC S 1.12 A note on OH measurements in v4.2 x T ¿e MLS OH measurements derive from observations in the 2.5-THz region of the spectrum. ¿e local IWP S oscillator signal driving the MLS 2.5-THz radiometers is provided by a methanol laser (pumped by a CO 2 laser). In December 2009, following more than ve years of operation, this laser began to show signs of aging T N S and was temporarily deactivated (prior to the 2004 Aura launch, the expected lifetime of this laser was only 2 O eighteen months). T S Upper stratospheric and mesospheric OH is strongly aected by solar activity [ Wang et al. , 2013], which O 3 was low during the solar 11-year minimum in 2009. In order to conserve the remaining lifetime of the THz T instrument for valuable measurements when the Sun becomes more active, we suspended OH measurements OH S from December 2009 to August 2011. As the Solar Cycle 24 rises with increasing solar activity, we reactivate T THz instrument for a 30-day continuous measurement during August – September in each year. ¿e reasons RHI S of having this measurement at this time of year are: T SO S . Since MLS OH measurements started in early August in 2004, in order to achieve the longest possible 1 2 T S T Aura Microwave Limb Sounder (MLS) 10 Level 2 Version 4.2 x Quality T

17 Help 1.12. A note on OH measurements in v4.2 x Overview annual OH data record for trend studies, we have to choose a time no earlier than August. ○ N, has Wang et al. . ¿e ground-based instrument used for MLS OH validation [ 2 , 2008], located at 34.4 Table the best OH measurement signal in summer (June – August) and the lowest signal in winter. Having the annual MLS OH measurement starting in August allows for continuing validation with the best BrO S possible collocated ground-based data. T We have performed the 30-day OH measurement in each August – September during 2011–2014 and plan to CH S 3 continue it during the declining phase of Solar Cycle 24. However, due to the aging of the instrument, OH Cl T data acquired since 2014 are signicantly noisier than the previous years and have poor spatial coverage at CH S mid-low latitudes. 3 CN T CH S 3 OH T ClO S T S CO T GPH S T H S 2 O T HCl S T HCN S T HNO S 3 T HO S 2 T HOCl S T IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 11 Level 2 Version 4.2 Quality T

18 Help Overview Table Chapter 2 Background reading for users of MLS version 4.x data BrO S T CH S 2.1 Aura MLS radiance observations 3 Cl Figures 2.1.1 and 2.1.2 show the spectral coverage of the MLS instrument. ¿e instrument consists of seven ra- T diometers observing emission in the 118GHz (R1A and R1B), 190GHz (R2), 240GHz (R3), 640GHz (R4) and CH S 3 2.5THz (R5H and R5V) regions. With the exception of the two 118GHz devices, these are “double sideband” CN radiometers. ¿is means that the observations from both above and below the local oscillator frequencies are T CH S combined to form an “intermediate frequency” signal. In the case of the 118-GHz radiometers, the signals 3 OH from the upper sideband (those frequencies above the 126GHz local oscillator) are suppressed. ¿ese in- ~ T termediate frequency signals are then split into separate “bands”. ¿e radiance levels within these bands are ClO S quantied by various spectrometers. In operation, the instrument performs a continuous vertical scan of both the GHz (for R1A–R4) and THz T S CO (R5H, R5V) antennæ from the surface to about 90km in a period of about 20s. ¿is is followed by about 5s Major Frame 25s cycle is known as a (MAF). During the of antenna retrace and calibration activity. ¿is ~ T GPH ~ (MIFs). Minor Frames 20s continuous scan, radiances are reported at 1/6s intervals known as S T H S 2.2 Brief review of theoretical basis 2 O T ¿e Level 2 algorithms implement a standard Optimal Estimation retrieval approach [ Rodgers , 1976, 2000] HCl S that seeks the “best” value for the state vector (the proles of temperature and abundances) based on an estimates of the state vector (from optimal combination of the t to the MLS radiance observations, a priori T HCN S climatological information and, for temperature in the troposphere and lower stratosphere, analysis elds), and constraints on the smoothness of the result. ¿is t must o en be arrived at in an iterative manner T HNO because of the non-linear nature of the Aura MLS measurement system. S An innovative aspect of the retrieval algorithms for Aura MLS arises from taking advantage of the fact 3 T that the MLS instrument looks in the forward direction from the spacecra . Figure 2.2.1 reviews the Aura HO S MLS measurement geometry and shows that each radiance observation is inuenced by the state of the atmo- 2 T Level 2 algorithms, the state vector consists of “chunks” of sphere for several consecutive proles. In the v4.2 x HOCl S several proles of atmospheric temperature and composition, which are then simultaneously retrieved from radiances measured in a similar number of MLS scans. Results from these “chunks” are then joined together T IWC S to produce the products at a granularity of one day (the chunks overlap in order to avoid “edge eects”). ¿e retrieval state vector consists of vertical proles of temperature and composition on xed pressure T IWP surfaces. Between these xed surfaces, the forward models assume that species abundances and temperature S varyfromsurfacetosurfaceinapiecewise-linearfashion(exceptfortheabundanceofH O,whichisassumed 2 T to vary linearly in the logarithm of the mixing ratio). ¿is has important implications for the interpretation N S 2 O of the data as was described in section 1.9. In addition to these proles, the pressure at the tangent point T for the mid-point of each minor frame is retrieved, based on both radiance observations and knowledge of S O 3 tangent point height from the MLS antenna position encoder and the Aura spacecra ephemeris and attitude T determination. OH S Most of the MLS data products are deduced from observations of spectral contrast, that is, variations T in radiance as a function of frequency for a given limb pointing. Many of the systematic errors in the MLS RHI S measurement system manifest themselves as a spectrally at or slowly spectrally varying error in radiance. T ¿is is true of eects arising both from the instrument (such as gain and oset during the limb scan) and from SO S the “forward model” (such as knowledge of continuum emission and the impact of some approximations 2 T S T Aura Microwave Limb Sounder (MLS) x 12 Level 2 Version 4.2 Quality T

19 Help 2.2. Brief review of theoretical basis Overview . . 10 MHz) ∼ Table 2545 R5V:2T5 R5H:2T5 2544 BrO S 2543 663 B20UF:PT B17UF:PT 0.2 MHz resolution over 191.9000 GHz 239.6600 GHz 642.8700 GHz 126.8000 GHz (lower sideband only) 2.5227816 THz ∼ R4:640 ( Digital Autocorrelator Spectrometer Single 0.5 GHz wide filter Standard 25 channel filter bank Mid band 11 channel filter bank 2542 662 T 2541 CH 661 ✸ ● polarizations. S B31UM:BRO R2:190 R3:240 R1[A/B]:118 R4:640 R5[H/V]:2T5 Green lines correspond to 30 hPa tangent point. Red lines correspond to 100 hPa tangent point. Paler labels indicate redundant signals or alternate Blue lines correspond to 10 hPa tangent point. Arrows indicate direction of channel numbering. Local oscillator frequencies: 3 B30UM:HO2 B14UF:O3 2540 660 Cl B13UF:HCL T 2539 659 208 CH S 2538 R2:190 658 207 3 CN B27UM:HCN 2537 657 206 B6UF:O3 T 140 2536 656 205 CH R1B:118 R1A:118 B19UF:OH B16UF:OH S 139 2535 B5UF:CLO 655 204 3 OH 251 138 2534 654 203 R3:240 250 T 137 2533 653 202 B4UF:HNO3 B12UF:N2O ClO S 249 ● 136 2532 652 201 B9UF:CO B25UD:CO B3UF:N2O ● 248 B2UF:H2O 135 B23UD:H2O 2531 651 200 T B18UF:OH B15UF:OH ✸ 247 ✸ B33UW:O3.C4 134 S CO 2530 650 199 B11UF:BRO B33UW:O3.C3 B29UM:HOCL B28UM:HO2 246 B10UF:CLO 133 2529 649 198 T B8UF:PT 245 132 2528 GPH 648 197 S ✸ 244 B33UW:O3.C2 131 2527 647 196 ● B7UF:O3 B24UD:O3 T 243 130 2526 646 ✸ 195 H S B33UW:O3.C1 2 242 129 2525 O 645 194 T 241 128 2524 644 193 HCl S 240 127 2523 643 192 Frequency / GHz 239 126 2522 642 T 191 HCN 238 S 125 2521 641 190 237 124 2520 640 ✸ 189 T B33LW:O3.C1 236 123 HNO ● 2519 639 188 S B7LF:O3 B24LD:O3 235 ✸ ✸ 122 2518 638 187 3 B34LW:PT.C1 B32LW:PT.C1 B33LW:O3.C2 234 T 121 2517 637 B8LF:PT 186 ✸ HO S 233 B32LW:PT.C2 B34LW:PT.C2 120 2516 636 185 B10LF:CLO ✸ B28LM:HO2 2 B29LM:HOCL 232 B11LF:BRO ✸ B33LW:O3.C3 119 2515 635 184 ● T B1LF:PT B21LF:PT B33LW:O3.C4 B26LD:PT B22LD:PT ● HOCl 231 B15LF:OH B18LF:OH 118 S 2514 634 B2LF:H2O 183 EOS MLS Spectral Coverage (split sideband) B23LD:H2O ● B9LF:CO B3LF:N2O B25LD:CO 230 ✸ 117 2513 633 182 B32LW:PT.C3 B34LW:PT.C3 B12LF:N2O T 0 229 B4LF:HNO3 116 2512 632 181 100 300 400 200 Radiance /K IWC S ✸ 115 2511 631 180 B32LW:PT.C4 B34LW:PT.C4 T B5LF:CLO 0 114 2510 630 179 100 300 400 200 B16LF:OH B19LF:OH Radiance /K IWP S 2509 629 178 B6LF:O3 2508 628 T 177 B27LM:HCN N S 2507 627 0 176 2 100 400 300 200 O Radiance /K 2506 626 T B13LF:HCL B14LF:O3 2505 625 S B30LM:HO2 O B31LM:BRO 3 2504 624 T 2503 0 623 200 300 400 100 OH S Radiance /K B20LF:PT B17LF:PT The ve panels show the spectral regions covered by the MLS radiometers. The grey boxes and other symbols denote the position of the various 2502 T 0 2501 50 100 150 200 250 Radiance /K RHI S T SO S Spectroscopic data provided by Mark Filipiak June 9, 2003. Nathaniel Livesey . . Figure 2.1.1: “bands” observed within the spectral regions covered by each radiometer. 2 T S T Aura Microwave Limb Sounder (MLS) 13 Level 2 Version 4.2 x Quality T

20 Help 2.3. The approach Core , CorePlusRn Overview . . EOS MLS Spectral Coverage (folded sideband) Table R1A:118 B22D:PT Blue lines correspond to 10 hPa tangent point. B26D:PT 400 R1B:118 ● B32W:PT.C3 B32W:PT.C4 B32W:PT.C2 B32W:PT.C1 Green lines correspond to 30 hPa tangent point. B34W:PT.C1 B34W:PT.C2 B34W:PT.C3 B34W:PT.C4 B1F:PT 300 B21F:PT ✸ ✸ ✸ ✸ Red lines correspond to 100 hPa tangent point. BrO S Paler labels indicate redundant signals or alternate 200 Radiance /K polarizations. 100 T CH 0 Standard 25 channel filter bank S 6 12 5 13 8 9 10 11 4 7 R2:190 Mid band 11 channel filter bank 3 B23D:H2O 200 Cl B27M:HCN ● Single 0.5 GHz wide filter ✸ B4F:HNO3 B2F:H2O T B6F:O3 B5F:CLO B3F:N2O ● Digital Autocorrelator Spectrometer 150 CH ∼ ( 10 MHz) ∼ 0.2 MHz resolution over S Arrows indicate direction of channel numbering. 100 3 CN Radiance /K Local oscillator frequencies: 50 R1[A/B]:118 126.8000 GHz (lower sideband only) T R2:190 191.9000 GHz 0 CH 10 7 13 11 15 9 14 16 8 12 S R3:240 239.6600 GHz B25D:CO B24D:O3 3 300 R4:640 642.8700 GHz R3:240 ● ● B33W:O3.C4 OH B33W:O3.C3 B33W:O3.C2 B33W:O3.C1 R5[H/V]:2T5 2.5227816 THz 250 ✸ ✸ ✸ ✸ B7F:O3 B8F:PT B9F:CO 200 T 150 ClO S Radiance /K 100 50 T 0 5 6 3 11 4 9 8 7 2 10 S CO 400 R4:640 B29M:HOCL B31M:BRO B28M:HO2 B30M:HO2 300 T B11F:BRO B14F:O3 B12F:N2O GPH B13F:HCL B10F:CLO S 200 Radiance /K 100 T H S 0 2 11 10 9 8 7 5 6 20 19 18 17 16 15 14 13 12 O 250 R5H:2T5 B16F:OH B17F:PT B15F:OH T B18F:OH B19F:OH B20F:PT R5V:2T5 200 HCl S 150 T 100 Radiance /K HCN S 50 0 18 19 20 21 22 6 7 8 9 10 11 12 13 14 15 16 17 T Intermediate Frequency / GHz June 9, 2003. Nathaniel Livesey HNO Spectroscopic data provided by Mark Filipiak S . . 3 T This is similar to gure 2.1.1, except that x-axes represent “intermediate frequency”. The signal at each Figure 2.1.2: HO S intermediate frequency represents a sum of the signals observed at that frequency both above and below the local 2 oscillator (below only in the case of the 118 GHz receivers.) T HOCl S T made in the forward model in order to increase its speed). In order to account for such eects, the v4.2 x IWC S algorithms also retrieve spectrally at (or slowly spectrally varying) corrections to the MLS radiances, either in terms of an additive radiance oset or an additive atmospheric extinction. T IWP S T approach 2.3 The Core , CorePlusRn N S 2 O 2.3.1 The need for separate “phases” T Many aspects of the MLS measurement system are linear in nature. In other words, there is a linear rela- S O 3 tionship between changes in aspects of the atmospheric state and consequent changes in the MLS radiance T observations. However, there are some components of the state vector whose impact on the radiances is non- OH S linear. ¿e most non-linear of these is the estimate of the tangent pressure for each MIF of observation. ¿e T impact of water vapor in the upper troposphere on the MLS radiance observations is also highly non-linear. RHI S Solving for these aspects of the state vector therefore requires several iterations. T ¿e computational eort involved in retrieval and forward models scales very rapidly (arguably as high SO S as cubically) as a function of the size of the measurement system (i.e., the number of elements in the state 2 T S T Aura Microwave Limb Sounder (MLS) x 14 Level 2 Version 4.2 Quality T

21 Help 2.3. The CorePlusRn , Core approach Overview 1000 500 Table orbit 0 -500 BrO S -1000 ‘Vertical’ distance / km 0 -4000 -500 1000 1500 -1000 -1500 2000 2500 500 -3500 -3000 -2500 -2000 3000 3500 4000 T ‘Horizontal’ distance / km CH S 150 3 Cl 100 T CH S 50 3 CN 0 T CH S -50 ‘Vertical’ distance / km 3 250 400 100 200 300 150 50 0 -50 -100 350 -400 -350 -300 -250 -200 -150 OH ‘Horizontal’ distance / km T ClO S Figure 2.2.1: The top diagram shows a section of one orbit. Three of the 120 limb ray paths per scan are indicated by the “horizontal” lines. The lower diagram shows an expansion of the boxed region above. The straight radial lines T denote the location of the retrieved atmospheric proles. The limb ray scan closest to each prole is that whose S CO color is the same as that of the arrow underneath. The thin black line under the central prole indicates the locus T of the limb tangent point for this scan, including the eects of refraction. GPH S T and measurement vectors). ¿us it is desirable to simplify retrievals involving strongly non-linear variables H S 2 O to a small subset of the complete system, in order to cut down on the eort involved in retrievals that require T many iterations. HCl S . In the case of the MLS v4.2 phases x For this and other reasons, most retrieval algorithms are split into re- T trievals, there are many such phases. ¿e rst group of phases (collectively known as “Core”) use the 118GHz HCN S 18 and 240GHz observations of O and O O, respectively, to establish estimates for temperature and tangent 2 T pressure. Upper tropospheric 190GHz radiances are used in these early phases to establish a rst order esti- HNO S mate of upper tropospheric humidity. ¿e “Core” phases also include “cloud screening” computations (based 3 ondierencesbetweenobservedandestimatedclear-skyradiances). ¿eseidentifyminorframeswhereradi- T HO S ances in a given radiometer have been subject to signicant (and currently poorly modeled) cloud scattering. 2 x Such minor frames are ignored in v4.2 processing in certain radiometers. Including information in such T HOCl cloud-contaminated conditions is a goal for future MLS data processing versions. S , where composition pro- ¿e “Core”, phases are followed by phases such as CorePlusR3 and CorePlusR5 T ) these later phases continue to les are retrieved from a given radiometer. Sometimes (e.g., for CorePlusR3 IWC S retrieve temperature and pressure, continuing using information from the 118GHz radiometers, as in “Core”. T In other phases (e.g., the CorePlusR2 and CorePlusR4 families of phases), the 118GHz information is ne- IWP S glected and temperature and pressure are constrained to the results of “Core”. ¿is choice is made based T on extensive testing aimed at maximizing the information yield from MLS while minimizing the impact of N S 2 inevitable systematic disagreements among the dierent radiometers, introduced by uncertain spectroscopy O T and/or calibration knowledge. S O Table 2.3.1 describes the phases in more detail. Many products (e.g., ozone) are produced in more than 3 one phase. All the separate measurements of these species are produced as diagnostic quantities, and labeled T OH S according to the spectral region from which they originated. For example, the ozone obtained from the dataset as retrieval is known in the v4.2 CorePlusR2 x O3-190 . In v4.2 x in order to reduce confusion for T users of MLS data, the algorithms also output “standard products”, which is typically a copy of one of the RHI S CorePlusRn phases. For example, the “standard” chlorine monoxide product is a copy of products from the T nitric acid, the standard product represents a hybrid of the results ClO-640 x the product. In the case of v4.2 SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 15 Level 2 Version 4.2 Quality T

22 Help 2.4. Forward models used in v4.2 x Overview from two phases. Details of which standard product is obtained from which phase are given in table 2.3.2. Table x 2.4 Forward models used in v4.2 ¿e retrieval algorithms in v4.2 make use of a variety of dierent forward models. ¿e most accurate is the x BrO S [2004]. ¿is is a hybrid line- so-called “full” forward model described in Read et al. [2004] and Schwartz et al. T by-line and channel averaged model that computes radiances on appropriate grids of frequency and tangent CH S pressure that are then convolved with the MLS frequency and angular responses. 3 Cl ¿is model is generally very time consuming, although for some comparatively “clean” spectral regions T the computational burden is small enough that the full forward model can be used in the operational re- CH S 3 retrieval algorithms, its use is restricted mainly to radiance channels whose focus is the trievals. In the v4.2 x CN upper troposphere and lower stratosphere, as these radiances generally have a non-linear relationship to the T CH state vector. S 3 OH For many of the MLS channels, a simpler “Linearized” forward model can be used. ¿is model invokes a T simplerst-orderTaylorseriestoestimateradiancesasafunctionofthedeviationofthestatefromoneofsev- ClO S eral pre-selected representative states. ¿e inputs to this model are pre-computed radiances and derivatives corresponding to the pre-selected states, generated by “o-line” runs of the full forward model. T S CO ¿is model is by its nature approximate, and its use is generally conned to the upper stratosphere and mesosphere where retrievals are more linear in nature. In addition, a “cloud” forward model can be invoked T GPH Wu and to model the eects of scattering from cloud particles in the troposphere and lower stratosphere [ S , 2004]. ¿is model was used in the simulation of radiances based on known model atmospheres for the Jiang T retrieval algorithms (the handling of clouds is described in more testing, but is not invoked in the v4.2 x x v4.2 H S 2 O detail in section 2.5). T HCl S 2.5 The handling of clouds in v4.2 x T HCN S ¿in clouds and atmospheric aerosols do not aect MLS atmospheric composition measurements as the typ- ical particle sizes are much smaller than the wavelengths of the radiation being observed. ¿e MLS v4.2 x T HNO algorithms can reliably retrieve composition in moderately cloudy cases (having small limb radiance pertur- S bations) and in the case of the Core+R3 retrieval this is handled by retrieving a frequency squared dependent 3 T extinction (including background atmospheric absorption from N , H O and unknown emitters). In the 2 2 HO S other retrieval phases, by contrast, a spectrally-at baseline is used. ¿ick clouds can aect the MLS radi- 2 T ances beyond the modeling capability of this approach, mainly through scattering processes. Such situations HOCl S and the aected radiances are excluded from the retrievals, or their inuence down-weighted. ¿e rst aspect of handling clouds in v4.2 x is therefore the agging of radiances that are believed to be T IWC S signicantly contaminated by cloud eects. To determine if a cloud is present in each MLS radiance mea- surement, we estimate the so-called cloud-induced radiance ( ). ¿is is dened as the dierence between T cir T the measured radiance and the radiance from a forward model calculation assuming clear-sky conditions. IWP S Specic window channels (those that are most transparent deepest into the atmosphere) in each radiometer T are chosen to set these ags. N S 2 O Cloud screening in the Core+R2 and Core+R3 phases is based on determining the highest altitude limb T view contaminated by cloud signals and either rejecting all radiances below that altitude or, in the case of less S O 3 impacted channels, inating their estimated radiance precisions. Clouds are detected by a combination of a T 2 radiance t χ quantity from Band 8 (240GHz isotopic oxygen line) retrieval of temperature and pointing, OH S and a cloud induced radiance calculation, also using 240GHz measurements. Cloud radiance scattering 18 T O line t, based on causes severe line shape distortions which are most reliably detectable in the Band 8 O RHI S extrapolating a tangent pressure estimate into the upper troposphere from the pressure information obtained T fromthe118GHzBand1measurements(unaectedbyclouds). RHifromapreviousphaseisusedtocompute SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 16 Level 2 Version 4.2 Quality T

23 Help 2.5. The handling of clouds in v4.2 x Overview Table BrO S T cir CH S 3 Cl T CH S 3 CN T CH S 3 OH 6 km-layer centered 400 – 600 hPa. Uses only T ~ ClO S T S CO T O retrieved down to 316 hPa, other species 100 hPa (note no 2 GPH S Comment Determines cloud ags for all phases H Retrievals down to 147 hPa saturated radiances Retrievals down to 147 hPa. This phase only performed when Retrievals down to 100 hPa Low vertical resolution (6/decade) Best P/T retrieval, lower limit 261 hPa T, pTan, GPH retrieval) Retrievals down to 147 hPa Initial estimate of P/T, lower limit 261 hPa Determines cloud screening for R2 (190 GHz) and R3 (240 GHz) MLS band 13 is operating Retrievals down to 147 hPa eld” cloud detection and to compute R3 T Retrieves RHi prole from 316 hPa and up, used mostly for “near Used for agging clouds in Core+R3 and later phases and Retrievals down to 316 hPa Retrieve RHi in a forms basis for cloud ice products. Initial trace gas estimator, lower limit 100 hPa Retrievals down to 68 hPa (147 hPa for Temperature) Retrievals down to 316 hPa O, RHi are “high resolution” (12 surfaces per decade change in pressure from T 2 H S 2 O T HCl S T HCN S T HNO S R2 (190 GHz) and R2 (190 GHz) — R1A & R1B (118 GHz), R3 (240 GHz) R3 (240 GHz) R2 (190 GHz) R4 (640 GHz) R1A & R1B (118 GHz) Measurements R2 (190 GHz) R3 (240 GHz) All radiometers R4 (640 GHz) R1A & R1B (118 GHz), R5H and R5V (2.5 THz) R2 (190 GHz) R4 (640 GHz) R4 (640 GHz) R1A & R1B (118 GHz), R1A & R1B (118 GHz), R2 (190 GHz) R3 (240 GHz) R3 (240 GHz) 3 T , The phases that form the v4.2x retrieval algorithms. 3 HO S , CN, , , , O 3 3 3 3 3 2 T 3 HOCl S , HNO 3 Table 2.3.1: , HCN, CH , CO, HNO T 3 O [2] 2 , HOCl, HNO IWC 3 S 2 is low resolution except for the Core+R3 phase. 3 Cl T , ClO, O 3 3 [1] CN, HCN, ClO, HNO IWP 3 S , HOCl, HCl, O 3 2 , CH 2 2 T Cl b , HNO 2 3 O, CH O, HNO 3 N , SO S 2 2 3 2 cir O CN, SO OH, ClO, BrO, HO , CH , RHi 3 3 2 2 2 2 , CO T O, SO O, N O, N 3 2 2 2 cir SO HCl, O CH N O T, pTan (GHz), GPH H ClO, BrO, HO HCN, O SO T, pTan (GHz), GPH, O T, pTan (GHz, THz), GPH, OH, O Upper tropospheric H SO UT RHi Cloud induced radiance, IWC, IWP H T RHi, T T, pTan (GHz), GPH CH none SO Target species S O 3 T OH S T RHI S Tangent pressure and Geopotential height have been abbreviated to pTan (GHz/THz) and GPH respectively. Minor state vector components such as “baseline” and/or On high vertical resolution grid T [1] [2] Methanol InitLowCloud Phase R3InitRHi CorePlusR5 CorePlusR4B FinalPTan CorePlusR4AB14 InitRHi CorePlusR4AB13 Hydrogen Cyanide HighCloud InitUTH CorePlusR3 CloudDetector InitR2 OzoneOnly InitPTan CorePlusR2 SO S ‘extinction’ have been omitted unless they are the specic focus of the phase. Temperature, IWC, H 1000 hPa to 1 hPa) unless otherwise stated. O 2 T S T Aura Microwave Limb Sounder (MLS) x 17 Level 2 Version 4.2 Quality T

24 Help 2.5. The handling of clouds in v4.2 x Overview Table BrO S T CH S 3 Cl T The origin of each of the “standard products” from v4.2x. Table 2.3.2: CH S 3 CN Product Origin Spectral region T CH S BrO CorePlusR4AB14 640 GHz 3 OH CH Cl CorePlusR4AB14 640 GHz 3 T 640 GHz CN CorePlusR4AB14 CH 3 ClO S CH Methanol 640 GHz OH 3 T S CO CorePlusR4AB14 640 GHz ClO T CO CorePlusR3 240 GHz GPH S CorePlusR3 118 & 240 GHz GPH T H H 190 GHz CorePlusR2 O S 2 2 O 640 GHz HCl CorePlusR4AB14 T HCl S HydrogenCyanide 190 GHz HCN T (15 hPa and less) CorePlusR2 190 GHz HNO HCN 3 S (larger than 15 hPa) CorePlusR3 240 GHz T CorePlusR4AB14 640 GHz HO 2 HNO S HOCl CorePlusR4AB14 640 GHz 3 T HighCloud 240 GHz IWC HO S IWP HighCloud 240 GHz 2 T HOCl N O 190 GHz CorePlusR2 2 S O OzoneOnly 240 GHz 3 T IWC S CorePlusR5 2.5 THz OH Computed from Temperature and T 190 GHz RHi IWP H O S 2 240 GHz CorePlusR3 SO 2 T N S Temperature CorePlusR3 118 & 240 GHz 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 18 Level 2 Version 4.2 Quality T

25 Help 2.6. The quantication of systematic uncertainty in MLS Level 2 data Overview MLS frequency channels and thresholds for cloud ag Table 2.5.1: Table USB/LSB frequency / GHz Low threshold High threshold Radiometer Cloud channel − T 115.3 (LSB only) B[32/34]W:PT.C4 K < none 4 R1[A/B]:118 cir BrO S T R2:190 B5F:ClO.C1 10 K 178.8 / 204.9 T > < − 20 K cir cir T 2 none χ R3:240 > 30 B8F:PT 233.4 – 234.5 / 244.8 – 245.9 CH S 3 10 635.9 / 649.8 B11F:BrO.C23 none K T − < R4:640 Cl cir T CH S 3 CN 2 ). ¿e vertical proles of χ and T are analyzed to nd the highest limb view a cloud induced radiance (T cir cir T aected by cloud scattering, and radiances below are handled accordingly. CH S is the estimation of cloud ice water content (IWC) and ice x ¿e other aspect of cloud handling in v4.2 3 OH computed by the retrieval in the water path (IWP) products from the nal HighCloud T phase. More cir T information on these products and their derivation is given in section 3.15. ClO S T 2.6 The quantication of systematic uncertainty in MLS Level 2 data S CO A major component of the validation of MLS data is the quantication of the various sources of systematic T GPH uncertainty. ¿esecan arisefrominstrumentalissues (e.g., radiometric calibration, eld ofviewcharacteriza- S tion), spectroscopic uncertainty, and through approximations in the retrieval formulation and implementa- T MLS products, x tion. A comprehensive quantication of these uncertainties has been performed for the v4.2 H S 2 O updatedfromthatdescribedintheMLSvalidationpapers(whichwerebasedonv2.2). ¿eindividualsections T of Chapter 3 detail the results of this quantication product-by-product. HCl S For each identied source of systematic uncertainty, its impact on MLS measurements of radiance (or T pointing where appropriate) has been quantied and modeled. ¿ese modeled impacts correspond to either HCN S 2 σ estimates of uncertainties in the relevant parameter(s), or an estimate of their maximum reasonable er- T ror(s) based on instrument knowledge and/or design requirements. Accordingly, the numbers reported for HNO S estimates of potential systematic error. each product in Chapter 3 reect 2 σ 3 For most of the uncertainty sources, the impact on MLS standard products has been quantied by run- T HO ning perturbed radiances through the MLS data processing algorithms. Other (typically smaller) uncertainty S 2 sources have been quantied by simple perturbation calculations. T Although the term “systematic uncertainty” is o en associated with consistent biases and/or scaling er- HOCl S rors, many sources of “systematic” error in the MLS measurement system give rise to additional scatter. For T spectroscopy, while being a bias on the fundamental parameter, will have an example, an error in the O 3 IWC S impact on the retrievals of species with weaker signals (e.g., CO) that is dependent on the amount and mor- T phology of atmospheric ozone. ¿e extent to which such terms can be expected to average down is estimated IWP S to rst order by these “full up studies” through their separate consideration of the bias and scatter each un- T certainty source introduces. N S ¿e results of these studies are summarized as “accuracy” (and in some cases additional contributions to 2 O “precision”) on a product by product basis in the next chapter. More details on the quantication for each T S O [2007] gives more product are given in the MLS validation papers. In addition Appendix A of Read et al. 3 specic details of the perturbations used in the study. T OH S 2.7 A brief note on the eld Quality T RHI S L2GP As described in section 1.6, the Quality eld in the les gives a measure of the t achieved between T the observed MLS radiances and those computed by the forward model given the retrieved MLS proles. SO 2 S statistic for all the radiances considered to have signicantly aected the Quality is computed from a χ 2 T S T Aura Microwave Limb Sounder (MLS) x 19 Level 2 Version 4.2 Quality T

26 Help 2.7. A brief note on the eld Quality Overview retrieved species (i.e., those close to the relevant spectral lines), normalized by dividing by the number of 2 , i.e., poor ts). is simply the reciprocal of this statistic (i.e., low values indicate large Quality χ radiances. Table 2 Ideally, the typical values of these normalized χ statistics will be around one, indicating that radiances Quality will therefore also ideally have a typical value of one. are typically tted to around their noise levels. For some species, however, because of uncertain knowledge of spectroscopy and/or instrument calibration, BrO S algorithms are known to be consistently unable to t some observed radiances to within their pre- the v4.2 x T dicted noise. In many of these cases, the noise reported on the radiances has been “inated” to allow the CH S 3 retrieval more leeway in tting to radiances known to be challenging. As the noise level is the denominator Cl 2 2 T statistics that are less than one and thus typical values χ statistic, these species will have typical χ in the CH S from one species to another do not reect of Quality higher than one. Accordingly, dierences in Quality 3 CN Quality the species’ relative validity, nor do version-to-version increases in for a given product necessarily T indicate improvements (or vice versa) CH S 3 OH T ClO S T S CO T GPH S T H S 2 O T HCl S T HCN S T HNO S 3 T HO S 2 T HOCl S T IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 20 Level 2 Version 4.2 Quality T

27 Help Overview Table Chapter 3 Product-specic information BrO S T CH S 3.1 Overview of species-specic discussion 3 Cl ¿issectiondescribeseachMLSv4.2 “standardproduct”inmoredetail. Anoverviewisgivenoftheexpected x T CH resolution, precision and accuracy of the data. ¿e resolution is characterized by the averaging kernels de- S 3 x scribed below. Precision is quantied through a combination of the precision estimated by the MLS v4.2 CN algorithms, through reference to the systematic uncertainty budget described in section 2.6, and through T CH S study of the actual MLS data (e.g., consideration of the observed scatter in regions where little natural vari- 3 OH ability is anticipated). T ¿e systematic uncertainty reported is generally based on the study described in section 2.6. However, in ClO S some cases larger disagreements are seen between MLS and correlative observations than these quantica- tions would imply. In such cases (e.g., MLS 215hPa CO) the uncertainty quoted reects these disagreements. T S CO A note on the averaging kernel plots T GPH S ¿e averaging kernels shown in this section describe both the horizontal (along track) and vertical (pressure) x data. While the averaging kernels vary somewhat from prole to prole, their resolution of the MLS v4.2 T H variationissucientlysmallthatthesesamplescanbeconsideredrepresentativeforallproles. ¿eaveraging S 2 O kernel plots are accompanied by estimates of the horizontal and vertical resolution of the product dened by T the full width at half maximum of the kernels. Each kernel plot also shows the integrated areas under the HCl S kernels. T HCN S T HNO S 3 T HO S 2 T HOCl S T IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 21 Level 2 Version 4.2 Quality T

28 Help Overview 3.2 Bromine monoxide (BrO) Table BrO Swath name: Useful range: 10 – 3.2 hPa (day/night dierences needed) BrO S < > Contact: [email protected] Email: Luis Millan, T CH S 3 Cl 3.2.1 Introduction T CH CoreR4AB13 ¿e standard product for BrO is taken from the 640-GHz retrieval. ¿e spectral signature of S 3 BrO in the MLS radiances is very small, leading to a very poor signal-to-noise ratio on individual MLS obser- CN vations. Signicant averaging (e.g., monthly zonal means) is required to obtain scientically useful results. T CH S Large biases of between 12 to 43pptv (typical BrO abundances range from 5 to 15pptv) are seen in the data. 3 OH ¿esebiasescanbeminimizedbytakingday/nightdierences. Forpressuresof4.6hPaandgreater, nighttime T BrO is negligible; however, for smaller pressures, nighttime BrO needs to be taken into account. Table 3.2.1 ClO S x BrO product. ¿e accuracy assessment summarizes the precision, accuracy, and resolution of the MLS v4.2 T , 2007]. Kovalenko et al. is based on v2.2 data, as described in the validation paper [ S CO Note, the v4.2 x “standard” BrO product (as with earlier versions) contains systematic biases and horizontal oscillations that present a larger challenge than for other species. ¿ose interested T GPH S in using MLS BrO in scientic studies are strongly advised to contact the MLS team before em- barking on their research. Dierent algorithms for BrO have been developed by the MLS team, T H , 2012], and those products are avail- Millán et al. aimed at ameliorating some of these artifacts [ S 2 O ” and it is available from: ML3DZMBRO able from the GSFC DISC. ¿e product short name is “ T ftp://acdisc.gsfc.nasa.gov/data/s4pa/Aura_MLS_Level3/ML3DZMBRO.004/ HCl S T 3.2.2 Vertical Resolution HCN S MLS BrO is about 5.5km in the 10 to 4.6hPa x Figure 3.2.1 shows that the vertical resolution for the v4.2 T pressure region, degrading to 6km at 3.2hPa. HNO S 3 3.2.3 Precision T HO S ¿e expected precision in a retrieved prole is calculated from radiance noise and reported for each retrieved 2 data point. ¿e value of the expected precision is agged negative or zero if it is worse than 50% of the value T HOCl S of the a priori precision. Figure 3.2.2 compares the expected precision (thick line) on an individual MLS BrO measurement with that deduced from observations of scatter in night-time observations (expected to be T ○ zero). Also shown are the expected precisions for daily, monthly, and yearly 10 zonal means. For the minimal IWC S ○ zonal mean, which corresponds to about 3,000 measurements, the averaging recommended, a monthly 10 T 4ppt. See Table 3.2.1 for more details. precision is about ± IWP S T 3.2.4 Accuracy N S 2 O ¿e accuracy of the MLS BrO product is summarized in Table 3.2.1. ¿e eect of each identied source T , 2007]. Read et al. of systematic error on MLS measurements of radiance has been quantied and modeled [ S O 2 estimatesofuncertaintiesineachMLSproduct,oranestimate σ ¿esequantiedeectscorrespondtoeither 3 T of the maximum reasonable uncertainty based on instrument knowledge and/or design requirements. More OH S [2007]. While that paper described v2.2 BrO, ndings are expected Kovalenko et al. discussion is given in T . ¿e potential systematic bias in MLS BrO measurements can be as high as x to be applicable also to v4.2 RHI S ± 43ppt at 10hPa, decreasing to about ± about 12pptv at 3.2hPa. ¿e systematic bias is dramatically reduced by subtracting the nighttime signal from the daytime signal. Taking the day/night dierence reduces the T SO S ± 16pptv at 10hPa and to ± 9pptv at 3.2hPa. If the MLS BrO data is used at 3.2 hPa, the systematic biases to 2 T S T Quality x 22 Level 2 Version 4.2 Aura Microwave Limb Sounder (MLS) T

29 Help 3.2. Bromine monoxide (BrO) Overview 0 N 70 Equator FWHM / km FWHM / km Table 8 0 6 -2 0 2 4 6 8 10 12 12 10 4 2 -2 0.1 BrO S T CH S 3 Cl T CH S 1.0 3 CN Pressure / hPa T CH S 3 OH T ClO S 10.0 T 0.6 0.4 1.2 1.0 0.8 0.6 0.2 0.0 -0.2 1.0 0.8 1.2 -0.2 0.0 0.2 0.4 S CO Kernel, Integrated kernel Kernel, Integrated kernel T GPH S ◦ Figure 3.2.1: N (right); Typical vertical averaging kernels for the MLS v4.2x BrO data at the equator (left) and at 70 T variation in the averaging kernels is suciently small that these are representative of typical proles. Colored lines H S 2 O show the averaging kernels as a function of MLS retrieval level, indicating the region of the atmosphere from which T information is contributing to the measurements on the individual retrieval surfaces, which are denoted by plus HCl S signs in corresponding colors. The dashed black line indicates the vertical resolution, determined from the full width at half maximum (FWHM) of the averaging kernels, approximately scaled into kilometers (top axes). The solid T HCN black line shows the integrated area under each kernel; values near unity imply that the majority of information S for that MLS data point has come from the measurements, whereas lower values imply substantial contributions T from a priori information. The low signal to noise for this product necessitates the use of signicant averaging (e.g., HNO S monthly zonal mean), making horizontal averaging kernels largely irrelevant. 3 T HO S 0.1 2 T HOCl S T 1.0 IWC S T IWP S Pressure / hPa 10.0 T N S 2 O T 100.0 1.0 100.0 1000.0 0.1 10.0 S O BrO precision / pptv 3 T OH S Figure 3.2.2: Comparison of the MLS v4.2x BrO precision as estimated from scatter in the retrieved data (circles) T with that expected from the retrieval (thick line), for a single prole. Also shown is the expected precision for the RHI S ◦ day/night dierence of 10 zonal mean proles averaged over a day (dotted line), a month (thin line) and a year T (dashed line). SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 23 Level 2 Version 4.2 Quality T

30 Help 3.2. Bromine monoxide (BrO) Overview day/night dierence value will need to be adjusted to compensate for the non-negligible nighttime BrO. We note that this method of taking day/night dierences is not applicable for polar summer and winter during Table periods of continuous sunlight or darkness. 3.2.5 Data screening BrO S Pressure range: 10–3.2 hPa T CH Values outside this range are not recommended for scientic use. S 3 Cl Averaging required: Signicant averaging (such as monthly zonal means) is required if useful scientic T CH data are sought. S 3 CN Diurnal dierences: For use in any scientic study, day / night or ascending / descending dierences T should be used to alleviate biases. CH S Notethat,for3.2hPa,thenon-zeronighttimeexpectedabundancesBrOneedstobetakenintoaccount. 3 OH Estimated precision: Only use values for which the estimated precision is a positive number. T ClO S information has a strong inuence are agged with negative or zero precision, Values where the a priori T and should not be used in scientic analyses (see Section 1.5). S CO Status ag: Only use proles for which the Status eld is an even number. T indicate that the prole should not be used in scientic studies. See Section 1.6 Odd values of Status GPH S for more information on the interpretation of the Status eld. T Clouds: Proles identied as being aected by clouds can be used with no restriction. H S 2 O eld is greater than 1.3 should be used. Quality: Only proles whose Quality T HCl S Convergence: Only proles whose Convergence eld is less than 1.05 should be used. T HCN S 3.2.6 Artifacts T SignicantbiasesexistintheBrOdata,asdiscussedabove. Day/night(orascending/descending)dierences HNO S Kovalenko et al. , must be used to reduce these. For 3.2 hPa, nighttime BrO needs to be taken into account [ 3 2007]. T x (and earlier) A systematic horizontal (i.e., prole-to-prole) oscillation has been discovered in MLS v4.2 HO S standard BrO product. ¿is presents a signicant challenge to the interpretation of the BrO observations. 2 T UsersarestronglyadvisedtocontacttheMLSteambeforeembarkingonresearchinvolvingtheMLSstandard HOCl S BrO product. As described above, use of other BrO products is preferable to using the Level 2 BrO data. T IWC S 3.2.7 Review of comparisons with other data sets T , from MLS measurements of BrO using a photochemical model, and We have calculated total bromine, Br y IWP S similarly inferred from balloon-borne measurements of BrO; good agreement is seen compared this with Br y , 2007]. [ Kovalenko et al. T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 24 Level 2 Version 4.2 Quality T

31 Help 3.2. Bromine monoxide (BrO) Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T ClO Table 3.2.1: Summary of the Aura MLS v4.2x BrO product. S Day/night T a S CO Pressure dierence Precision Vertical Comments b range accuracy / pptv res. / km T GPH S / pptv T – 2.2hPa and less – – Unsuitable for scientic use H S Need to account for 2 O 3.2hPa ± 9 6 5 ± non-negligible night T time BrO HCl S T ± 5.5 4 12 4.6 ± HCN S ± 5.5 6.8 14 ± 4 10 5.5 ± 4 ± 16 T HNO S 150–15hPa – – – Unsuitable for scientic use 1000–215hPa – – Not retrieved – 3 T HO S a ○ ¿e precision quoted is for a 10 monthly zonal mean 2 b T Because of large biases in the data, the daytime and nighttime BrO data are unsuitable for scientic use, so day/night dierences HOCl must be used. Note that day/night dierences are not useful for polar winter and summer, where BrO does not undergo a diurnal S variation. T IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 25 Level 2 Version 4.2 Quality T

32 Help Overview 3.3 Methyl chloride (CH Cl) 3 Table Swath name: CH3Cl 147 – 4.6 hPa Useful range: BrO S [email protected] > Michelle Santee, Contact: Email: < T CH S 3 Cl 3.3.1 Introduction T CH 0.1–0.4ppbv) negative bias at re- ~ ¿e v2.2 MLS ClO measurements were characterized by a substantial ( S 3 [2008] suggested that contamination from Santee et al. trieval levels below (i.e., pressures larger than) 22hPa. CN an interfering species such as CH Cl, which has lines in two wing channels of the 640-GHz band used to T 3 CH S measure ClO, could have given rise to the bias; they showed results from early v3 algorithms in which CH Cl 3 3 OH was also retrieved that demonstrated signicant reduction in the bias in lower stratospheric ClO. Further T renements in the v3.3 algorithms yielded not only an improved ClO product, but also a reliable x x and v3.4 ClO S MLS CH x and v3.4 Cl measurements were assessed Cl. ¿e quality and reliability of the v3.3 x retrieval of CH 3 3 T [2013]. Santee et al. in detail by S CO As is the case for ClO, the standard CH Cl product is derived from radiances measured by the radiometer 3 x CH Cl data are scientically useful over the range 147 to 4.6hPa. A centered near 640GHz. ¿e MLS v4.2 3 T GPH S summary of the estimated precision, resolution (vertical and horizontal), and systematic uncertainty of the CH x Cl measurements as a function of altitude is given in Table 3.3.1. v4.2 3 T H S 2 O x /v3.4 3.3.2 Dierences between v3.3 and v4.2 x x T /v3.4 x Changes in CH Cl between v4.2 x result from modications to upstream retrievals of tangent and v3.3 x HCl 3 S pressure etc., and small updates to the spectroscopy database for some molecules. Dierences in the retrieved T Cl abundances between the two versions range from less than 10pptv at 68hPa and lower pressures to CH HCN 3 S 50pptv at 147hPa (Figure 3.3.1), or typically within ± ~ mixing ratios generally smaller. x 10%, with v4.2 T HNO S 3.3.3 Resolution 3 T , 2000]; the two- ¿e resolution of the retrieved data can be described using “averaging kernels” [e.g., Rodgers HO S dimensional nature of the MLS data processing system means that the kernels describe both vertical and 2 horizontal resolution. Values of the integrated kernel near unity indicate that the majority of information T HOCl S for that level has come from the measurements themselves and not the a priori; Figure 3.3.2 shows that the measurementsdominateatpressuresgreaterthan3.2hPa,abovewhichleveltheintegratedkerneldropsbelow T 0.5. Smoothing, imposed on the retrieval system in both the vertical and horizontal directions to enhance IWC S retrievalstabilityandprecision, degradestheinherentresolutionofthemeasurements. ¿us,althoughCH Cl 3 T 2.7km), the ~ measurements are reported at six pressure levels per decade change in pressure (spacing of IWP S vertical resolution of the v4.2 CH Cl data as determined from the full width at half maximum of the rows of x 3 T 4–6km in most of the lower stratosphere, degrading ~ the averaging kernel matrix shown in Figure 3.3.2 is N S 2 8–10km at and above (i.e., at pressures less than) 15hPa. ¿e averaging kernels are fairly symmetric, to ~ O T and for the most part they peak at their nominal position. However, overlap in the averaging kernels for the S O 100 and 147hPa retrieval surfaces indicates that the 147hPa retrieval does not provide as much independent 3 T information as is given by retrievals at higher altitudes. Figure 3.3.2 also shows horizontal averaging kernels, OH S ~ from which the along-track horizontal resolution is determined to be 600km at 147hPa, ~ 450–500km 550–850km at and above 15hPa. ¿e cross-track resolution, set by the width of from 100 to 22hPa, and ~ T RHI S the eld of view of the 640-GHz radiometer, is ~ 3km. ¿e along-track separation between adjacent retrieved ○ 165km), whereas the longitudinal separation of MLS measurements, set by ~ great circle angle ( proles is 1.5 T ○ ○ SO S the Aura orbit, is 10 over low and middle latitudes, with much ner sampling in the polar regions. –20 2 T S T Aura Microwave Limb Sounder (MLS) x 26 Level 2 Version 4.2 Quality T

33 Help 3.3. Methyl chloride (CH Cl) 3 Overview Table BrO S T CH S 3 Cl T CH S 3 CN Averages for July, 2009 Averages for July, 2009 T Cl v4.1 CH Cl v3.3 CH 3 3 1.0 1.0 CH S 3 OH T ClO S 10.0 10.0 T S CO Pressure / hPa Pressure / hPa T GPH S T 100.0 100.0 H S o o o o o o o o o o o o 90 60 N 30 N 90 EQ S 30 S S 60 S 30 S EQ 30 N 60 N 90 N N S 60 90 2 Latitude Latitude O T 550 0 50 100 550 500 450 400 350 300 250 200 150 100 50 0 150 500 450 400 350 300 250 200 CH Cl / pptv CH Cl / pptv 3 3 HCl S Averages for July, 2009 T CH Cl Difference (v3.3-v4.1) 3 1.0 HCN S T HNO S Cl proles for a Zonal averages of MLS CH Figure 3.3.1: 3 3 T 10.0 representative month (July 2009), showing the MLS v4.2x HO S Pressure / hPa mixing ratios (upper left), the v3.3x and v3.4x mixing ra- 2 tios (upper right), and their dierences in pptv (v3.3x and T HOCl v3.4x minus v4.2x, lower left) as a function of pressure. S 100.0 T IWC o o o o o o S S 30 30 S N 60 S 60 90 N 90 EQ N Latitude T 5 35 25 45 -5 15 Cl Difference / pptv CH 3 IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 27 Level 2 Version 4.2 Quality T

34 Help 3.3. Methyl chloride (CH Cl) 3 Overview Table Equator FWHM / km FWHM / km 800 600 400 200 0 1000 1200 -2 0 2 4 6 8 10 12 0.1 BrO S T CH S 1.0 3 Cl T CH S 3 CN 10.0 T Pressure / hPa CH S 3 OH 100.0 T ClO S T 1000.0 1.0 -4 0.6 1.2 -2 0 2 0.2 -0.2 0.4 0.0 0.8 4 S CO Profile number Kernel, Integrated kernel 0 70 N T FWHM / km FWHM / km GPH 200 -2 600 800 1000 1200 0 2 4 6 8 10 12 400 0 S 0.1 T H S 2 O 1.0 T HCl S T HCN 10.0 S T Pressure / hPa HNO S 100.0 3 T HO S 2 1000.0 T HOCl -2 1.2 2 0.8 0.6 0.4 0.2 0.0 -0.2 -4 4 0 1.0 S Kernel, Integrated kernel Profile number T IWC S Typical two-dimensional (vertical and horizontal along-track) averaging kernels for the MLS v4.2x Figure 3.3.2: T ◦ N (lower); variation in the averaging kernels is suciently small that Cl data at the equator (upper) and at 70 CH 3 IWP S these are representative of typical proles. Colored lines show the averaging kernels as a function of MLS retrieval T level, indicating the region of the atmosphere from which information is contributing to the measurements on N S the individual retrieval surfaces, which are denoted by plus signs in corresponding colors. The dashed black line 2 O indicates the resolution, determined from the full width at half maximum (FWHM) of the averaging kernels, approx- T imately scaled into kilometers (top axes). (Left) Vertical averaging kernels (integrated in the horizontal dimension S O 3 for ve along-track proles) and resolution. The solid black line shows the integrated area under each kernel (hori- T zontally and vertically); values near unity imply that the majority of information for that MLS data point has come OH S from the measurements, whereas lower values imply substantial contributions from a priori information. (Right) T Horizontal averaging kernels (integrated in the vertical dimension) and resolution. The horizontal averaging ker- RHI S nels are shown scaled such that a unit averaging kernel amplitude is equivalent to a factor of 10 change in pressure. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 28 Level 2 Version 4.2 Quality T

35 Help 3.3. Methyl chloride (CH Cl) 3 Overview 0 1 21852 21865 Table 21866 21866 21866 ASC BrO Mean S 21866 Difference 10 DSC Observed 21866 T Scatter 21866 CH S Reported Precision 21866 3 Pressure / hPa Cl 21866 T 21866 CH S 21866 3 100 CN 21866 T CH 250 450 50 -50 -50 0 150 250 200 150 100 50 350 S 3 Cl / pptv Difference (ASC-DSC) / pptv CH 3 OH T (left) Ensemble mean proles for ascending (light red) and descending (dark red) orbit matching Figure 3.3.3: ClO S pairs of MLS v4.2x CH Cl proles averaged over several months of a representative year of data (2005). Symbols 3 T indicate MLS retrieval pressure levels. (right) Mean dierences (ascending descending) in pptv (green solid line). − S CO Also shown are the standard deviations about the mean dierences (light blue solid line) and the root sum square (RSS) of the precisions calculated by the retrieval algorithm for the two sets of proles (light blue dashed line). T ” GPH 1 / The observed scatter about the mean dierences and the reported precision values have been scaled by 2 (to S convert from standard deviations of dierences into standard deviations of individual data points); hence the light T blue solid line represents the statistical repeatability of the MLS measurements, and the light blue dashed line H S 2 σ 1 precision for a single prole. The thin black lines mark zero in each panel. The number represents the expected O T of crossing pairs of measurements being compared at each pressure level is noted in the space between the panels. HCl S T 3.3.4 Precision HCN S Cldataisestimated empiricallybycomparingprolesmeasuredatthe intersec- ¿eprecisionofthe MLSCH 3 T tions of ascending (day) and descending (night) portions of the orbit. Under ideal conditions (i.e., a quiescent HNO S atmosphere), the standard deviation about the mean dierences between such matched prole pairs provides 3 a measure of the precision of the individual data points. In practice, however, real changes in the atmosphere T HO S may occur over the 12h interval between the intersecting measurement points, in which case the observed 2 scatter provides an upper limit on the estimate of precision, assuming that the a priori has a negligible in- T HOCl uence on the retrieval (a reasonable assumption at least at pressures greater than 3.2hPa). ¿e precision S estimates were found to be essentially invariant with time; results for a representative year of data are shown T in Figure 3.3.3. ¿e observed standard deviation values are ~ 100pptv or less throughout the vertical domain. IWC S Mean dierences between paired crossing proles are ~ 10pptv or less except at the lowest two levels, where T descending they reach 20–25pptv. Given the large number of data points being compared, these ascending − IWP S dierences are substantially larger than the standard error of the mean, implying the presence of signicant T systematic biases. ¿ese biases likely arise from the cumulative eect of various factors in the retrieval sys- N S 2 tem that can vary diurnally or along the orbit, such as interferences from temperature or other atmospheric O T O and ClO), or thermal emissions from the MLS antenna. constituents (e.g., H 2 S O ¿e precision reported for each data point by the Level 2 data processing system exceeds the observation- 3 ally determined precision throughout the vertical range (Figure 3.3.3), indicating that the vertical smoothing T OH S (regularization) applied to stabilize the retrieval system and improve the precision has a non-negligible in- uence. Because the reported precisions take into account occasional variations in instrument performance, T RHI S the best estimate of the precision of an individual data point is the value quoted for that point in the L2GP les, but it should be borne in mind that this approach slightly overestimates the actual measurement noise. T ¿e estimates reported here represent the precisions at each pressure level of a single prole; precision can SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 29 Level 2 Version 4.2 Quality T

36 Help 3.3. Methyl chloride (CH Cl) 3 Overview √ N N times the generally be improved by averaging, with the precision of an average of proles being 1/ precision of an individual prole (although the actual standard error of the mean can in some cases be even Table smaller [ Toohey and von Clarmann , 2013]). 3.3.5 Range BrO S Cl is retrieved over the range 147 to 0.001hPa, because of the degradation in resolution and Although CH 3 T expected precision, the reduction in independent information contributed by the measurements, and the CH S results of simulations using synthetic data as input radiances to test the closure of the retrieval system, the 3 Cl data are not deemed reliable at retrieval pressures less than 4.6hPa. Despite the overlap in the averaging T CH S kernels for the 147 and 100hPa surfaces discussed above, the simulations show that the retrieved CH Cl 3 3 CN values track the variations in the “truth” eld at both levels. Moreover, the retrievals (of real data) at 147hPa T display signicant features not seen at 100hPa (not shown) that are believed to represent actual atmospheric CH S x Cl data may be used for scientic studies between 147 variations. ¿us, we recommend that the v4.2 CH 3 3 OH and 4.6hPa (inclusive), although the reduced sensitivity at the extremes of this range, as well as the relatively T coarse vertical resolution of the retrieved proles, should be borne in mind. ClO S 3.3.6 Accuracy T S CO ¿e eects of various sources of systematic uncertainty (e.g., instrumental issues, spectroscopic uncertainty, and approximations in the retrieval formulation and implementation) on the MLS v4.2 CH x Cl measure- T 3 GPH S Santee et al. ments have been quantied through a comprehensive set of retrievals of synthetic radiances; see x CH [2013] for details of a similar analysis conducted on MLS v3.3 Cl data. ¿e overall systematic x and v3.4 3 T H uncertainty, or accuracy, is calculated by combining (RSS) the contributions from both the expected biases S 2 O and the additional scatter each source of uncertainty may introduce into the data. In aggregate, the factors T considered in these simulations are estimated to give rise to total systematic uncertainty of approximately HCl S Cl data in the upper troposphere / lower stratosphere (see Table 3.3.1). x CH 30–45% in the MLS v4.2 3 T HCN S 3.3.7 Review of comparisons with other datasets T Extensive comparisons of MLS v3.3 x and v3.4 CH Cl data with a variety of dierent platforms (balloon- x 3 HNO S Santee et al. borne, aircra , and satellite instruments) were presented by [2013]. 3 T 3.3.8 Data screening HO S Pressure range: 147–4.6hPa 2 T Values outside this range are not recommended for scientic use. HOCl S Estimated precision: Only use values for which the estimated precision is a positive number. T IWC Values where the a priori information has a strong inuence are agged with negative or zero precision, S and should not be used in scientic analyses (see Section 1.5). T IWP S eld is zero. Status Status ag: Only use proles for which the T We recommend that all proles with nonzero values of be discarded, because of the potential Status N S impact of cloud artifacts at lower levels. Note, however, that rejecting in their entirety all proles with 2 O nonzero Status may be unnecessarily severe at and above (i.e., at pressures equal to or smaller than) T S O 46hPa, where clouds have negligible impact; thus otherwise good-quality proles with nonzero but 3 values may be used without restriction at those levels as long as they are removed at larger even Status T OH S pressures. See Section 1.6 for more information on the interpretation of the Status eld. T Quality: Only proles whose Quality eld is greater than 1.3 should be used. RHI S Cl Quality (unchanged from v3.3 x and v3.4 x ) typically excludes less than 1% of CH ¿is threshold for 3 T proles on a daily basis; note that it potentially discards some “good” data points while not necessarily SO S identifying all “bad” ones. 2 T S T Aura Microwave Limb Sounder (MLS) x 30 Level 2 Version 4.2 Quality T

37 Help 3.3. Methyl chloride (CH Cl) 3 Overview Convergence: Only proles whose Convergence eld is less than 1.05 should be used. ) discards very few Convergence On a typical day this threshold for x (unchanged from v3.3 and v3.4 x Table (0.3% or less) of the CH Cl proles, most (but not all) of which are also ltered out by the other quality 3 control measures. BrO S 3.3.9 Artifacts T CH − descending dierences imply the presence of systematic biases at the bottom Signicant ascending • S 3 Cl two retrieval pressure levels. However, various analyses performed separately on sets of ascending- T only and descending-only measurements suggest that, although signicant, these biases will have little CH S Cl measurements. or no impact on most scientic conclusions based on the MLS CH 3 3 CN T 3.3.10 Desired improvements for future data version(s) CH S 3 • Cl measurements in Minimize the impact of thick clouds on the retrievals to further improve the CH OH 3 the upper troposphere and lowermost stratosphere. T ClO S T S CO T GPH S T H S 2 O T HCl S T HCN S T HNO S 3 T HO S 2 T HOCl S T IWC S Cl Characteristics Summary of Aura MLS v4.2x CH Table 3.3.1: 3 T IWP S Systematic Resolution b Pressure Precision Known Artifacts c a Uncertainty H × V T / hPa / pptv or Other Comments N / % / km S 2 O 3.2–0.001 — Unsuitable for scientic use — — T 550–850 × 8–10 15–4.6 30–75 ± 100 ± S O 3 100 ± 450–500 × 100–22 4.5–6.5 ± 30–50 T 100 ± 45 147 4 × ± 600 OH S 1000–215 — — — Not retrieved T RHI S a Vertical and Horizontal resolution in along-track direction. b T Precision on individual proles. c SO S σ Values should be interpreted as 2- estimates of the probable magnitude. 2 T S T Aura Microwave Limb Sounder (MLS) x 31 Level 2 Version 4.2 Quality T

38 Help Overview 3.4 Methyl cyanide (CH CN) 3 Table Swath name: CH3CN Useful range: 46 – 1.0 hPa BrO S > [email protected] Email: Michelle Santee, Contact: < T CH S 3 Cl 3.4.1 Introduction T CH and v3.4 CN product is taken from radiances measured by the ra- x , the standard CH In v4.2 , as in v3.3 x x 3 S 3 diometer centered near 640GHz. ¿e CH CN retrieval is largely unchanged in v4.2 x . Although the data CN 3 x CN retrievals are deemed to be scientically useful over have not been validated extensively, the v4.2 CH T 3 CH S the range 46 to 1hPa, except in the winter polar regions, where they may exhibit large biases below 10hPa. 3 OH Data retrieved at higher pressures may be used with caution in certain circumstances. A summary of the T estimated precision, resolution (vertical and horizontal), and systematic uncertainty of the v4.2 x CH CN 3 ClO S measurements as a function of altitude is given in Table 3.4.1. T S CO x and v4.2 x /v3.4 x 3.4.2 Dierences between v3.3 CNbetweenv4.2 x andv3.3 x ChangesinCH /v3.4 x resultfrommodicationstoupstreamretrievalsoftangent T 3 GPH S pressure etc., and small updates to the spectroscopy database for some molecules. Dierences in the retrieved CH CNabundancesbetweenthetwoversionsaretypicallylessthan10pptv,butcanreach20pptv(stillwithin 3 T H 5%) at 147hPa, particularly at higher latitudes (Figure 3.4.1). ± S 2 O T 3.4.3 Resolution HCl S ¿e resolution of the retrieved data can be described using “averaging kernels” [e.g., , 2000]; the two- Rodgers T dimensional nature of the MLS data processing system means that the kernels describe both vertical and HCN S horizontal resolution. Values of the integrated kernel near unity indicate that the majority of information T for that level has come from the measurements themselves and not the a priori; Figure 3.4.2 shows that the HNO S measurements dominate throughout most of the vertical range. Smoothing, imposed on the retrieval system 3 T in both the vertical and horizontal directions to enhance retrieval stability and precision, degrades the in- HO S herent resolution of the measurements. ¿us, although CH CN measurements are reported at six pressure 3 2 CH CN data x 2.7km), the vertical resolution of the v4.2 ~ levels per decade change in pressure (spacing of T 3 HOCl S as determined from the full width at half maximum of the rows of the averaging kernel matrix shown in ~ 5–6km in the lower stratosphere, and then further worsens Figure 3.4.2 degrades from 4km at 147hPa to T to ~ 7–8km in the upper stratosphere. Substantial overlap in the averaging kernels for the 100 and 147hPa IWC S retrieval surfaces (which both peak at 100hPa) indicates that the 147hPa retrieval does not provide as much T independent information as is given by retrievals at higher altitudes. Figure 3.4.2 also shows horizontal av- IWP S ~ eraging kernels, from which the along-track horizontal resolution is determined to be 400–700km over T most of the vertical range. ¿e cross-track resolution, set by the width of the eld of view of the 640-GHz N S ○ 2 ~ 3km. ¿e along-track separation between adjacent retrieved proles is 1.5 great circle angle radiometer, is O ○ ○ T ~ over 165km), whereas the longitudinal separation of MLS measurements, set by the Aura orbit, is 10 ( –20 S O low and middle latitudes, with much ner sampling in the polar regions. 3 T OH S 3.4.4 Precision T ¿e precision of the MLS CH CN data is estimated empirically by comparing proles measured at the in- 3 RHI S tersections of ascending (day) and descending (night) portions of the orbit. Under ideal conditions (i.e., a quiescentatmosphere), thestandarddeviationaboutthemeandierencesbetweensuchmatchedprolepairs T SO S provides a measure of the precision of the individual data points. In practice, however, real changes in the 2 T S T Aura Microwave Limb Sounder (MLS) x 32 Level 2 Version 4.2 Quality T

39 Help 3.4. Methyl cyanide (CH CN) 3 Overview Table BrO S T CH S 3 Cl T CH S 3 CN Averages for July, 2009 Averages for July, 2009 T CN v3.3 CH CH CN v4.1 3 3 1.0 1.0 CH S 3 OH T ClO S 10.0 10.0 T S CO Pressure / hPa Pressure / hPa T GPH S T 100.0 100.0 H S o o o o o o o o o o o o 90 S N S EQ 30 60 N 60 90 90 N S 60 90 S 30 S S EQ 30 N N 60 30 N 2 Latitude Latitude O T 200 100 150 200 0 250 0 300 350 400 50 50 100 400 150 350 300 250 CN / pptv CH CN / pptv CH 3 3 HCl S Averages for July, 2009 T CN Difference (v3.3-v4.1) CH 3 1.0 HCN S T HNO S CN proles for a Zonal averages of MLS CH Figure 3.4.1: 3 3 T 10.0 representative month (July 2009), showing the MLS v4.2x HO S Pressure / hPa mixing ratios (upper left), the v3.3x and v3.4x mixing ra- 2 tios (upper right), and their dierences in pptv (v3.3x and T HOCl v3.4x minus v4.2x, lower left) as a function of pressure. S 100.0 T IWC o o o o o o S 30 N 60 90 N S 60 N S 30 EQ S 90 Latitude T -10.0 -5.0 0.0 5.0 -15.0 10.0 CN Difference / pptv CH 3 IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 33 Level 2 Version 4.2 Quality T

40 Help 3.4. Methyl cyanide (CH CN) 3 Overview Table Equator FWHM / km FWHM / km 800 600 400 200 0 1000 1200 -2 0 2 4 6 8 10 12 0.1 BrO S T CH S 1.0 3 Cl T CH S 3 CN 10.0 T Pressure / hPa CH S 3 OH 100.0 T ClO S T 1000.0 1.0 -4 0.6 1.2 -2 0 2 0.2 -0.2 0.4 0.0 0.8 4 S CO Profile number Kernel, Integrated kernel 0 70 N T FWHM / km FWHM / km GPH 200 -2 600 800 1000 1200 0 2 4 6 8 10 12 400 0 S 0.1 T H S 2 O 1.0 T HCl S T HCN 10.0 S T Pressure / hPa HNO S 100.0 3 T HO S 2 1000.0 T HOCl -2 1.2 2 0.8 0.6 0.4 0.2 0.0 -0.2 -4 4 0 1.0 S Kernel, Integrated kernel Profile number T IWC S Typical two-dimensional (vertical and horizontal along-track) averaging kernels for the MLS v4.2x Figure 3.4.2: T ◦ N (lower); variation in the averaging kernels is suciently small that CN data at the equator (upper) and at 70 CH 3 IWP S these are representative of typical proles. Colored lines show the averaging kernels as a function of MLS retrieval T level, indicating the region of the atmosphere from which information is contributing to the measurements on N S the individual retrieval surfaces, which are denoted by plus signs in corresponding colors. The dashed black line 2 O indicates the resolution, determined from the full width at half maximum (FWHM) of the averaging kernels, approx- T imately scaled into kilometers (top axes). (Left) Vertical averaging kernels (integrated in the horizontal dimension S O 3 for ve along-track proles) and resolution. The solid black line shows the integrated area under each kernel (hori- T zontally and vertically); values near unity imply that the majority of information for that MLS data point has come OH S from the measurements, whereas lower values imply substantial contributions from a priori information. (Right) T Horizontal averaging kernels (integrated in the vertical dimension) and resolution. The horizontal averaging ker- RHI S nels are shown scaled such that a unit averaging kernel amplitude is equivalent to a factor of 10 change in pressure. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 34 Level 2 Version 4.2 Quality T

41 Help 3.4. Methyl cyanide (CH CN) 3 Overview 21866 1 21866 21866 Table 21866 21866 21866 BrO Mean S 21866 Difference 10 Observed 21866 T ASC Scatter CH 21866 DSC S Reported Precision 3 21866 Cl Pressure / hPa 21866 T CH 21866 S 3 21866 100 CN 21866 T CH S 0 0 400 -100 -50 100 200 150 200 100 50 300 3 CN / pptv Difference (ASC-DSC) / pptv CH 3 OH T Figure 3.4.3: (left) Ensemble mean proles for ascending (light red) and descending (dark red) orbit matching ClO S pairs of MLS v4.2x CH CN proles averaged over several months of a representative year of data (2005). Symbols 3 T indicate MLS retrieval pressure levels. (right) Mean dierences (ascending − descending) in pptv (green solid line). S CO Also shown are the standard deviations about the mean dierences (light blue solid line) and the root sum square (RSS) of the precisions calculated by the retrieval algorithm for the two sets of proles (light blue dashed line). T ” GPH S 2 (to The observed scatter about the mean dierences and the reported precision values have been scaled by 1/ convert from standard deviations of dierences into standard deviations of individual data points); hence the light T blue solid line represents the statistical repeatability of the MLS measurements, and the light blue dashed line H S 2 represents the expected precision for a single prole. The thin black lines mark zero in each panel. The number σ 1 O T of crossing pairs of measurements being compared at each pressure level is noted in the space between the panels. HCl S T atmosphere may occur over the 12h interval between the intersecting measurement points, in which case the HCN S observed scatter provides an upper limit on the estimate of precision, assuming that the a priori has a neg- T ligible inuence on the retrieval (a reasonable assumption throughout the retrieval range for CH CN). ¿e 3 HNO S precision estimates were found to be essentially invariant with time; results for a representative year of data 3 50pptv or less throughout most of the ~ are shown in Figure 3.4.3. ¿e observed standard deviation values are T HO S vertical domain, increasing to 100pptv at 147 and 1hPa. Mean dierences between paired crossing proles 2 are negligible, indicating the absence of signicant ascending − descending osets. T HOCl ¿e precision reported for each data point by the Level 2 data processing system exceeds the observation- S ally determined precision throughout the vertical range (Figure 3.4.3), indicating that the vertical smoothing T (regularization) applied to stabilize the retrieval system and improve the precision has a non-negligible in- IWC S uence. Because the reported precisions take into account occasional variations in instrument performance, T the best estimate of the precision of an individual data point is the value quoted for that point in the L2GP IWP S les, but it should be borne in mind that this approach slightly overestimates the actual measurement noise. T ¿e estimates reported here represent the precisions at each pressure level of a single prole; precision can √ N S 2 proles being 1/ N generally be improved by averaging, with the precision of an average of times the N O T precision of an individual prole (although the actual standard error of the mean can in some cases be even S O , 2013]). smaller [ Toohey and von Clarmann 3 T 3.4.5 Range OH S CN is retrieved (and reported in the L2GP les) over the range 147 to 0.001hPa, on the basis Although CH 3 T RHI S of the drop o in precision and resolution, the lack of independent information contributed by the measure- ments, and the results of simulations using synthetic data as input radiances to test the closure of the retrieval T x system, the data are not deemed reliable at the extremes of the retrieval range. ¿us we recommend that v4.2 SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 35 Level 2 Version 4.2 Quality T

42 Help 3.4. Methyl cyanide (CH CN) 3 Overview CH CN be used for scientic studies only at the levels between 46 and 1hPa, except in the winter polar re- 3 gions, where they may exhibit large biases below 10hPa. However, although the 147, 100, and 68hPa retrievals Table are not generally recommended, they may be scientically useful in some circumstances. For example, the ○ ± data display unphysical sharp latitudinal gradients at 30 at 100 and 68hPa, yet the large-scale longitudi- nal variations within the tropics are probably robust. Similarly, conned regions of signicant enhancement BrO S at 147hPa unaccompanied by comparably enhanced values at 100hPa may reect real atmospheric features. T Indeed, many of the “hotspots” apparent in MLS CH CN measurements in the upper troposphere / lower CH 3 S 3 stratosphere closely track similar enhancements in other pollution markers measured by MLS, such as CO Cl T and CH CN data at the lowest retrieval levels (147–68hPa) should only be used in con- Cl. ¿e v4.2 CH x 3 3 CH S sultation with the MLS science team. 3 CN T 3.4.6 Accuracy CH S ¿e eects of various sources of systematic uncertainty (e.g., instrumental issues, spectroscopic uncertainty, 3 OH CN measure- CH x and approximations in the retrieval formulation and implementation) on the MLS v4.2 3 T ments have been quantied through a comprehensive set of retrievals of synthetic radiances. ¿e overall ClO S systematic uncertainty, or accuracy, is calculated by combining (RSS) the contributions from both the ex- T pected biases and the additional scatter each source of uncertainty may introduce into the data. In aggregate, S CO the factors considered in these simulations are estimated to give rise to total systematic uncertainty of ap- T x proximately 100–200% in the MLS v4.2 CN data (see Table 3.4.1). CH 3 GPH S T 3.4.7 Review of comparisons with other datasets H S 2 Detailed quantication of dierences from correlative data sets has not been performed, but preliminary O T CN measurements are biased substantially high in the UTLS relative comparisons suggest that the MLS CH 3 HCl S , 2003], balloon-borne [ Singh et al. Kleinböhl et al. , 2005], and ACE-FTS satellite [ Harrison and to airborne [ CN measurements, as do results from a two-dimensional chemistry transport model and , 2013] CH Bernath T 3 HCN S (noncoincident)CH CNretrievalsfromthepredecessorMLSinstrumentontheUpperAtmosphereResearch 3 Satellite (UARS) [ , 2001] (not shown). Furthermore, the zonal-mean morphology of the Aura Livesey et al. T HNO S MLS CH CN at the lowest levels does not agree well with that either observed by UARS MLS or predicted by 3 the model. 3 T HO S 3.4.8 Data screening 2 T Pressure range: 46–1.0hPa HOCl S Values outside this range are not recommended for scientic use. ¿e CH CN data at 147–68hPa 3 T may be useful under certain circumstances but should not be analyzed in scientic studies without IWC S signicant discussion with the MLS science team. T Estimated precision: Only use values for which the estimated precision is a positive number. IWP S Values where the information has a strong inuence are agged with negative or zero precision, a priori T and should not be used in scientic analyses (see Section 1.5). N S 2 O Status ag: Only use proles for which the Status eld is zero. T S We recommend that all proles with nonzero values of Status be discarded, because of the potential O 3 impact of cloud artifacts at lower levels. Note, however, that rejecting in their entirety all proles with T Status nonzero may be unnecessarily severe at and above (i.e., at pressures equal to or smaller than) OH S 46hPa, where clouds have negligible impact; thus otherwise good-quality proles with nonzero but T Status values may be used without restriction at those levels as long as they are removed at larger even RHI S eld. Status pressures. See Section 1.6 for more information on the interpretation of the T SO S Quality Quality: Only proles whose eld is greater than 1.4 should be used. 2 T S T Aura Microwave Limb Sounder (MLS) x 36 Level 2 Version 4.2 Quality T

43 Help 3.4. Methyl cyanide (CH CN) 3 Overview x Quality CN x ¿isthresholdfor andv3.4 (unchangedfromv3.3 )typicallyexcludeslessthan1%ofCH 3 proles on a daily basis; note that it potentially discards some “good” data points while not necessarily Table identifying all “bad” ones. Convergence eld is less than 1.05 should be used. Convergence: Only proles whose BrO S and v3.4 Convergence On a typical day this threshold for ) discards very few (unchanged from v3.3 x x T CN proles, many (but not all) of which are ltered out by the other quality (0.3% or less) of the CH 3 CH S control measures. 3 Cl T 3.4.9 Artifacts CH S ○ 3 . ¿e retrievals at 100 and 68hPa are characterized by unphysical sharp latitudinal gradients at ± • 30 CN T • Substantial biases may be present in the mixing ratios in the winter polar regions for retrieval levels in CH S the range 100–15hPa. 3 OH T 3.4.10 Desired improvements for future data version(s) ClO S Improve the CH CN retrievals at 147–68hPa. • 3 T S CO T GPH S T H S 2 O T HCl S T HCN S T HNO S 3 T HO S 2 T HOCl S CN Characteristics Summary of Aura MLS v4.2x CH Table 3.4.1: 3 T IWC S Systematic Resolution b Pressure Precision Known Artifacts c a Uncertainty × H V T / hPa / pptv or Other Comments IWP S / % / km Unsuitable for scientic use — — 0.68–0.001 — T N S 1.0 Consult with MLS science team 200 ± 100 ± 800 × 6 2 O × 100–200 Consult with MLS science team 7–8 10–1.5 ± 50 ± 400–700 T 100–200 Consult with MLS science team 400–500 ± 50 ± × 46–15 5–6.5 S O 3 × ± 50 ± 100–68 5.5 100 550–700 Consult with MLS science team T ± 100 ± 200 Consult with MLS science team 147 4 × 750 OH S 1000–215 — — — Not retrieved T RHI S a Vertical and Horizontal resolution in along-track direction. b T Precision on individual proles. c SO S estimates of the probable magnitude. Values should be interpreted as 2- σ 2 T S T Aura Microwave Limb Sounder (MLS) x 37 Level 2 Version 4.2 Quality T

44 Help Overview 3.5 Methanol (CH OH) 3 Table Swath name: CH3OH 147 – 100 hPa Useful range: BrO S > Contact: < Michelle Santee, [email protected] Email: T CH S 3 Cl 3.5.1 Introduction T OH product is taken from radiances measured by the radiometer centered near 640GHz. ¿e standard CH 3 CH S 3 However, as the methanol spectral signature in this region is very similar to that of ClO, additional mea- CN surements from the 190-GHz radiometer (which has channels sensitive to ClO but not CH OH) are used to 3 T CH decouple the CH , and it has not been fully OH and ClO information. CH x OH is a new product in v4.2 3 3 S 3 validated. ¿us the scientic utility of the v4.2 OH measurements remains to be determined. A x MLS CH OH 3 summary of the estimated precision, resolution (vertical and horizontal), and systematic uncertainty of the T ClO S OH measurements as a function of altitude is given in Table 3.5.1. v4.2 CH x 3 T 3.5.2 Resolution S CO , 2000]; the two- ¿e resolution of the retrieved data can be described using “averaging kernels” [e.g., Rodgers T dimensional nature of the MLS data processing system means that the kernels describe both vertical and GPH S horizontal resolution. Values of the integrated kernel near unity indicate that the majority of information for T that level has come from the measurements themselves and not the a priori; Figure 3.5.1 shows that the mea- H S 2 surements dominate throughout most of the vertical range. Smoothing, imposed on the retrieval system in O T both the vertical and horizontal directions to enhance retrieval stability and precision, degrades the inherent HCl S resolution of the measurements. ¿us, although CH OH measurements are reported at six pressure levels per 3 T ~ CH 2.7km), the vertical resolution of the v4.2 OH data as deter- decade change in pressure (spacing of x 3 HCN S mined from the full width at half maximum of the rows of the averaging kernel matrix shown in Figure 3.5.1 is ~ 4–5km at 147 and 100hPa. Figure 3.5.1 also shows horizontal averaging kernels, from which the along-track T HNO S horizontal resolution is determined to be 350km. ¿e cross-track resolution, set by the width of the eld of ~ view of the 640-GHz radiometer, is ~ 3km. ¿e along-track separation between adjacent retrieved proles is 3 T ○ 1.5 great circle angle ( ~ 165km), whereas the longitudinal separation of MLS measurements, set by the Aura HO S ○ ○ –20 orbit, is 10 over low and middle latitudes, with much ner sampling in the polar regions. 2 T HOCl S 3.5.3 Precision OH data is estimated empirically by comparing proles measured at the in- ¿e precision of the MLS CH 3 T IWC S tersections of ascending (day) and descending (night) portions of the orbit. Under ideal conditions (i.e., a quiescent atmosphere), the standard deviation about the mean dierences between such matched prole T IWP pairs provides a measure of the precision of the individual data points. In practice, however, real changes S in the atmosphere may occur over the 12h interval between the intersecting measurement points, in which T case the observed scatter provides an upper limit on the estimate of precision, assuming that the a priori has N S 2 O a negligible inuence on the retrieval (a reasonable assumption throughout the retrieval range for CH OH). 3 T ¿e precision estimates were found to be essentially invariant with time; results for several months of a rep- S O 3 resentative year of data are shown in Figure 3.5.2. For the most part, dierences between paired proles are T small, implying the absence of signicant systematic ascending / descending biases. ¿e observed standard OH S deviation is 1ppbv at 147hPa. ¿e observationally determined precision agrees well with that reported for ~ T each data point by the Level 2 data processing system. ¿e estimates reported here represent the precisions at RHI S each pressure level of a single prole; precision can generally be improved by averaging, with the precision of √ T proles being 1/ N an average of times the precision of an individual prole (although the actual standard N SO S Toohey and von Clarmann error of the mean can in some cases be even smaller [ , 2013]). 2 T S T Aura Microwave Limb Sounder (MLS) x 38 Level 2 Version 4.2 Quality T

45 Help 3.5. Methanol (CH OH) 3 Overview Table Equator FWHM / km FWHM / km 800 600 400 200 0 1000 1200 -2 0 2 4 6 8 10 12 0.1 BrO S T CH S 1.0 3 Cl T CH S 3 CN 10.0 T Pressure / hPa CH S 3 OH 100.0 T ClO S T 1000.0 1.0 -4 0.6 1.2 -2 0 2 0.2 -0.2 0.4 0.0 0.8 4 S CO Profile number Kernel, Integrated kernel 0 70 N T FWHM / km FWHM / km GPH 200 -2 600 800 1000 1200 0 2 4 6 8 10 12 400 0 S 0.1 T H S 2 O 1.0 T HCl S T HCN 10.0 S T Pressure / hPa HNO S 100.0 3 T HO S 2 1000.0 T HOCl -2 1.2 2 0.8 0.6 0.4 0.2 0.0 -0.2 -4 4 0 1.0 S Kernel, Integrated kernel Profile number T IWC S Typical two-dimensional (vertical and horizontal along-track) averaging kernels for the MLS v4.2x Figure 3.5.1: T ◦ N (lower); variation in the averaging kernels is suciently small that OH data at the equator (upper) and at 70 CH 3 IWP S these are representative of typical proles. Colored lines show the averaging kernels as a function of MLS retrieval T level, indicating the region of the atmosphere from which information is contributing to the measurements on N S the individual retrieval surfaces, which are denoted by plus signs in corresponding colors. The dashed black line 2 O indicates the resolution, determined from the full width at half maximum (FWHM) of the averaging kernels, approx- T imately scaled into kilometers (top axes). (Left) Vertical averaging kernels (integrated in the horizontal dimension S O 3 for ve along-track proles) and resolution. The solid black line shows the integrated area under each kernel (hori- T zontally and vertically); values near unity imply that the majority of information for that MLS data point has come OH S from the measurements, whereas lower values imply substantial contributions from a priori information. (Right) T Horizontal averaging kernels (integrated in the vertical dimension) and resolution. The horizontal averaging ker- RHI S nels are shown scaled such that a unit averaging kernel amplitude is equivalent to a factor of 10 change in pressure. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 39 Level 2 Version 4.2 Quality T

46 Help 3.5. Methanol (CH OH) 3 Overview 20319 1 20319 20319 Table 20319 20319 20319 BrO S Mean 20319 Difference 10 T Observed 20319 ASC Scatter CH S 20319 DSC Reported 3 Precision Cl 20319 Pressure / hPa T 20319 CH S 20319 3 CN 20319 100 20319 T CH S -0.5 0.5 -1.0 1.0 0.0 1.5 1.0 0.5 -0.5 0.0 3 OH Difference (ASC-DSC) / ppbv CH OH / ppbv 3 T ClO S Figure 3.5.2: (left) Ensemble mean proles for ascending (light red) and descending (dark red) orbit matching pairs of MLS v4.2x CH OH proles averaged over several months of a representative year of data (2005). Symbols 3 T indicate MLS retrieval pressure levels. (right) Mean dierences (ascending − descending) in pptv (green solid line). S CO Also shown are the standard deviations about the mean dierences (light blue solid line) and the root sum square T (RSS) of the precisions calculated by the retrieval algorithm for the two sets of proles (light blue dashed line). ” GPH S The observed scatter about the mean dierences and the reported precision values have been scaled by 1/ (to 2 convert from standard deviations of dierences into standard deviations of individual data points); hence the light T H blue solid line represents the statistical repeatability of the MLS measurements, and the light blue dashed line S 2 O σ precision for a single prole. The thin black lines mark zero in each panel. The number 1 represents the expected T of crossing pairs of measurements being compared at each pressure level is noted in the space between the panels. HCl S T HCN 3.5.4 Range S OH is retrieved (and reported in the L2GP les) over the range 147 to 0.001hPa. However, zonal mean CH 3 T HNO mixing ratios are negative everywhere at retrieval pressures less than 100hPa, as well as at middle and high S latitudes at 100hPa. ¿us, with rare exceptions (such as during extreme events), the v4 CH OH data are 3 3 T considered potentially useful for scientic studies only at 147hPa and at low latitudes at 100hPa. HO S 2 T 3.5.5 Accuracy HOCl S ¿e eects of various sources of systematic uncertainty (e.g., instrumental issues, spectroscopic uncertainty, OH measure- x CH and approximations in the retrieval formulation and implementation) on the MLS v4.2 3 T IWC S ments have been quantied through a comprehensive set of retrievals of synthetic radiances. ¿e overall systematic uncertainty, or accuracy, is calculated by combining (RSS) the contributions from both the ex- T IWP pected biases and the additional scatter each source of uncertainty may introduce into the data. In aggregate, S the factors considered in these simulations are estimated to give rise to total systematic uncertainty of ap- T CH x proximately 100–250% in the MLS v4.2 OH data (see Table 3.5.1). N 3 S 2 O T 3.5.6 Review of comparisons with other datasets S O 3 Detailed comparisons with correlative data sets have not been undertaken, but preliminary comparisons with T version 3 ACE-FTS data suggest that the MLS values may be biased substantially high at 147hPa (not shown). OH S T 3.5.7 Data screening RHI S Do not use: x CH OH data should only be used in consultation with the MLS science team. ¿e v4.2 3 T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 40 Level 2 Version 4.2 Quality T

47 Help 3.5. Methanol (CH OH) 3 Overview 3.5.8 Artifacts • OH measurements has not been performed, but x A complete assessment of artifacts in the v4.2 CH 3 Table it is known that zonal mean mixing ratios are negative everywhere at retrieval pressures less than or equal to 68hPa, as well as at middle and high latitudes at 100hPa. BrO S 3.5.9 Desired improvements for future data version(s) T CH OH retrieval remain to be determined. Specic improvements in the CH • S 3 3 Cl T CH S 3 CN T CH S 3 OH T ClO S T S CO T GPH S T H S 2 O T HCl S T HCN S T HNO S 3 T HO S 2 T HOCl S T IWC S T OH Characteristics Summary of Aura MLS v4.2x CH Table 3.5.1: IWP 3 S Systematic Resolution b T Pressure Precision Known Artifacts c a N Uncertainty × V H S 2 / hPa / ppbv or Other Comments O / % / km T — 68–0.001 — — Unsuitable for scientic use S O 3 ± 1 ± × Consult with MLS Science Team 5 100 350 100 T 3 × 147 150 250 Consult with MLS Science Team ± ± 1 OH S Not retrieved 1000–215 — — — T RHI S a Vertical and Horizontal resolution in along-track direction. b T Precision on individual proles. c SO S Values should be interpreted as 2- estimates of the probable magnitude. σ 2 T S T Aura Microwave Limb Sounder (MLS) x 41 Level 2 Version 4.2 Quality T

48 Help Overview 3.6 Chlorine Monoxide (ClO) Table ClO Swath name: Useful range: 147 – 1.0 hPa BrO S Michelle Santee, [email protected] > < Email: Contact: T CH S 3 Cl 3.6.1 Introduction T CH ¿e quality and reliability of the version 2 (v2.2) Aura MLS ClO measurements were assessed in detail by S 3 [ Livesey et al. , 2013]; in x Santee et al. x and v3.4 [2008]. ¿e ClO product was signicantly improved in v3.3 CN 0.1–0.4ppbv) negative bias present in the v2.2 ClO values at retrieval levels below ~ particular, the substantial ( T CH S Cl, which (i.e., pressureslargerthan)22hPawasmitigatedtoalargeextent, primarilythroughretrievalofCH 3 3 OH was a new MLS product in v3.3 x and v3.4 x . ¿e ClO retrieval is largely unchanged over much of the prole T in v4.2 . x ClO S the standard ClO product is derived from radiances measured by the As in previous versions, in v4.2 x T radiometer centered near 640GHz. (ClO is also retrieved using radiances from the 190-GHz radiometer, but S CO thosedatahavepoorerprecision.) ¿eMLSv4.2 ClOdataarescienticallyusefulovertherange147to1hPa. x A summary of the estimated precision, resolution (vertical and horizontal), and systematic uncertainty of the T GPH S ClO measurements as a function of altitude is given in Table 3.6.1. v4.2 x T 3.6.2 Resolution H S 2 O Rodgers ¿e resolution of the retrieved data can be described using “averaging kernels” [e.g., , 2000]; the two- T dimensional nature of the MLS data processing system means that the kernels describe both vertical and HCl S horizontal resolution. Smoothing, imposed on the retrieval system in both the vertical and horizontal di- T rections to enhance retrieval stability and precision, degrades the inherent resolution of the measurements. HCN S ¿us, although ClO measurements are reported at six pressure levels per decade change in pressure (spacing T x of ~ 2.7km), the vertical resolution of the v4.2 ClO data as determined from the full width at half maximum HNO S 3–4.5km throughout the retrieval range ~ of the rows of the averaging kernel matrix shown in Figure 3.6.1 is 3 x ~ (with a mean of 3.5km). As in v3.3 x x the averaging kernels are sharply peaked at all levels, and v3.4 , in v4.2 T HO S including 147hPa. ¿us, although some degree of overlap is present, the 147hPa surface provides indepen- 2 . Figure 3.6.1 also shows horizontal averaging kernels, from which the along-track x dent information in v4.2 T HOCl S 250–500km over most of the vertical range. ¿e cross-track reso- ~ horizontal resolution is determined to be ~ 3km. ¿e along-track separation lution, set by the width of the eld of view of the 640-GHz radiometer, is T ○ great circle angle ( ~ 165km), whereas the longitudinal separation of between adjacent retrieved proles is 1.5 IWC S ○ ○ over low and middle latitudes, with much ner sampling –20 MLS measurements, set by the Aura orbit, is 10 T in the polar regions. IWP S T 3.6.3 Precision N S 2 ¿e precision of the MLS ClO measurements is estimated empirically by computing the standard deviation O ○ T -wide latitude band centered around the equator. For of the descending (i.e., nighttime) proles in the 20 S O this region and time of day, natural atmospheric variability should be negligible relative to the measurement 3 T noise. As shown in Figure 3.6.2, the observed scatter in the data is essentially unchanged in v4.2 x , rising from OH S ~ ~ 0.3ppbv at 1hPa and 147hPa. ¿e smoothing of the retrieval is 0.1ppbv over the interval 100 – 3hPa to turned o above 1hPa, and as a consequence the precision rises steeply above this level. ¿e scatter in the data T RHI S is essentially invariant with time, as seen by comparing the results for the dierent days shown in Figure 3.6.2. ¿e single-prole precision estimates cited here are, to rst order, independent of latitude and season, T SO S but of course the scientic utility of individual MLS proles (i.e., signal to noise) varies with ClO abundance. 2 T S T Aura Microwave Limb Sounder (MLS) x 42 Level 2 Version 4.2 Quality T

49 Help 3.6. Chlorine Monoxide (ClO) Overview Table Equator FWHM / km FWHM / km 800 600 400 200 0 1200 1000 12 -2 0 2 4 6 8 10 0.01 BrO S T CH 0.10 S 3 Cl T CH 1.00 S 3 CN T 10.00 Pressure / hPa CH S 3 OH 100.00 T ClO S T 1000.00 0.8 -4 1.0 0.2 1.2 -2 0 0.0 2 -0.2 0.4 0.6 4 S CO Profile number Kernel, Integrated kernel 0 70 N T FWHM / km FWHM / km GPH 0 200 -2 600 800 1000 1200 0 2 4 6 8 10 12 400 S 0.01 T H S 2 O 0.10 T HCl S 1.00 T HCN S 10.00 T Pressure / hPa HNO S 3 100.00 T HO S 2 1000.00 T HOCl 4 0.8 2 0.4 0.2 0.0 -0.2 -2 0 -4 1.2 1.0 0.6 S Profile number Kernel, Integrated kernel T IWC S Typical two-dimensional (vertical and horizontal along-track) averaging kernels for the MLS v4.2x ClO Figure 3.6.1: T ◦ N (lower); variation in the averaging kernels is suciently small that these data at the equator (upper) and at 70 IWP S are representative of typical proles. Colored lines show the averaging kernels as a function of MLS retrieval level, T indicating the region of the atmosphere from which information is contributing to the measurements on the indi- N S vidual retrieval surfaces, which are denoted by plus signs in corresponding colors. The dashed black line indicates 2 O the resolution, determined from the full width at half maximum (FWHM) of the averaging kernels, approximately T scaled into kilometers (top axes). (Left) Vertical averaging kernels (integrated in the horizontal dimension for ve S O 3 along-track proles) and resolution. The solid black line shows the integrated area under each kernel (horizontally T and vertically); values near unity imply that the majority of information for that MLS data point has come from the OH S measurements, whereas lower values imply substantial contributions from a priori information. (Right) Horizon- T tal averaging kernels (integrated in the vertical dimension) and resolution. The horizontal averaging kernels are RHI S shown scaled such that a unit averaging kernel amplitude is equivalent to a factor of 10 change in pressure. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 43 Level 2 Version 4.2 Quality T

50 Help 3.6. Chlorine Monoxide (ClO) Overview Table 1 BrO S T CH S 10 v4 v3.3 / v3.4 3 Cl 2009d015 15 Jan 2009 T Pressure / hPa 15 Apr 2009 2009d105 CH S 15 Jul 2009 2009d196 3 CN 2009d288 15 Oct 2009 T 100 CH S 3 OH 0.4 0.6 0.8 1.0 0.6 1.0 0.8 0.2 0.0 0.2 0.4 T ClO Precision / ppbv ClO Precision / ppbv ClO S Precision of the (left) v4.2x and (right) v3.3x and v3.4x MLS ClO measurements for four representative Figure 3.6.2: T days in dierent seasons (see legend). Solid lines depict the observed scatter in nighttime-only measurements S CO obtained in a narrow equatorial band (see text); dotted lines depict the theoretical precision estimated by the T retrieval algorithm. GPH S T Outside of the lower stratospheric winter polar vortices, within which ClO is o en strongly enhanced, the H S 2 O single-prole precision exceeds typical ClO mixing ratios, necessitating the use of averages for scientic stud- T ies. Precision can generally be improved by averaging, with the precision of an average of proles being N HCl √ S 1/ times the precision of an individual prole (although the actual standard error of the mean can in some N T Toohey and von Clarmann , 2013]). cases be even smaller [ HCN S ¿e observational determination of the precision is compared in Figure 3.6.2 to the theoretical precision T values reported by the Level 2 data processing algorithms. ¿e predicted precision exceeds the observed HNO S scatter, particularly above 15hPa, indicating that the vertical smoothing (regularization) applied to stabilize 3 the retrieval and improve the precision has a nonnegligible inuence on the results at these levels. Because the T HO theoretical precisions take into account occasional variations in instrument performance, the best estimate S 2 of the precision of an individual data point is the value quoted for that point in the L2GP les, but it should T HOCl be borne in mind that this approach slightly overestimates the actual measurement noise. S T 3.6.4 Accuracy IWC S ¿e eects of various sources of systematic uncertainty (e.g., instrumental issues, spectroscopic uncertainty, T and approximations in the retrieval formulation and implementation) on the MLS v4.2 ClO measurements x IWP S Santee et al. [2008] have been quantied through a comprehensive set of retrievals of synthetic radiances; see T for details of a similar analysis conducted on MLS v2.2 ClO data. ¿e overall systematic uncertainty, or N S 2 accuracy, iscalculatedbycombining(RSS)thecontributionsfromboththeexpectedbiasesandtheadditional O T scatter each source of uncertainty may introduce into the data. In aggregate, the factors considered in these S O simulations are estimated to give rise to a total systematic uncertainty ranging from approximately 0.02 to 3 ClO data (see Table 3.6.1). x 0.4ppbv, depending on the level, in the MLS v4.2 T OH S x Dierencesbetweenv4.2 andv3.3 x andv3.4 x ClOmixingratiosaregenerallylessthan0.04ppbv(usually considerably so) at and above (i.e., at pressures lower than) 68hPa, including around the secondary peak in T the ClO prole in the upper stratosphere (Figure 3.6.3). Signicant dierences (as much as 0.1ppbv) are seen RHI S at the lowest retrieval levels, however. Figure 3.6.4 depicts nighttime mixing ratios for a single representative T day in Southern Hemisphere late spring / Northern Hemisphere early autumn for which ClO is not enhanced SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 44 Level 2 Version 4.2 Quality T

51 Help 3.6. Chlorine Monoxide (ClO) Overview Table BrO S T CH S 3 Cl T CH S 3 CN Averages for April, 2009 Averages for April, 2009 T ClO v3.3 ClO v4.1 CH S 3 OH 1.0 1.0 T ClO S T 10.0 10.0 S CO Pressure / hPa Pressure / hPa T GPH S T 100.0 100.0 H S o o o o o o o o o o o o EQ 30 90 S 90 S 60 N S 30 60 S EQ 30 30 N 60 N N 90 90 N S 60 N S 2 Latitude Latitude O T 0.3 0.4 0.5 0.6 0.7 0.7 -0.1 0.0 0.1 0.6 -0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 ClO / ppbv ClO / ppbv HCl S Averages for April, 2009 T ClO Difference (v3.3-v4.1) HCN S 1.0 T Figure 3.6.3: Zonal averages of all MLS ClO proles (day- HNO S time and nighttime) from the month of April 2009, show- 3 ing the MLS v4.2x mixing ratios (upper left), the v3.3x and T 10.0 HO v3.4x mixing ratios (upper right), and their dierences in S Pressure / hPa 2 ppbv (v3.3x and v3.4x minus v4.2x, lower left) as a func- T tion of pressure (no bias corrections have been applied HOCl S here). The month shown represents typical conditions during the Northern Hemisphere late spring / Southern T 100.0 IWC Hemisphere early autumn, when ClO is not enhanced in o o o o o o S 90 N 60 N 90 30 EQ N 30 S S 60 S Latitude the polar lower stratosphere in either hemisphere. T -0.06 -0.02 0.02 0.06 0.10 ClO Difference / ppbv IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 45 Level 2 Version 4.2 Quality T

52 Help 3.6. Chlorine Monoxide (ClO) Overview 5-Nov-2009 -- 2009d309 0.8 0.8 31 hPa 21 hPa v4 v4 Table 0.6 0.6 v3.3 / v3.4 v3.3 / v3.4 0.4 0.4 0.2 0.2 0.0 0.0 BrO S -0.2 -0.2 Nighttime ClO / pbbv Nighttime ClO / pbbv -0.4 -0.4 T 0 -50 50 0 -50 50 CH S 0.8 0.8 46 hPa v4 68 hPa v4 3 0.6 0.6 v3.3 / v3.4 v3.3 / v3.4 Cl 0.4 0.4 T CH 0.2 0.2 S 0.0 0.0 3 CN -0.2 -0.2 Nighttime ClO / pbbv Nighttime ClO / pbbv T -0.4 -0.4 50 -50 0 0 -50 50 CH S 0.8 0.8 146 hPa v4 100 hPa v4 3 OH 0.6 0.6 v3.3 / v3.4 v3.3 / v3.4 0.4 0.4 T 0.2 0.2 ClO S 0.0 0.0 -0.2 -0.2 T Nighttime ClO / pbbv Nighttime ClO / pbbv -0.4 -0.4 S CO 0 0 50 50 -50 -50 Latitude / deg Latitude / deg T GPH S Figure 3.6.4: Nighttime v4.2x (red) and v3.3x and v3.4x (black) MLS ClO data as a function of latitude for the six T lowest retrieval pressure surfaces (21–147 hPa). The date shown is representative of a typical Southern Hemisphere H S late spring / Northern Hemisphere early autumn day for which ClO is not enhanced in the polar lower stratosphere 2 O in either hemisphere. T HCl S T in the polar lower stratosphere in either hemisphere; a similar plot for a corresponding day in Northern HCN S Hemisphere late spring / Southern Hemisphere early autumn gives very similar results (not shown). ¿is gure demonstrates that a small, somewhat latitude-dependent, negative bias is still evident at 68 and 100hPa T HNO S x x in v4.2 x , of approximately the same magnitude as in v3.3 and v3.4 at 68hPa but a little larger at 100hPa, ClO displays a strongly latitudinally x x especially in the Northern Hemisphere. At 147hPa, v3.3 and v3.4 3 T varying bias that is positive at middle and low latitudes and slightly negative in the polar regions (at least HO S x , on some days). In v4.2 x , ClO mixing ratios at 147hPa are slightly lower than they were in v3.3 x and v3.4 2 T and consequently the low-latitude positive bias has been reduced, although it remains substantial (exceeding HOCl S 0.7ppbv in some cases), and the high-latitude negative bias has been exacerbated. T In many cases the bias can be essentially eliminated by subtracting daily gridded or zonal-mean night- IWC S time values from the individual daytime measurements. ¿is is not a practical approach under conditions of continuous daylight in the summer or continuous darkness in the winter at high latitudes, however. More- T IWP S over, under certain circumstances inside the winter polar vortices, chlorine activation leads to nonnegligible − ClO abundances even at night. In this case, taking day night dierences considerably reduces the apparent T N degree of chlorine activation. It is instead recommended that the estimated value of the bias be subtracted S 2 O from the individual measurements at each aected retrieval level. T x MLSClOdata, weshowin Toinvestigatethemagnitudeofandtemporalvariationsinthebiasinthev4.2 S O 3 Figure 3.6.5 monthly zonal means of MLS nighttime ClO measurements for pressure levels 147–46hPa. Each T panel represents a calendar month; data taken during that month over the course of the Aura mission to date OH S ○ ○ . Figure 3.6.6 is a similar ± (2005–2016) have been binned and averaged in 5 -wide latitude bands between 85 T plot but encompasses all of the MLS nighttime ClO data for pressure levels 147–10hPa. Although Figure 3.6.5 RHI S ~ reveals slight month-to-month variability, it is relatively small (ranging from less than 0.04ppbv at 147hPa ○ T to less than 50 0.01ppbv at 68hPa) at ~ ± (the highest latitudes for which month-to-month variability can be SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 46 Level 2 Version 4.2 Quality T

53 Help 3.6. Chlorine Monoxide (ClO) Overview Jan Feb 146 0.6 0.6 Table 100 0.4 0.4 68 46 0.2 0.2 BrO S 0.0 0.0 v4 ClO bias / ppbv v4 ClO bias / ppbv T -0.2 -0.2 CH S -50 50 70 90 -90 10 30 50 70 90 -90 -70 -50 -30 -10 10 -30 -70 30 -10 3 Cl Apr Mar 0.6 0.6 T CH S 0.4 0.4 3 CN 0.2 0.2 T 0.0 0.0 CH S v4 ClO bias / ppbv v4 ClO bias / ppbv 3 -0.2 -0.2 OH 10 50 70 90 70 -90 -70 -50 90 -10 -30 50 10 30 30 -90 -70 -30 -50 -10 T Jun May ClO S 0.6 0.6 0.4 0.4 T S CO 0.2 0.2 T 0.0 0.0 v4 ClO bias / ppbv v4 ClO bias / ppbv GPH S -0.2 -0.2 30 50 70 50 70 90 90 -90 -70 -50 -30 -10 10 30 -90 -70 -50 -30 -10 10 T H S Jul Aug 0.6 0.6 2 O T 0.4 0.4 HCl S 0.2 0.2 T 0.0 0.0 HCN v4 ClO bias / ppbv v4 ClO bias / ppbv S -0.2 -0.2 30 50 70 -90 -70 90 70 50 30 10 -10 90 -50 -30 -30 -10 -50 -70 -90 10 T HNO S Oct Sep 0.6 0.6 3 0.4 0.4 T HO S 0.2 0.2 2 T 0.0 0.0 v4 ClO bias / ppbv v4 ClO bias / ppbv HOCl S -0.2 -0.2 10 -70 -50 -30 -10 10 30 50 70 90 90 30 -10 70 -90 -90 -70 -50 -30 50 T IWC Nov Dec S 0.6 0.6 0.4 0.4 T IWP S 0.2 0.2 T 0.0 0.0 v4 ClO bias / ppbv v4 ClO bias / ppbv N S 2 -0.2 -0.2 O -90 -90 70 90 30 10 50 -10 -30 -50 -70 90 70 50 30 10 -10 -30 -50 -70 T Latitude / degrees Latitude / degrees S O 3 ◦ -wide geographic latitude bands on the 147, 100, 68, and Figure3.6.5: Estimates of the bias in MLS v4.2x ClO data in 5 T 46 hPa MLS retrieval pressure surfaces (see legend). Each panel shows monthly zonal means of v4.2x MLS nighttime OH S ◦ (solar zenith angle (SZA) 100 ) ClO measurements averaged over a 12-yr period (2005–2016, lled circles). The grey > T line marks the zero level. The colored solid lines denote the global mean bias estimate at each level. To ensure that RHI S ◦ ClO was not enhanced, consideration was restricted to latitudes equatorward of 50 S for the days between 1 May ◦ T N for the days between 1 December and 1 April. and 1 November and to latitudes equatorward of 50 SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 47 Level 2 Version 4.2 Quality T

54 Help 3.6. Chlorine Monoxide (ClO) Overview 0.2 146 100 Table 68 46 0.1 31 BrO S 21 14 T 10 CH S 0.0 3 Cl T CH S v4 ClO bias / ppbv 3 -0.1 CN T CH S 3 OH -0.2 T -10 -30 -50 -70 10 -90 90 70 50 30 ClO S Latitude / degrees T ◦ S CO Figure 3.6.6: Estimates of the bias in MLS v4.2x ClO data in 5 -wide geographic latitude bands on MLS retrieval pressure surfaces from 147 to 10 hPa (see legend) calculated over a 12-yr period (2005–2016). To ensure that ClO T ◦ GPH was not enhanced, consideration was restricted to latitudes equatorward of 50 S for the days between 1 May and S ◦ 1 November and to latitudes equatorward of 50 N for the days between 1 December and 1 April. The colored solid T lines denote the global mean bias estimate at each level. The large positive bias at low latitudes at 147 hPa is cut o H S 2 in this gure. O T HCl S quantied). Moreover, themagnitudeofthebiasvarieslittlefromoneyeartothenext(notshown). ¿erefore, T we compute 12-yr climatological bias estimates. To guide the eye, the global mean climatological bias value HCN S is indicated for each level (colored solid lines) in both gures. As discussed above, the magnitude, and at T 147hPa even the sign, of the bias varies with latitude, and Figures 3.6.5 and 3.6.6 underscore the necessity of HNO S applying latitudinally and altitudinally varying bias corrections. Given the degree of seasonal variability seen 3 in Figure 3.6.5, monthly varying bias estimates would also be desirable; as noted previously, however, it is T HO not possible to directly quantify the bias in the polar regions during much of the year. Attempts to estimate S 2 the bias through approaches other than examination of nighttime ClO measurements (e.g., via correlations T with potentially interfering species, which themselves may vary strongly with season) have so far met with HOCl S only limited success. Consequently, we report time-invariant but altitude- and latitude-dependent v4.2 ClO x T bias corrections, which are adequate for most studies. An ASCII le containing the estimated bias values is IWC S available from the MLS web site. T IWP S 3.6.5 Review of comparisons with other data sets Extensive comparisons of MLS v2.2 ClO data with a variety of dierent platforms (ground-based, balloon- T N S borne, aircra , and satellite instruments) were presented by [2008]. A subset of those compar- Santee et al. 2 O x isons with v3.3 and v3.4 x ClO data were reported by [2013]. Livesey et al. T S O 3 3.6.6 Data screening T Pressure range: 147–1.0hPa OH S Values outside this range are not recommended for scientic use. T RHI S Estimated precision: Only use values for which the estimated precision is a positive number. T information has a strong inuence are agged with negative or zero precision, Values where the a priori SO S and should not be used in scientic analyses (see Section 1.5). 2 T S T Aura Microwave Limb Sounder (MLS) x 48 Level 2 Version 4.2 Quality T

55 Help 3.6. Chlorine Monoxide (ClO) Overview Status Status ag: Only use proles for which the eld is zero. Status We recommend that all proles with nonzero values of be discarded, because of the potential Table impact of cloud artifacts at lower levels. Note, however, that rejecting in their entirety all proles with nonzero Status may be unnecessarily severe at and above (i.e., at pressures equal to or smaller than) BrO S 46hPa, where clouds have negligible impact; thus otherwise good-quality proles with nonzero but Status values may be used without restriction at those levels as long as they are removed at larger even T CH eld. Status pressures. See Section 1.6 for more information on the interpretation of the S 3 Cl Quality: Only proles whose eld is greater than 1.3 should be used. Quality T CH x and v3.4 x (unchanged from v3.3 Quality ¿is threshold for ) typically excludes less than 1% of ClO S 3 CN proles on a daily basis; note that it potentially discards some “good” data points while not necessarily T identifying all “bad” ones. CH S eld is less than 1.05 should be used. Convergence Convergence: Only proles whose 3 OH (unchanged from v3.3 x Convergence and v3.4 ) discards very few On a typical day this threshold for x T (0.3%orless)oftheClOproles, many(butnotall)ofwhicharelteredoutbytheotherqualitycontrol ClO S measures. T S CO 3.6.7 Artifacts T • x Signicant biases are present in both daytime and nighttime v4.2 ClO mixing ratios at and below GPH S (i.e., pressures larger than) 68hPa. ¿e bias should be corrected by subtracting from the individual T measurements at each aected retrieval level the altitude- and latitude-dependent bias estimates given H S 2 O in an ASCII le available from the MLS web site. T HCl S 3.6.8 Desired improvements for future data version(s) T • Reduce the biases present at the lowest retrieval levels (147–68hPa). HCN S T HNO S 3 T Summary of Aura MLS v4.2x ClO Characteristics Table 3.6.1: HO S Systematic Resolution b 2 Pressure Precision Known Artifacts c a T Uncertainty H × V / hPa / ppbv or Other Comments HOCl S / ppbv / km Unsuitable for scientic use — — — 0.68–0.001 T 0.02 0.3 500 ± × ± 1.0 3 IWC S 15–1.5 3.5–4 × 250–400 ± 0.1 0.02–0.03 ± T × 0.05 0.1 22 3 ± 300–400 ± IWP S 46–32 ± 300–400 × 3 0.15 ± 0.1 d 0.2 68 3 × 450 ± Latitude-dependent bias 0.1 ± T d N S 0.25 Latitude-dependent bias 100 3 × 500 ± 0.1 ± 2 O d ± 147 4.5 600 0.3 ± 0.4 Latitude-dependent bias × T — 1000–215 Not retrieved — — S O 3 a T Vertical and Horizontal resolution in along-track direction. b OH S Precision on individual proles, determined from observed scatter in nighttime (descending) data in a region of minimal atmo- spheric variability. T c estimates of the probable magnitude and, at the higher pressures, are the uncertainties a er σ Values should be interpreted as 2- RHI S subtraction of the known bias. d T Correct for the bias by subtracting from the individual measurements at this level the latitude-dependent bias estimates given SO S in the ASCII le available from the MLS website 2 T S T Aura Microwave Limb Sounder (MLS) x 49 Level 2 Version 4.2 Quality T

56 Help Overview 3.7 Carbon monoxide (CO) Table CO Swath name: Useful range: 215 – 0.0046 hPa BrO S Contact: > Hugh C. Pumphrey (stratosphere/mesosphere), Email: < [email protected] T Email: > < [email protected] Michael Schwartz (troposphere), CH S 3 Cl T 3.7.1 Introduction CH S 3 CN Carbon monoxide is retrieved from radiance measurements of two bands in the MLS 240GHz radiometer. Full details are given in Pumphrey et al. [2008]. Livesey et al. [2007] and T CH S 3 OH 3.7.2 Dierences between v4.2 and v03.x x T x x x Intheuppertroposphere(UT),theprimarydierencebetweenv4.2 COisthereduction /v3.4 COandv3.3 ClO S in the frequency and severity of artifacts caused by deep convective clouds. ¿is was accomplished through T a modication to the manner in which cloud signals are modeled in the retrievals. Screening of data has S CO been simplied, fewer proles are marked “bad” and fewer proles that appear to contain cloud artifacts pass through the recommended screening. T GPH S A comparison of the two versions in the UTLS for February 2009, shown in Figure 3.7.1, includes non- spurious high values from the plume of the “Black Saturday” re, that are seen in both versions (so near the T H black 1:1 line). Scattering from large particles in convective cores produces mostly high outliers in v03.30 S 2 O CO at 215hPa, 147hPa and 100hPa and low outliers at 68hPa and 46hPa that are greatly reduced in v4.2 , x T resultinginthecloudsofpointsawayfromthe1:1line. ¿e“BlackSaturday”plumeproducesUTLSCOvalues HCl S outside of the typically encountered, allowing comparison of non-spurious high values from the two version, T mixing ratios. At the highest values of unobscured by the v03.30 cloud-induced scatter present at lower v4.2 x HCN S x x and v3.4 are lower than those from v3.3 x the “Black Saturday” tails at 100hPa and 68hPa, values from v4.2 T 10%, but fractional dierences between the versions are smaller at lower mixing ratios. ¿e screening by ∼ HNO S and v3.4 x x recommended for v3.3 results in a signicant amount of missing data in the tropics, and when the 3 screening is included, a signicant fraction of the recommended IWC-based component of v3.3 x and v3.4 x T HO S “Black Saturday” non-spurious high values are screened out. 2 T HOCl 3.7.3 Resolution S Figure 3.7.2 shows the horizontal and vertical averaging kernels for v4.2 x MLS CO. Vertical and horizontal T resolution are essentially unchanged from v3.3 x x . ¿e vertical resolution is in the range 3.5–5km and v3.4 IWC S from the upper troposphere to the lower mesosphere, degrading to 6–7km in the upper mesosphere. Down T to the 215hPa level, the vertical averaging kernels are sharply peaked at the level being retrieved, but while IWP S the 316-hPa measurement contains contribution from 316hPa, it has a larger contribution from 215hPa and a T negative contribution around 100hPa of similar magnitude to that at 316hPa. ¿e retrieved value at 316hPa N S 2 is thus more an extrapolation of the prole higher in the UTLS than it is an independent measurement at O T 316hPa, and it is not recommended for scientic use. ¿e horizontal resolution is about 200km in the meso- S O sphere, degrading slowly to 300km with decreasing height in the stratosphere and more rapidly to about 3 T 460km at 100hPa and 690km at 215hPa. OH S 3.7.4 Precision T RHI S a posteriori ¿e MLS data are supplied with an estimated precision (the eld ) which is the L2gpPrecision precision as returned by the optimal estimation. ¿e precision of the v4.2 x CO is very similar to that of T SO S x v3.3 /v3.4 x and is usually smaller (better) than that of v2.2. In all versions, the precision is greater than the 2 T S T Aura Microwave Limb Sounder (MLS) x 50 Level 2 Version 4.2 Quality T

57 Help 3.7. Carbon monoxide (CO) Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T ClO S T S CO T GPH S T H S 2 O T HCl S T HCN S T HNO S 3 T HO S 2 T HOCl S T IWC S Figure 3.7.1: Version v03.30 CO scattered against v04.20 CO for data from February 2009, when the “Black Saturday” T re provided a plume of non-spurious high-CO outliers, unprecedented in the MLS record. This data is unscreened. IWP S The 316-hPa retrieval level is shown, but is not recommended for scientic use. T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 51 Level 2 Version 4.2 Quality T

58 Help 3.7. Carbon monoxide (CO) Overview Table Equator FWHM / km FWHM / km 800 600 400 200 0 1000 1200 12 10 -2 0 2 4 6 8 0.001 BrO S T 0.010 CH S 3 Cl 0.100 T CH S 3 CN 1.000 T Pressure / hPa CH S 10.000 3 OH T 100.000 ClO S T 1000.000 0.2 1.2 -0.2 0.8 0.4 -4 -2 0.0 0 0.6 1.0 2 4 S CO Profile number Kernel, Integrated kernel 0 N 70 T FWHM / km FWHM / km GPH 0 -2 200 400 600 800 1000 1200 0 2 4 6 8 10 12 S 0.001 T H S 0.010 2 O T HCl S 0.100 T HCN 1.000 S T Pressure / hPa HNO 10.000 S 3 T 100.000 HO S 2 1000.000 T HOCl -2 0.8 0.6 0.4 0.2 0.0 -0.2 -4 1.0 4 2 0 1.2 S Kernel, Integrated kernel Profile number T IWC S Typical two-dimensional (vertical and horizontal along-track) averaging kernels for the MLS v4.2x CO Figure 3.7.2: T ◦ data at the equator (upper) and at 70 N (lower); variation in the averaging kernels is suciently small that these IWP S are representative of typical proles. Colored lines show the averaging kernels as a function of MLS retrieval level, T indicating the region of the atmosphere from which information is contributing to the measurements on the indi- N S vidual retrieval surfaces, which are denoted by plus signs in corresponding colors. The dashed black line indicates 2 O the resolution, determined from the full width at half maximum (FWHM) of the averaging kernels, approximately T scaled into kilometers (top axes). (Left) Vertical averaging kernels (integrated in the horizontal dimension for ve S O 3 along-track proles) and resolution. The solid black line shows the integrated area under each kernel (horizontally T and vertically); values near unity imply that the majority of information for that MLS data point has come from the OH S measurements, whereas lower values imply substantial contributions from a priori information. (Right) Horizon- T tal averaging kernels (integrated in the vertical dimension) and resolution. The horizontal averaging kernels are RHI S shown scaled such that a unit averaging kernel amplitude is equivalent to a factor of 10 change in pressure. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 52 Level 2 Version 4.2 Quality T

59 Help 3.7. Carbon monoxide (CO) Overview v04.20 v04.20 v04.20 v04.20 v04.20 v04.20 0.001 0.001 v03.30 v03.30 v03.30 v03.30 v03.30 v03.30 v02.21 v02.21 v02.21 v02.21 v02.21 v02.21 Table 80 80 0.01 0.01 ) ) BrO hPa hPa 0.1 0.1 S 60 60 2005d028 2006d059 T 1 1 CH Pressure Pressure S ( ( 10 10 3 40 40 Cl log log Approx Altitude / km Approx Altitude / km T 10 10 CH S 3 Std dev Std dev Std dev Std dev Std dev Std dev CN 20 20 Precision Precision Precision Precision Precision Precision 100 100 T Mean VMR Mean VMR Mean VMR Mean VMR Mean VMR Mean VMR CH S 3 5e+00 5e+00 5e−01 5e−02 5e−02 5e−01 5e−03 5e−03 OH T Precision and scatter (ppmv) Precision and scatter (ppmv) ClO S Scatter (standard deviation) and (estimated) precision for MLS v4.2x (black), /prevvShalsh (blue) and Figure 3.7.3: T ◦ of the equator on 28 January 2005 and 28 v2.2 (red) CO. The statistics shown are generated from all proles within 20 S CO February 2006. Proles of the mean volume mixing ratio (VMR) are shown for comparison. The vertical co-ordinate T is hPa )) so that 16 km on the axis is exactly 100 hPa. − 3 log 16 ( ( Pressure / 10 GPH S T scatter observed in the data in regions of low natural variability. Where the estimated precision is greater H S 2 O than 50% of the a priori precision the data will be inuenced by the a priori to an undesirably large extent. In T such cases, L2gpPrecision is set to be negative (or zero in some cases) to indicate that the data should not be HCl S used. Figure 3.7.3 shows both the scatter and estimated precision for CO, with typical proles for comparison. T Note that the random errors are larger than 100% of the mixing ratio for much of the vertical range, meaning HCN S that signicant averaging (e.g., daily zonal mean or weekly map) is needed to make use of the data. T HNO S 3.7.5 Accuracy 3 ¿e estimated accuracy is summarized in Table 3.7.1. In the middle atmosphere the accuracies are estimated T HO S by comparisons with the ACE-FTS instrument; see Pumphrey et al. [2007] for further details. ¿e MLS v2.2 2 CO data at 215hPa showed high (factor of ~ 2) biases compared to other observations. ¿e morphology, T HOCl however, was generally realistic [ Livesey et al. , 2008]. In v4.2 x (as in v3.3 x /v3.4 x ) this bias has been essentially S eliminated through a change in the approach to modeling the background radiance upon which the CO T spectral line sits, and a small reduction in the number of MLS spectral channels considered in the retrieval. IWC S Tropospheric (215–100hPa) accuracies in Table 3.7.1 are 2- estimates obtained by propagating parameter σ T uncertainties through a model of the measurement system. IWP S T 3.7.6 Data screening N S 2 Pressure range: 215–0.0046hPa. O T Values outside this range are not recommended for scientic use. S O 3 Estimated precision: Only use values for which the estimated precision is a positive number. T information has a strong inuence are agged with negative or zero precision, a priori Values where the OH S and should not be used in scientic analyses (see Section 1.5). T RHI S Status Status ag: Only use proles for which the eld is an even number. T indicate that the prole should not be used in scientic studies. See Section 1.6 Status Odd values of SO S eld. Status for more information on the interpretation of the 2 T S T Aura Microwave Limb Sounder (MLS) x 53 Level 2 Version 4.2 Quality T

60 Help 3.7. Carbon monoxide (CO) Overview Cloudshavenoimpactforpressuresof31hPaorless. v4.2 Clouds: ismuchlesssusceptibletocloud-induced x artifacts than was v03.x, so screening of the CO product is considerably simplied, needing only appli- Table rules as described below. Convergence and Quality cation of the standard Quality: Only use proles with greater than 1.5. Quality BrO S ∼ 2% of proles in the Proles with Quality less than or equal to 1.5 comprise 0.7% of all data, and T tropics. At 215 hPa in the tropics, the rejected proles have mixing ratios that are, on average, 7% CH S higher than other tropical values and the occurrence rates of high outliers with mixing ratios greater 3 Cl than 150ppbv is about twice that of the rest of the tropical ensemble. At 100hPa there is no signicant T dierence in the means of tropical proles with Quality less than or greater than 1.5. CH S 3 CN Convergence: Only proles whose Convergence eld is less than 1.03 should be used. T criterion rejects fewer than 0.1% of proles. Almost all of the retrievals in the phase Convergence ¿is CH S CO converge to their target, so that produces v4.2 x is of limited use in screening. Convergence 3 OH T 3.7.7 Artifacts ClO S • Positive systematic error of 20–50% throughout the mesosphere. T S • Negative systematic error of 50–70% near 30hPa. CO T • Retrieved proles are rather jagged, especially between 1hPa (48km) and 0.1hPa (64km). ¿e greater GPH S x smoothing applied in v4.2 x /v3.4 ) compared to v2.2 has reduced this problem considerably x (and v3.3 but has not eliminated it entirely. T H S 2 • ¿ere is a tendency for negative values to occur at the level below a large positive value. ¿e most O T striking examples occur in the polar vortex, where air with high CO mixing ratios descends to the mid- HCl S stratosphere. ¿is problem is slightly worse in v4.2 x /v3.4 x and v3.3 x than in v2.2 – this was considered T an acceptable trade-o for the less jagged proles obtained over most of the middle atmosphere. HCN S • Upper-tropospheric v4.2 x CO retrieved values still show some anomalous sensitivity to thick clouds T and associated with deep convection. ¿e screening procedure based upon Quality Convergence that HNO S is described above is generally eective in removing these artifacts. 3 T HO 3.7.8 Review of comparisons with other datasets S 2 In the upper troposphere, comparisons with various in situ CO observations (NASA DC-8, WB-57 and the T HOCl MLS v2.2 215hPa CO product was biased high by a factor of earlier 2. MOZAIC dataset) indicate that the ~ S x , this is largely eliminated in v4.2 x . As in v3.3 x and v3.4 T In the mesosphere, comparisons of the earlier v2.2 MLS CO with ODIN-SMR and ACE-FTS suggest a IWC S positive bias: 30%–50% against ACE-FTS, 50%–100% against SMR. Near 31hPa, the MLS values are lower T than SMR and ACE-FTS by at least 70%. ¿e MLS values have not changed much between v2.2 and v4.2 x IWP S in the middle atmosphere, so these comparisons may mostly be considered valid for v4.2 . What change x T there was between v2.2 and later versions consists of a slight lowering of the MLS values, bringing them N S 2 slightly towards the ACE-FTS data; 20% is now a better estimate of the MLS-ACE bias in much of the middle O T atmosphere (compared to 30% with v2.2). ¿e dierence between v3.3 x /v3.4 x and v4.2 x in the mesosphere S O is small. 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 54 Level 2 Version 4.2 Quality T

61 Help 3.7. Carbon monoxide (CO) Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T Table 3.7.1: Data quality summary for MLS v4.2x CO. ClO S Comment Pressure Resolution / km Precision Accuracy / T S Vert Horiz. × ppbv / hPa CO T . < 001 — 0 — — Not retrieved GPH S 0.0022-0.001 — — Unsuitable for scientic use — T × 0.0046 50% + 20% to + 6000 200 6.2 H S 3400 0.01 5.9 × 200 50% + 20% to + 2 O 0.046 5.9 × 200 1000 + 20% to + 50% T HCl S × 200 640 + 20% to + 50% 0.14 3.8 + 4.1 20% to 250 1 × 50% 120 + T HCN 10 5.0 × 440 16 ± 10 % S − 4.9 × 400 9 31 70% to − 50% T a 4.9 × 19ppbv and 30% ± 100 ± 14 450 HNO S 26ppbv and 570 16 ± 5.1 ± 30% 147 × 3 T 215 5.4 × 690 19 ± 38ppbv and ± 30% HO S — 316 — — Unsuitable for scientic use 2 T — — Not retrieved 316 — > HOCl S a Estimated 215–100 hPa systematic uncertainty is the RSS of the additive and multiplicative terms. T IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 55 Level 2 Version 4.2 Quality T

62 Help Overview 3.8 Geopotential Height (GPH) Table GPH Swath name: Useful range: 261 – 0.001 hPa BrO S Email: < [email protected] > Contact: Michael J. Schwartz, T CH S 3 Cl 3.8.1 Introduction T CH Geopotential height (GPH) is retrieved, along with temperature and the related assignment of tangent pres- S 3 sures to limb views, primarily from bands near O GPH x spectral lines at 118-GHz and 234GHz. ¿e v4.2 CN 2 Schwartz product is generally similar to both the v03.x product and to the v02.2 product that is described in T CH S et al. [2008]. 3 OH ¿e heights of surfaces of constant geopotential are a property of the Earth’s gravitational eld and do T not depend upon atmospheric conditions. Geopotential dierences between surfaces are equal to the inte- ClO S gral with height of the gravitational acceleration, g . GPH is geopotential dierence from the Earth’s surface T geopotential to a given location, scaled by the mean-sealevel gravitational acceleration, g , to give units of 0 S CO height. MLS products, including GPH and temperature, are reported on pressure surfaces, but pressure, temper- T GPH S ature and height depend upon one another through assumed hydrostatic balance and the gas law. T 1 R H S (3.1) , ~ P g dh = T dP S S 2 O g Mg o o T HCl S where M is the molar mass of air and R is the gas constant. ¿us a retrieval of a temperature prole on xed pressure surfaces also determines GPH dierences be- T HCN S tween those surfaces, and so determines the GPH prole to within an additive constant. Absolute pointing cannot be inferred from radiometric information and must be obtained, for each prole, from the spacecra T HNO orbit/attitude system and the instrument scan model. By convention, this additional degree of freedom is S taken to be the height of the 100-hPa reference level, but this choice is arbitrary, as the reference is an additive 3 T oset to the entire prole. HO S x Table 3.8.1 summarizes v4.2 measurement precision, modeled accuracy and observed biases. ¿e fol- 2 T lowing sections provide details. HOCl S and previous versions x 3.8.2 Dierences between v4.2 T IWC S CorePlusR3 GPH product is taken from the x ¿e v4.2 retrieval phase along with a number of standard prod- ucts that depend upon radiances of the 240-GHz radiometer. ¿e previously released v02.2x and v03.x GPH T IWP products were taken from a preliminary “Core” retrieval phase. Since these GPH proles are retrieved simul- S products and are assumed in the CorePlusR2 and CorePlusR4 retrievals, taneously with the CorePlusR3 T GPH and the retrieved constituents should be, in some sense, more internally consistent than they were in N S 2 O previous versions. T ¿ese previous versions of MLS GPH were referenced to the WGS84 ellipsoid, a simple model of the S O 3 x uses a surface of constant geopotential, Earth’s surface but not a surface of constant geopotential. v4.2 T the Earth Gravitational Model 1996 (EGM96) geoid, as its zero GPH surface. ¿e magnitude of the static, OH S latitude/longitude-dependentdierencebetweenthegeoidandellipsoidexceeds ± 100minplaces,asisshown T in Figure 3.8.1. ¿is EGM96-WGS84 dierence is a bias that should be subtracted from previous versions’ RHI S x GPH before they are compared to v4.2 GPH. ¿ese dierences, if not corrected, will result in an erroneous T additive contribution to geostrophic winds calculated from v03.x and v02.x GPH at all levels. SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 56 Level 2 Version 4.2 Quality T

63 Help 3.8. Geopotential Height (GPH) Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH The height of the EGM96 geoid from the WGS84 ellipsoid. These values should be subtracted from Figure 3.8.1: T MLS v03.x and earlier GPH values to give geopotential heights referenced to the Earth’s mean sea level geopotential. ClO S v4.2x GPH does not include this bias term. T S CO 80 80 T GPH S 60 60 T 40 40 H S 2 O T 20 20 HCl S 0 0 T HCN Latitude Latitude S −20 −20 T HNO S −40 −40 3 T −60 −60 HO S 2 −80 −80 T J J A S O N D A J D N O S A J J M F M F J M A M HOCl S 2005 2009 T IWC S −10 0 10 20 30 40 50 70 60 (m) T IWP S Figure 3.8.2: Dierences between v4.2x and ellipsoid-referenced v03.30 100-hPa GPH for 2005 and 2009 are shown T as functions of latitude and day of year. N S 2 O T GPH values are found to be somewhat x A er correction of v03.3x for the EGM96-WGS84 bias, v4.2 S O 3 higher at high latitudes than are those of the v03.3x product, with typical 100-hPa mean dierences of 50m in T the high northern latitudes, 30m in high southern latitudes and close to zero bias in the tropics. Latitudinally OH S and ellipsoid-referenced v3.3 x x and seasonally dependent dierences between v4.2 100-hPa GPH are /v3.4 x T shown in Figure 3.8.2 for the years 2005 and 2009. ¿ese dierences are very similar for ascending and RHI S descending parts of the orbits, and the two are averaged in this gure. ¿e details of these dierences vary T between years, but the morphology is similar. SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 57 Level 2 Version 4.2 Quality T

64 Help 3.8. Geopotential Height (GPH) Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T ClO S T S CO Annually-repeating biases between MLS 100-hPa GPH and GEOS-5.2 100-hPa GPH as functions of as- Figure 3.8.3: T cending and descending latitude and of day of year. These biases were computed based upon data from the years GPH S 2005–2012. T H S 3.8.3 Comparison to GEOS-5 Analysis 2 O T A er removal of the geoid correction from v03.3x GPH, there remains a latitudinally and seasonally-varying HCl S bias between MLS and GEOS-5 that is also present in v4.2 x GPH. Figure 3.8.3 shows this pattern in the . dierence between geoid-corrected v03.3x and GEOS-5 100-hPa GPH. ¿is pattern is very similar in v4.2 x T HCN S Some of this annually-repeating “bias” may result from annually-repeating geophysical information cap- tured by MLS that is not in the GEOS-5 analysis. However, the large ascending-descending dierences are T HNO almost certainly not primarily diurnal eects. ¿e delity with which these patterns annually repeat suggests S that they may result from thermal distortions of the spacecra /instrument that vary annually due to the ec- 3 T centricity of the Earth’s orbit, or perhaps from orbit/attitude errors from the star trackers that provide the HO S spacecra -attitude information. ¿e same star elds are viewed each year at the same date and orbit position. 2 T Figure 3.8.4 shows daily-binned dierences between v4.2 and GEOS-5.9 100-hPa GPH as functions of x HOCl S orbital position. ¿e patterns are extremely similar in the two years and to the annually repeating v3.3 x and v3.4 x 80m lower than 2005 due to a decreasing trend in MLS residual shown in Figure 3.8.3, but 2009 is ∼ T IWC S x GPH. v4.2 ¿e monthly-averaged dierences between MLS v4.2 x 100-hPa GPH and GEOS-5.9 100-hPa GPH for 11 T IWP years of Januarys are shown in Figure 3.8.5. ¿ere is a clear downward trend of ∼ 100m in the rst 4 years of S the mission, 2005–2008, with a further 15m decrease in 2009–2014. ¿is trend is in the MLS GPH and is not ∼ T GPH, for which the entire MLS record has evident in the GEOS-5.9 time series. Analysis of v3.3 x and v3.4 x N S 2 O been processed, shows the same global decreasing trend. It is important to emphasize that MLS radiances do T not provide absolute pointing information, but only GPH dierences between pointing. Absolute pointing S O 3 comes from spacecra ephemeris and attitude data and the instrument scan model. ¿e origins of the trend T is a topic of current research, but it is likely not primarily of geophysical origin. OH S When both the trend and the annually repeating bias are removed from v3.3 x and v3.4 x , the standard T deviation of the residual dierences between the bias-corrected MLS and GEOS-5 100-hPa GPH is on the RHI S x . order of 20m, and a similar result is expected for v4.2 T ¿e last four proles of each day show increased rates of outliers relative to GEOS-5 temperature, com- SO S pared to other prole positions in the day in version 4.20 (xed in version 4.22). ¿ese outliers can be as large 2 T S T Aura Microwave Limb Sounder (MLS) x 58 Level 2 Version 4.2 Quality T

65 Help 3.8. Geopotential Height (GPH) Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T ClO S T S CO T GPH S T H Daily-binned dierences between v4.2x and GEOS-5.9 100-hPa GPH as functions of orbital position. Figure 3.8.4: S 2 O Orbital position, from the bottom of each panel, ascends from the equator to the northern terminator, descends T through the equator to the southern terminator and then ascends to the equator. HCl S T HCN S 220 T HNO S 200 3 T HO S 2 180 T HOCl S 160 T IWC S T 140 IWP S T 120 N S 2 O T MLS v04.20 100−hPa GPH minus GEOS−5.9 GPH (m) S O 100 3 T OH S 80 2008 2009 2010 2011 2012 2006 2013 2014 2015 2007 2005 T RHI S T January monthly-binned dierences between v4.2x and GEOS-5.9 100-hPa GPH. Figure 3.8.5: SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 59 Level 2 Version 4.2 Quality T

66 Help 3.8. Geopotential Height (GPH) Overview as 50K at at some levels. A recommendation to discard the last 4 proles of each day for both temperature and GPH (as well as RHI) has been added to recommended screening for v4.20 (not needed for v4.22). Table 3.8.4 Vertical resolution ¿e GPH prole is vertically-integrated temperature, so its vertical resolution is not well-dened. ¿e vertical BrO S resolution of the underlying temperature given in Section 3.22 is repeated in Table 3.8.1. T CH S 3.8.5 Precision 3 Cl T GPH precision is summarized in Table 3.8.1. Precision is the random component of measurements x MLS v4.2 CH S that will average-down if a measurement is repeated. ¿e retrieval algorithms produce an estimate of GPH 3 CN precision only for the 100hPa reference level, as this is the only element included in the MLS “state vector.” T Precisions reported for this level in the L2GP les are typically less than 8m, are believed to be optimistically CH S small, and may underestimate the impact of scatter in the absolute pointing information. A precision of 3 OH 20m is reported for 100hPa in Table 3.8.1, column 2. GPH precisions at other standard-product prole levels T (summarized in column 2 of Table 3.8.1) are calculated from the GPH precision at the reference level and ClO S the prole of temperature precisions, but values in the table are inated somewhat from those in the L2GP T les in the troposphere and stratosphere. Estimated precisions remain at 20–25m up to 1hPa and degrade S CO to 110m at 0.01hPa. O-diagonal elements of the temperature/GPH error covariance matrix are neglected in T 5m near 100hPa.) this GPH-precision-prole calculation, but resulting errors are believed to be small ( ~ GPH S T 3.8.6 Accuracy H S ¿e accuracy of the v2.2 GPH was modeled based upon consideration of a variety of sources of systematic 2 O T Schwartz et al. accuracy is believed to be substantially similar and the re- error, as discussed in x [2008]. v4.2 HCl S sults of the v2.2 calculations are given in column four of Table3.8.1. Of the error sources considered, modeled amplier non-linearity had the largest impact, just as is the case with the calculation for temperature. T HCN S ¿e second terms in column four are model-based estimates of the bias magnitude from other sources including uncertainty in pointing/eld-of-view, uncertainty in spectroscopic parameters, and retrieval nu- T HNO merics. ¿e combined bias magnitudes due to these sources is 100–150m. S “Observedbiasuncertainty”inTable3.8.1isanestimateofbiasbaseduponcomparisonswithanalysesand 3 T withotherpreviously-validatedsatellite-basedmeasurements. ¿esecomparisonsweremadeusingMLSv2.2, HO S GPH are generally less than 50m from 261–0.1hPa and reasonably x but as the biases between v2.2 and v4.2 2 T x constant, these results hold for v4.2 as well. ¿e primary sources of correlative data were the Goddard HOCl S Earth Observing System, Version 5.0.1 data assimilation system (GEOS-5) [ Rienecker et al. , 2007], used in the troposphere and lower stratosphere, and the Sounding of the Atmosphere using Broadband Radiometry T IWC S Mlynczak and Russell (SABER) [ , 1995], used in the upper stratosphere through the mesosphere. MLS has a 150m high bias relative to analyses (GEOS-5) at 100-hPa that drops to 100m at 1hPa. Biases with respect to T SABER are small at 0.1hPa but increasingly negative at higher levels, reaching -600m at 0.001hPa, but with IWP S signicant latitudinal and seasonal variability. T N S 2 O 3.8.7 Known Artifacts T To maintain its position in the “A-Train” constellation of satellites, the Aura spacecra periodically executes S O 3 “inclination adjustment maneuvers” (IAMs) involving large changes in spacecra yaw and ring of thrusters. T ¿ere were 39 such maneuvers in the rst 10 years of operation. Large systematic errors in spacecra reported OH S attitude have been found in the rst orbit that follows the return to nominally good attitude data a er these T maneuvers, apparently due to the ringing of a Kalman lter in the attitude measurement system. Since MLS RHI S GPH relies on spacecra attitude for its absolute reference, these attitude errors result in large systematic T errors in GPH, typically reaching +3500m in the rst eighth of the orbit and -30m in the second quarter SO S orbit a er the nominal “end of maneuver”. ¿e large positive errors almost always occur on ascending orbits 2 T S T Aura Microwave Limb Sounder (MLS) 60 Level 2 Version 4.2 x Quality T

67 Help 3.8. Geopotential Height (GPH) Overview from near Australia, crossing the Indian Ocean and, at 3500m amplitude, are large enough to signicantly ∼ bias even long-term averages. A le of start and stop times for data to be avoided is maintained on the MLS Table . website: http://mls.jpl.nasa.gov/cal/issues.txt 3.8.8 Data screening BrO S GPH should be screened in much the same way as is temperature: T CH S Pressure range: 261–0.001hPa. 3 Cl T Values outside this range are not recommended for scientic use. CH S 3 Estimated precision: Only use values for which the estimated precision is a positive number. CN a priori information has a strong inuence are agged with negative or zero preci- Values where the T CH S sion, and should not be used in scientic analyses (see Section 1.5).GPH precision is set negative at 3 OH and beyond any level in the integration of temperature away from the 100-hPa reference level where T temperature has negative precision. ClO S eld is an even number. Status ag: Only use proles for which the Status T Status Odd values of indicate that the prole should not be used in scientic studies. See Section 1.6 S CO for more information on the interpretation of the eld. Status T GPH GPH measurements in the upper troposphere Clouds: ¿ick clouds are believed have some impact on v4.2 x S (261–100hPa). Tropospheric GPH may be screened using the ice water content (IWC) product, re- T 3 jecting proles between 261–100hPa for which the 215hPa value of IWC is greater than 0.005mg/m . H S 2 O greater than 0.2 for the 83hPa level and smaller pressures, and Quality Only use proles with Quality: T HCl Quality proles with greater than 0.9 at larger pressures of 100hPa and larger. S T Convergence: Only proles whose eld is less than 1.03 should be used. Convergence HCN S Use of this threshold typically discards less than 0.1% of proles and is primarily a safeguard against T proles with extremely poor convergence. HNO S End of day (v4.20 only): ¿e last four proles of each day show greatly increased rates of large temperature 3 T . Accordingly the last four proles of the GPH product each day should not be a priori departure from HO S used. Note that this issue was xed in v4.22. 2 T Large errors in reported spacecra attitude have been found following the spacecra inclination ad- IAMs: HOCl S justment maneuvers that occur ∼ 4 times per year. A le of times at which GPH should not be used is http://mls.jpl.nasa.gov/cal/issues.txt . maintained on the MLS website: T IWC S 3.8.9 Desired improvements for future data version(s) T IWP S Reduction of seasonally and latitudinally-repeating systematic errors in GPH that may be the result of errors in the absolute pointing information from the spacecra attitude and ephemeris data stream is an area of T N ongoing research. Undertanding the decreasing trend in MLS global mean GPH, particularly in the rst four S 2 O years of the mission is an area of active research. T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 61 Level 2 Version 4.2 Quality T

68 Help 3.8. Geopotential Height (GPH) Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T ClO S T Summary of MLS GPH product. Table 3.8.1: S CO Observed Resolution a T Precision Comments Region GPH accuracy Horiz. × Vert. S / meters / m / km T H — — < — Unsuitable for scientic use 0.001hPa S 2 O × 0.001hPa 10–13 450 220 ± 160 − T 100 − 0.01hPa 8–12 × 185 ± 110 HCl S ± 0.1hPa 6 × 165 45 0 T 100 25 1hPa 7 × 165 ± HCN S 20 ± 165 × 4.3 10hPa 100 T 5.2 × 165 ± 20 150 100hPa HNO S × 170 ± 20 150 261hPa 5.3 3 Unsuitable for scientic use 1000–316hPa — — — T HO S a 2 Precision on individual proles T HOCl S T IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 62 Level 2 Version 4.2 Quality T

69 Help Overview 3.9 Water Vapor (H O) 2 Table H2O Swath name: Useful range: 316 – 0.002 hPa BrO S Email: < [email protected] > Alyn Lambert (stratosphere/mesosphere), Contact: T Email: [email protected] > William Read (troposphere), < CH S 3 Cl T 3.9.1 Introduction CH S 3 CN ¿e standard water vapor product is taken from the 190-GHz “ ” retrieval. ¿e vertical grid for CorePlusR2 O is: 12 levels per decade change in pressure (LPD) for 1000–1 hPa, 6LPD for 1.0–0.1hPa, and 3LPD for H T 2 CH ○ − 5 S hPa. ¿e horizontal grid is every 1.5 0.1–10 O is unusual among MLS products in along the orbit track. H 2 3 OH that it is assumed that the logarithm of the mixing ratio, and not mixing ratio itself, varies linearly with log T pressure; although, H VMR). Vertical and O, like the other MLS constituents is retrieved in VMR (not log 2 ClO S space. horizontal interpolations of H log O should be performed in ) VMR ( 2 T x O between 1000 and 383hPa is taken from a retrieval of relative humidity with respect ¿e MLS v4.2 H 2 S CO to ice (RHi) product, converted to specic humidity using the Go-Gratch vapor pressure over ice equation. ¿is RHi retrieval is not vertically resolved, and all levels between 1000 and 383hPa are assumed to have the T GPH S Read same RHi. See Section 3.20 for more information. Validation of MLS v2.2 water vapor is presented in [2007]. ¿is section reiterates the key information from those studies, and et al. [2007] and Lambert et al. T H H x . Table 3.9.1 gives a summary of MLS v4.2 O precision, resolution, and accuracy. x updates them for v4.2 S 2 2 O T x and v3.4 x 3.9.2 Changes from v3.3 HCl S /v3.4 x and v3.3 x O dierences between v4.2 x relate to cloud screening and improvements in ¿e main H 2 T ” initial upper tropospheric and lower stratospheric humidity estimation phase that produces the “ InitUTH HCN S O prole retrieved in the nal retrieval phase that the rst guess and sets smoothing constraints on the H 2 T produces the standard product. ¿ere have been no spectroscopy changes either in linewidth or contin- HNO S and v3.3 x /v3.4 x . An improved cloud detection methodology has been uum charaterization between v4.2 x 3 x that does a better job of rejecting cloudy radiances that are likely to cause poor ts and developed for v4.2 T HO S corrupted proles. ¿e number of erroneous spikes in the upper troposphere has been reduced in v4.2 x rel- 2 ative to v3.3 x and v3.4 x . ¿e InitUTH O phase retrieves H O on a more coarse 6-LPD grid. ¿e InitUTH H 2 2 T HOCl S is used for the initial guess for the nal 12-LPD retrieval and sets smoothing constraints for the prole shape phase has been expanded a priori instead of smoothing to the shape of a climatological prole. ¿e InitUTH T to use more channels and better forward model representations to improve its retrieval accuracy. In addi- IWC S ) to 10hPa tion, the retrieval range has been expanded to have its top moved from 100hPa (in v3.3 x and v3.4 x T ), for a more seamless transition from the troposphere to the stratosphere. Simulation studies show (in v4.2 x IWP S that these changes have improved the agreement between the “truth” proles used to produce the simulated T radiances and the retrieved proles in the 316–215-hPa levels. N S 2 x and v2.2. At most levels, the average dierence is small x Figure 3.9.1 compares MLS v4.2 x to v3.3 /v3.4 O T x and v3.4 x than v3.3 x – less than 10%. ¿e 316–215-hPa levels are 10–20% drier in v4.2 . ¿is dierence in S O . ¿e extremely x behavior is replicated in simulation studies where the agreement with truth is better in v4.2 3 T x . Unfor- O at high latitudes seen in previous versions are still present in v4.2 low values of the 215-hPa H 2 OH S tunately the high latitude behavior is not replicated in simulation studies and therefore is a systematic error that is not understood. T RHI S Humidity data at pressures greater than 316hPa are derived from a broad layer relative humidity retrieval (using low limb viewing MLS wing channel radiances) similar to that obtained from NOAA operational hu- T SO Read et al. midity sounders such as TOVS. As noted in [ , 2007], the v2.2 retrieval at pressures larger than S 2 T S T Aura Microwave Limb Sounder (MLS) x 63 Level 2 Version 4.2 Quality T

70 Help 3.9. Water Vapor (H O) 2 Overview Table Precision Differences Profile 180W--180E, 60N--90N 12 v03.30 v02.21 17 v04.20 BrO S 26 38 56 T 82 CH S 121 3 Cl Pressure (hPa) 177 261 T 383 CH S 0 10 0.1 1.0 10.0 20 1 -40 -20 100 40 Precision Profile Differences 3 CN H2O (ppmv) (%) H2O (ppmv) 180W--180E, 30N--60N 12 v03.30 v02.21 T 17 v04.20 26 CH S 38 3 OH 56 82 T 121 ClO Pressure (hPa) S 177 261 383 T 10 100 20 0 -20 1.0 10.0 1 40 0.1 -40 Differences Profile Precision S CO H2O (ppmv) (%) H2O (ppmv) 180W--180E, 30S--30N 12 v03.30 v02.21 T 17 v04.20 GPH 26 S 38 56 T 82 H S 121 2 O Pressure (hPa) 177 T 261 383 HCl S 0 20 40 10 100 1 -40 0.1 1.0 10.0 -20 Differences Profile Precision H2O (ppmv) (%) H2O (ppmv) T 180W--180E, 60S--30S 12 v03.30 v02.21 HCN 17 S v04.20 26 38 T 56 HNO S 82 121 3 Pressure (hPa) 177 T 261 HO S 383 2 -20 100 10 1 0.1 1.0 10.0 -40 0 20 40 Precision Profile Differences T (%) H2O (ppmv) H2O (ppmv) 180W--180E, 90S--60S 12 HOCl v03.30 S v02.21 17 v04.20 26 T 38 IWC 56 S 82 121 T Pressure (hPa) 177 IWP S 261 383 T 10 1 -40 -20 0 20 40 100 0.1 1.0 10.0 N H2O (ppmv) (%) H2O (ppmv) S 2 O T S O 3 A comparison of v3.3x/v3.4x (blue) to v2.2 (green) and v4.2x (red) water vapor for Jan-Feb-Mar 2005 in Figure 3.9.1: T 5 latitude bands. Other time periods are similar. The left panel compares mean proles, the center shows the mean OH S dierence (red and green diamonds) surrounded by each version’s estimated precision, and the right panel shows the estimated retrieval precision (solid and bullets) and measured variability (dotted) which includes atmospheric T variability about the mean prole. RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 64 Level 2 Version 4.2 Quality T

71 Help 3.9. Water Vapor (H O) 2 Overview 30% too high, based on comparisons with AIRS. ¿e accuracy of this retrieval ~ 316hPa was likely to be the assumed and v3.4 x is highly sensitive to the transmission eciency of the MLS optics system. In v3.3 x Table value of the MLS antenna transmission eciency was adjusted empirically (within the uncertainty range es- tablishedfromMLScalibration)togivebetteragreementwithAIRSinthetropics. Inv4.2 continuum theN x 2 was adjusted only for this phase to minimize the clear sky cloud induced radiance bias. ¿is retrieval is used BrO S as an a priori and prole constraint for the humidity prole at pressures greater than 316hPa, which are not T O product retrieval. retrieved in the standard H CH 2 S 3 ¿e third panel in Figure 3.9.1 shows the mean estimated single prole precision and the measured Cl T variability (which includes instrument noise and atmospheric variability). ¿e precisions for v4.2 x and CH S /v3.4 x are nearly identical except for pressures greater than 68hPa where v4.2 v3.3 x x is producing lower val- 3 CN O linewidth. ues. Version 2 used a more coarse retrieval grid above 22hPa and a dierent value of the H 2 T Together these account for H O precision dierences at these altitudes between v2 and later versions. 2 CH S Figures 3.9.2 and 3.9.3 show a comparison of H O zonal means from v2, v3, and v4 retrievals for 100– 2 3 OH O tends to show only minor departures from v3. Ver- 2.6hPa and 2.2–0.002-hPa, respectively. Version 4 H 2 T O shows some vertical oscillatory behavior at some levels that was removed when an improved sion 2 H 2 ClO S linewidth was used in v3. ¿e latitude structure tends to be very similar in all cases except at 0.002hPa for T v2.2. S CO T 3.9.3 Resolution GPH S ¿e spatial resolution is obtained from examination of the averaging kernel matrices shown in Figure 3.9.4. T ¿e vertical resolution for H O is in the range 1.3–3.6km from 316–0.22hPa and degrades to 6–11km for 2 H S 170–350km for pressures greater pressures lower than 0.22hPa. ¿e along track horizontal resolution is ~ 2 O T than 4.6hPa, and degrades to 400–740km at smaller pressures. ¿e resolutions in Table 3.9.1 are a smoothed ○ HCl S N values shown in Figure 3.9.4. ¿e horizontal cross-track resolution is average of the equatorial and 70 7km, the width of the MLS 190-GHz eld-of-view for all pressures. ¿e longitudinal separation of the MLS T ○ ○ HCN S –20 measurements is 10 over middle and lower latitudes, with much ner sampling in polar regions. T 3.9.4 Precision HNO S O data. For pressures hPa, the H 100 Table 3.9.1 summarizes the estimated precision of the MLS v4.2 x ≥ 2 3 T 1 precisions given are the scatter about the mean of coincident comparison dierences, which are larger σ HO S σ 1 hPa the precisions are the 83 ≤ , 2007]. For pressures Read et al. than the formal retrieval precisions [ 2 T Lambert et al. scatter of coincident ascending/descending MLS prole dierences [ , 2007]. ¿e individual HOCl S Level 2 precisions are set to negative values or zero in situations when the retrieved precision is larger than 50% of the a priori precision – an indication that the data are biased toward the a priori value. T IWC S 3.9.5 Accuracy T IWP ¿e values for accuracy are based primarily on two sources: comparisons with validated instruments and a S full systematic error analysis performed on the MLS measurement system as described (for v2, but the same T ) in Read et al. [2007] and Lambert et al. [2007]. For pressures between approach has been used for v4.2 x N S 2 O 316–178hPa, comparisons between AIRS v6 and MLS v4.2 x % agreement in the zonal mean (MLS 20 show < T ○ usually drier for low concentrations and wetter for high concentrations) equatorward of 40 with MLS having S O 3 large dry biases at higher latitudes. T ¿e values in Table 3.9.1 for pressures between 316 and 178hPa are AIRS validated accuracies which are OH S better than those theoretically possible for the MLS measurement system. For the pressure range 178–83hPa, T the quoted values come directly from the systematic error analysis performed on the MLS measurement RHI S system. Comparisons between MLS and frost point hygrometers show that, when the tropopause is between T O at the tropopause level tends to have a large (20–30%) dry biases. Future work 215 and 147hPa, MLS H 2 SO S to understand this behavior is planned. An estimate of the accuracy between 121–83hPa is also from the 2 T S T Aura Microwave Limb Sounder (MLS) x 65 Level 2 Version 4.2 Quality T

72 Help 3.9. Water Vapor (H O) 2 Overview Table Zonal Means for Data Over March, 2009 BrO S H2O, v03.3x H2O, v04.2x H2O, v02.2x T CH S 100 hPa 56 hPa 83 hPa 68 hPa 3 5.0 5.0 5.0 5.0 Cl 4.8 4.5 4.5 T 4.5 4.6 4.0 CH S 4.0 3.5 4.0 4.4 3 CN O / ppmv 3.5 2 4.2 3.0 H 3.5 T 3.0 2.5 4.0 CH S 3.8 3.0 2.0 2.5 o o o o o o o o o o o o o o o o 3 45 EQ 45 45 N 90 90 N N S 45 N S EQ 45 90 N 45 S 90 90 N N 90 EQ S 45 N 90 S S 90 S EQ S 45 OH 46 hPa 38 hPa 26 hPa 32 hPa 5.2 5.5 5.5 6.0 T 5.0 ClO S 5.0 5.5 5.0 4.8 4.5 5.0 T 4.5 4.6 O / ppmv 4.5 4.0 S 2 CO 4.4 H 4.0 4.0 3.5 4.2 T 3.0 4.0 3.5 3.5 GPH S o o o o o o o o o o o o o o o o N 45 90 S N 90 90 N S 45 S S EQ 45 45 N 90 45 N S EQ 90 45 S 45 EQ S EQ 45 N N 90 S N 90 90 18 hPa 22 hPa 12 hPa 15 hPa T 6.0 6.0 6.0 6.0 H S 5.5 2 5.5 5.5 5.5 O 5.0 T 5.0 5.0 5.0 HCl S O / ppmv 4.5 2 H 4.5 4.5 4.5 4.0 T 4.0 4.0 3.5 4.0 HCN S o o o o o o o o o o o o o o o o 90 EQ S 45 45 S EQ 45 S 45 90 N N 90 N N 45 90 90 N 90 N EQ 90 N S 45 EQ S 45 45 S N 90 S S 8.3 hPa 6.8 hPa 10 hPa 5.6 hPa T 6.0 7.0 6.5 6.5 HNO S 6.5 5.5 6.0 6.0 3 6.0 T 5.5 5.0 5.5 O / ppmv HO 5.5 S 2 H 5.0 4.5 5.0 2 5.0 T 4.0 4.5 4.5 4.5 HOCl S o o o o o o o o o o o o o o o o 45 S S EQ 45 S N 90 EQ 90 45 N N N 90 90 N EQ N 90 45 N S 90 45 S S 45 45 S EQ 45 S N 90 90 2.6 hPa 3.2 hPa 3.8 hPa 4.6 hPa 8.0 7.5 7.5 7.0 T IWC S 6.5 7.0 7.5 7.0 6.5 6.0 6.5 7.0 T O / ppmv 6.0 5.5 6.5 6.0 IWP 2 S H 5.5 5.5 6.0 5.0 T 5.0 5.0 5.5 4.5 N o o o o o o o o o o o o o o o o S S S EQ 45 N N 90 45 N 90 90 90 S S 45 45 S EQ 45 S N 90 S N EQ 45 EQ N 90 45 N S 90 N 90 45 2 O Latitude / Degrees Latitude / Degrees Latitude / Degrees Latitude / Degrees T S O 3 T Figure 3.9.2: A comparison of v4.2x (red) to v3.3x/v3.4x (blue) and v2.2 (green) water vapor zonal means for March OH S 2009. Each panel represents a pressure as noted above. The pressures shown range from 100 hPa to 2.6 hPa. T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 66 Level 2 Version 4.2 Quality T

73 Help 3.9. Water Vapor (H O) 2 Overview Table Zonal Means for Data Over March, 2009 BrO S H2O, v02.2x H2O, v04.2x H2O, v03.3x T CH S 1.8 hPa 1.2 hPa 2.2 hPa 1.5 hPa 3 8.5 8.0 8.0 7.5 Cl T 7.5 7.0 8.0 7.5 CH S 3 7.0 CN 7.0 6.5 7.5 O / ppmv T 2 6.5 H CH S 6.5 7.0 6.0 6.0 3 OH 6.0 5.5 5.5 6.5 T o o o o o o o o o o o o o o o o S EQ 45 N N 45 90 N N 90 S N 90 90 S 45 45 S EQ 45 S N 90 45 N 90 S EQ 45 S N 90 90 S 45 EQ ClO S 1.0 hPa 0.68 hPa 0.32 hPa 0.46 hPa 8 8.5 8.0 8.5 T 7.5 7 S CO 8.0 8.0 7.0 6 T 7.5 6.5 7.5 GPH S O / ppmv 2 5 H 6.0 T 7.0 7.0 4 5.5 H S 2 O 6.5 5.0 3 6.5 o o o o o o o o o o o o o o o o T N 90 S 45 S EQ 45 N 90 N 90 S 45 S EQ 45 N 90 N 90 S S EQ 45 90 N 90 N 90 S 45 S EQ 45 N 45 HCl 0.15 hPa 0.22 hPa 0.10 hPa 0.046 hPa S 8 8 8 8 T HCN 6 6 6 6 S T 4 4 4 4 HNO O / ppmv 2 S H 2 2 2 2 3 T HO S 0 0 0 0 o o o o o o o o o o o o o o o o N S EQ 45 90 N 90 90 S S 45 N S EQ 45 EQ N 90 45 N 90 N 90 S 90 N S 45 S S EQ 45 45 N 90 45 2 T 4.64E-03 hPa 0.022 hPa 0.010 hPa 2.15E-03 hPa 5 1.8 3.0 6 HOCl S 1.6 5 2.5 4 T 1.4 IWC 4 2.0 S 3 1.2 O / ppmv 2 T 3 1.5 H 1.0 IWP S 2 1.0 2 0.8 T 1 1 0.6 0.5 N o o o o o o o o o o o o o o o o S S 45 45 S 45 90 S S EQ 45 N N 90 90 90 N S N N 90 N EQ 45 EQ S 90 45 S N 90 45 90 N S 45 EQ 2 O Latitude / Degrees Latitude / Degrees Latitude / Degrees Latitude / Degrees T S O 3 T Figure 3.9.3: A comparison of v4.2x (red) to v3.3x/v3.4x (blue) and v2.2 (green) water vapor zonal means for March OH S 2009. Each panel represents a pressure as noted above. The pressures shown range from 2.2 hPa to 0.002 hPa. T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) 67 Level 2 Version 4.2 x Quality T

74 Help 3.9. Water Vapor (H O) 2 Overview Table Equator FWHM / km FWHM / km 800 600 400 200 0 1000 1200 -2 0 2 4 6 8 10 12 0.1 BrO S T CH S 1.0 3 Cl T CH S 3 CN 10.0 T Pressure / hPa CH S 3 OH 100.0 T ClO S T 1000.0 1.0 -4 0.6 1.2 -2 0 2 0.2 -0.2 0.4 0.0 0.8 4 S CO Profile number Kernel, Integrated kernel 0 70 N T FWHM / km FWHM / km GPH 200 -2 600 800 1000 1200 0 2 4 6 8 10 12 400 0 S 0.1 T H S 2 O 1.0 T HCl S T HCN 10.0 S T Pressure / hPa HNO S 100.0 3 T HO S 2 1000.0 T HOCl -2 1.2 2 0.8 0.6 0.4 0.2 0.0 -0.2 -4 4 0 1.0 S Kernel, Integrated kernel Profile number T IWC S O Typical two-dimensional (vertical and horizontal along-track) averaging kernels for the MLS v4.2x H Figure 3.9.4: 2 T ◦ data at the equator (upper) and at 70 N (lower); variation in the averaging kernels is suciently small that these IWP S are representative of typical proles. Colored lines show the averaging kernels as a function of MLS retrieval level, T indicating the region of the atmosphere from which information is contributing to the measurements on the indi- N S vidual retrieval surfaces, which are denoted by plus signs in corresponding colors. The dashed black line indicates 2 O the resolution, determined from the full width at half maximum (FWHM) of the averaging kernels, approximately T scaled into kilometers (top axes). (Left) Vertical averaging kernels (integrated in the horizontal dimension for ve S O 3 along-track proles) and resolution. The solid black line shows the integrated area under each kernel (horizontally T and vertically); values near unity imply that the majority of information for that MLS data point has come from the OH S measurements, whereas lower values imply substantial contributions from a priori information. (Right) Horizon- T tal averaging kernels (integrated in the vertical dimension) and resolution. The horizontal averaging kernels are RHI S shown scaled such that a unit averaging kernel amplitude is equivalent to a factor of 10 change in pressure. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 68 Level 2 Version 4.2 Quality T

75 Help 3.9. Water Vapor (H O) 2 Overview systematic error analysis performed on the MLS measurement system. Comparisons among in situ sensors on the WB-57 high altitude aircra and frostpoint hygrometers own on balloons show 30% disagreements Table – well in excess of the estimated accuracy of each instrument including MLS – near the tropopause and lower stratosphere. ¿e balloon based frost point hygrometer shows agreement better than indicated in Table 3.9.1. Read ¿e validation paper describes in detail why a 30% spread is inconsistent with the MLS measurements [ BrO S , 2007]. For pressures less than 83hPa, the accuracy is based on the systematic error analysis. et al. T Please note the discussion of a potential dri in the MLS water vapor measurements discussed below. CH S 3 Cl 3.9.6 A note about the water vapor averaging kernels T CH S ¿e averaging kernels are most applicable to linear and moderately non-linear retrieval problems. ¿e MLS 3 CN H O retrievals are mostly linear, except when the atmosphere approaches opaqueness. ¿is occurs when the 2 T limb tangent of the instrument eld of view is less than 10km above the Earth’s surface and the atmosphere is CH S moist. In such cases the measurements are very nonlinear and o en times the averaging kernel calculations 3 OH for the lowest retrieved levels are numerically unstable, and unrepresentative of the actual MLS measurement T system. ¿is is the case, for example, for the 316-hPa and 262-hPa equatorial kernels shown in Figure 3.9.4. ClO S Analysis of simulated MLS observations indicates that application of the averaging kernel provides little ben- T et to analyses involving the 316-hPa and 262-hPa levels, and accordingly we recommend not applying the S CO (unrepresentative) kernels at those levels. T GPH S 3.9.7 Review of comparisons with other datasets Figure 3.9.5 shows a latitude-value zonal mean comparison among several satellite data sets. ¿e satellite T H datasets include MLS v4.2 x , AIRS v6, ACE-FTS v3.5, MIPAS IMK v4, HALOE v19, Odin SMR continuum S 2 O O.ACE-FTSandMLSshowthebestagreementwitheachotherthrough- H O,andOdinSMRlineresolvedH 2 2 T out the full vertical range and over most latitudes. MLS also agrees well with AIRS v6 over the lower latitudes. HCl S MLS however does have a signicant dry bias at high latitudes which is o en a factor of two. MIPAS-IMK T at most altitudes and latitudes show similar behavior to MLS except near the cold point tropopause where HCN S MLS shows lower values in the tropics whereas MIPAS does not. HALOE is much too dry in the tropics but T 100 behaves like MLS for pressures hPa. ¿e Odin SMR continuum product (100–68 hPa) does not agree ≤ HNO S wellwiththeothersatellitebasedproducts. ¿eOdin-SMRlineresolvedproduct(pressureslessthanorequal 3 to 46hPa) usually agrees within 10% of MLS v4 for all heights and latitudes. T HO S x is shown in Figures 3.9.6 and 3.9.7 that compare multi-month maps One potential improvement in v4.2 2 of MLS with AIRS v6. ¿e very moist regions over the tropics in v4 have become a bit drier and in better T HOCl S agreement with AIRS. and MLS x O is similar to the MLS v3.3 x H x Apart from the dierences noted above, the MLS v4.2 /v3.4 2 T Lambert et al. [2007] and Read et al. v2.2 products, the latter described and validated in [2007]. ¿e H O 2 IWC S validation paper used AIRS v4 and MLS v2 humidities and both of them have become more dry in their T subsequent versions. A revised validation paper for H O is not planned in the near future and users are 2 IWP S encouraged to read Read et al. [2007] and [2007] for more information. MLS v4.2 x is a part of Lambert et al. T the second SPARC Water Vapor Assessment (WAVAS-II) activity. N S 2 O 3.9.8 Potential drift in lower stratospheric water vapor T S O Hurst et al. [2016] discuss how comparisons between MLS lower stratospheric water vapor observations and 3 1 instruments from 2004 to 2015 reveal signs of a potential dri be- those from balloon-borne CFH and FPH T -1 OH S tweenthetwosetsofmeasurements, withMLSwatervaporincreasingatarateofaround0.03–0.07ppmvyr relative to the frostpoint sondes (roughly 0.6 to 1.5% per year), starting around 2009. Comparisons with the T ACE-FTS instrument, and with measurements of upper stratospheric water vapor from ground-based mi- RHI S crowave sensors, provide evidence for a smaller ( ~ +0.2% per year) dri . Investigations by the MLS team have T SO 1 S Cryogenic Frostpoint Hygrometer (CFH) and Frost Point Hygrometer (FPH) 2 T S T Aura Microwave Limb Sounder (MLS) x 69 Level 2 Version 4.2 Quality T

76 Help 3.9. Water Vapor (H O) 2 Overview 316 hPa 383 hPa 261 hPa 215 hPa 1000 1000 10000 Table 100 100 1000 100 BrO S 10 10 10 100 T CH S 3 1 1 1 10 Cl 50 0 50 -50 -50 -50 0 50 0 50 -50 0 T CH 100 hPa 121 hPa 179 hPa 147 hPa S 100 20 8 10 3 CN 8 6 15 T CH 6 S 4 10 10 3 OH 4 T 2 5 2 ClO S 0 0 1 0 -50 -50 50 0 50 -50 0 0 50 50 0 -50 T 68 hPa 46 hPa 56 hPa 83 hPa S CO 6 7 6 7 T 6 6 5 5 GPH S 5 5 4 4 T VMR (ppmv) 4 4 H S 2 O 3 3 3 3 T 2 2 2 2 HCl S -50 -50 -50 0 50 0 50 50 0 50 0 -50 T 26 hPa 22 hPa 32 hPa 38 hPa 7 7 7 7 HCN S 6 6 6 6 T HNO 5 5 5 5 S 4 4 4 4 3 T 3 3 3 3 HO S 2 2 2 2 2 T 50 -50 -50 50 0 50 50 0 -50 0 -50 0 HOCl S 15 hPa 12 hPa 18 hPa 10 hPa 7 7 7 7 T 6 6 6 6 IWC S 5 5 5 5 T 4 4 4 4 IWP S 3 3 3 3 T 2 2 2 2 N S 2 50 -50 50 0 -50 0 -50 0 50 -50 50 0 O T S Latitude O 3 T OH S A comparison of MLS v4.2x (red) water vapor for Jan-Feb-Mar 2005 with other satellite observations Figure 3.9.5: T shown as latitude-value zonal means. Each panel represents a pressure surface. The satellites are: AIRS v6 (dark RHI S blue), ACE-FTS v3.5 (light blue), MIPAS IMK v4 (yellow-green), HALOE v19 (cyan), Odin SMR 544 GHz continuum product (orange open diamonds), and Odin SMR line resolved product (orange solid bullets). T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 70 Level 2 Version 4.2 Quality T

77 Help 3.9. Water Vapor (H O) 2 Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T ClO S T S CO T GPH S T H S 2 O T HCl S T HCN S T HNO S 3 T HO S 2 T HOCl S T IWC S T IWP S T N S 2 O T S O 3 T OH S Mapped elds from AIRS v6 (left), MLS v2.2 (center-left), MLS v3.3x/v3.4x (center-right), and MLS v4.2x Figure 3.9.6: T RHI (right) pressures between 300–150 hPa for January and February 2005. Note that the AIRS measurement is mostly S O greater than 10 ppmv. suitable for H 2 T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 71 Level 2 Version 4.2 Quality T

78 Help 3.9. Water Vapor (H O) 2 Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T ClO S T S CO T GPH S T H S 2 O T HCl S T HCN S T HNO S 3 T HO S 2 T HOCl S T IWC S T IWP S T N S 2 O T S O 3 T OH S Mapped elds from AIRS v6 (left), MLS v2.2 (center-left), MLS v3.3x/v3.4x (center-right), and MLS v4.2x Figure 3.9.7: T RHI (right) pressures between 300 – 150 hPa for June through August 2005. Note that the AIRS measurement is mostly S O greater than 10 ppmv. suitable for H 2 T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 72 Level 2 Version 4.2 Quality T

79 Help 3.9. Water Vapor (H O) 2 Overview indicated that this latter, smaller, dri is consistent with observed changes in instrument performance (cor- rection of which is a goal for future versions of the MLS data processing so ware). However, these observed Table instrument changes cannot (at least in our current understanding) account for the larger dri reported by . ¿ese issues are under active investigation by the MLS team. Hurst et al. BrO S 3.9.9 Data screening T Pressure range: 316–0.002hPa. CH S 3 Cl Values outside this range are not recommended for scientic use. T CH Estimated precision: Only use values for which the estimated precision is a positive number. S 3 CN a priori Values where the information has a strong inuence are agged with negative or zero precision, T and should not be used in scientic analyses (see Section 1.5). CH S Status ag: Only use proles for which the Status eld is an even number. 3 OH Status Odd values of indicate that the prole should not be used in scientic studies. See Section 1.6 T ClO eld. Status for more information on the interpretation of the S T Ignore status bit 16 (high cloud) or bit 32 (low cloud) set indicating the presence of clouds. See Clouds: S CO artifacts for more details. T than0.7shouldbeusedinscienticstudies. greater Quality Quality eld: Onlyproleswithavalueofthe GPH S , as described at the end of Note that this is a change from the value originally recommended for v4.2 x T this subsubsection. Also, note the “additional screening” information given below. H S 2 O less than 2.0 should be used in scientic Convergence Convergence eld: Only proles with a value of the T studies. HCl S Status Convergence , and Quality , Even a er application of the Additional screening to avoid outliers: T HCN O dataset. ¿ese x H screening described above, there remain some unphysical proles in the MLS v4.2 S 2 are characterized by unrealistically low water vapor mixing ratios in the upper troposphere and strato- T sphere. ¿ese should be excluded from analyses by simply neglecting any H O prole where the mix- HNO 2 S ing ratio at any pressure at or larger than 1hPa (i.e., altitudes below that pressure level) is less than 3 T 0.101ppmv. Note that this additional screening is an additional step beyond those originally recom- HO S mended for v4.2 O. H x 2 2 T HOCl S Updated Quality screening recommendation O (see A byproduct of the aging in MLS that gives rise to the dri in the water vapor product and 190-GHz N 2 T O product exhibits a decreasing trend (most notable in the metric for the H Quality section 3.17) is that the IWC 2 S years following ~ 2009). ¿is decrease is not correlated with any actual decline in the accuracy of the MLS T O product, which appears largely unchanged (dri issues aside). However, application of the previously H 2 IWP S threshold leads to rejection of an unreasonably large fraction of the H O proles Quality recommended 1.45 2 T Quality in the more recent years of the Aura mission, particularly during southern winter. ¿e revised 0.7 N S 2 O threshold restores these proles without erroneously identifying any obviously poor proles as acceptable. T O observations agged as acceptable according to the “old” and Figure 3.9.8 compares the amounts of H 2 S O “new” screening methods over the life of the Aura mission to date. ¿e new screening method rejects fewer 3 T O proles than the old one. In addition the old method rejects more proles in the later part of the mission H 2 OH S than in the earlier years, a behavior the new method avoids. T RHI S 3.9.10 Artifacts O measurements become unreliable. ¿is is given in Ta- ¿ere is a minimum concentration where MLS H T 2 SO S O” column. ¿e lowest allowable H O value is 0.1ppmv. Dierences between ble 3.9.1 under the “Min. H 2 2 2 T S T Aura Microwave Limb Sounder (MLS) x 73 Level 2 Version 4.2 Quality T

80 Help 3.9. Water Vapor (H O) 2 Overview 100 Table 95 BrO S T 90 CH S 3 Cl T Good Data Yield (%) 85 CH S 3 CN T 80 CH S 2018 2016 2014 2012 2010 2006 2008 2004 3 OH T ClO S A time series of daily good data yields using the previous data quality screening for v04.2x (grey) and Figure 3.9.8: the current revised quality and prole value screening method (black). T S CO T O constraint and the true atmospheric state can cause errors at 316 and the retrieved middle tropospheric H 2 GPH S < 261hPa. ¿e error manifests either as dry ( 1ppmv) or moist spikes in an orbital time series. Such data are T o en accompanied with good quality and status. H S < Clouds in the eld of view degrade the data in unpredictable ways. Many instances of quality 1.45 occur 2 O in the presence of clouds where the cloud screening has not rejected the radiances. Cloud radiative scattering T HCl S distorts the spectral lineshape causing poor ts and low quality values. However, not all MLS signals are status O ( bit 16 or 32 set between 316– obviously aected. Coincident comparisons of MLS cloud agged H 2 T HCN 215hPa) with good quality AIRS show a small mean bias of 10% but exhibit a 50% increase in variability for S theindividualdierences. ¿ereforeusersshouldbeawarethat, althoughtheoverallbiasesformeasurements T HNO inside clouds are similar to that for clear sky, individual proles will exhibit greater variability about the true S atmospheric humidity. 3 T Note the discussion of the dri in MLS stratospheric water vapor above. HO S 2 T 3.9.11 Desired improvements for future data version(s) HOCl S We want to improve high latitude performance of 261 and 215 hPa H O in future work. Possibly a vertically 2 hPa at high latitudes during dry conditions. 316 > O product for pressure resolved H 2 T IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 74 Level 2 Version 4.2 Quality T

81 Help 3.9. Water Vapor (H O) 2 Overview Table BrO S T Summary of MLS v4.2x H Table 3.9.1: O product. 2 CH S 3 a Cl Min. / Accuracy / Pressure / Resolution Precision Comments b T ppmv % hPa H / km V × / % CH S 3 Unsuitable for scientic use 0.001 — — — ≤ — CN 10.3 0.002 40 0.1 350 152 × T CH 585 81 23 0.004 × 10.8 0.1 S 3 OH 0.010 8.8 × 725 55 17 0.1 × 745 41 19 0.1 0.021 8.0 T ClO S 0.1 35 540 × 7.4 0.046 17 5.8 × 745 0.10 14 0.1 24 T S CO 0.1 0.21 3.6 × 670 19 11 × 0.46 505 12 8 0.1 3.3 T GPH 2.5 × 400 6 7 0.1 1.0 S 2.2 3.5 × 385 4 7 0.1 T 3.4 × 350 4.6 5 0.1 4 H S 2 O 2.8 × 290 6 19 0.1 10 T 0.1 265 5 22 5 3.2 × HCl S 0.1 4 5 × 3.2 46 230 T 190 × 68 5 9 0.1 3.1 HCN S 0.1 7 9 190 × 3.1 83 T 8 3.0 × 100 15 198 0.1 HNO S 2.6 × 193 20 12 0.1 121 3 2.3 15 20 147 0.1 188 × T HO S 3 25 183 × 20 1.7 178 2 ○ T 60 Large low bias for latitudes > 1.6 178 40 25 3 215 × HOCl S ○ 60 > Large low bias for latitudes 173 × 1.4 261 35 20 4 T Occasionallyerroneouslowvalue IWC S < 1 ppmvandhighvalueiersare T 7 316 × 1.3 168 65 15 retrievedinthetropics, usuallyin IWP S clouds. T N Unsuitable for scientic use — — — 316 > — S 2 O T a Precision for a single MLS prole S O b x MLS. Minimum H O is an estimate of the minimum H O concentration measurable by v4.2 3 2 2 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 75 Level 2 Version 4.2 Quality T

82 Help Overview 3.10 Hydrogen Chloride (HCl) Table Swath name: HCl 100 – 0.32 hPa Useful range: BrO S [email protected] Lucien Froidevaux, < Contact: Email: > T CH S 3 Cl 3.10.1 Introduction T CH ¿ere has been very little change in the v4.2 HCl retrieval results, in comparison to v3.3/v3.4. We provide S 3 below sample mean HCl distributions for the two data versions, and their dierences. Otherwise, previous CN information regarding this species remains largely unchanged; the main points are mentioned here mainly T CH S for new data users. 3 OH ¿e MLS v3.3/v3.4 retrievals of the HCl standard product (from the 640GHz radiometer) use channels T from band 14, as a result of the deterioration observed since early 2006 in nearby band 13, originally targeted ClO S (with narrower channels than band 14) at the main HCl emission line center. Full measurement days with T band 13 on a er February 15, 2006, are as follows: March 15, 2006 (2006d074), April 14, 2006 (2006d104), S CO January6through8,2009(2009d006through2009d008),andJanuary24through27,2010(2010d024through 2010d027). For days prior to February 16, 2006 and for the few days (as listed above) when band 13 gets T GPH S HCl-640-B13 product (stored in the turned on therea er, the MLS Level 2 so ware also produces a separate le), using the band 13 radiances. ¿is product has slightly better precision and vertical resolution L2GP-DGG T H in the upper stratosphere than the standard HCl product. ¿e MLS team plans to turn band 13 on very S 2 O infrequently (possibly only one more time), as its lifetime is estimated at a few days to a few weeks at most, T based on the channel counts and channel noise characteristics observed during the 3-day turn-on period HCl S in late January, 2010; band 13 should provide useful trend information for upper stratospheric data, given T its narrower channels. Upper stratospheric trends from the (uninterrupted from 2004 to present) band 14 HCN S retrievals are not reliable enough and are too small, compared to band 13 data and expectations, as well as T versus ACE-FTS HCl data. Scientic usage of the MLS standard HCl (band 14) dataset should therefore be HNO S restricted to the lower stratosphere; in this region, studies of tendencies and longer-term trends are justied 3 (e.g., see the HCl comparisons discussed by Mahieu et al. [2014]). T HO S See Figure 3.10.1 for an illustration of the trend dierences between these two MLS band measurements 2 of HCl. In the lower stratosphere, however, variations in the two HCl products are closer together, although T HOCl there is also more seasonal variability; we believe that the band 14 retrievals are quite appropriate to use in S studiesofseasonal(andlatitudinal)changes(e.g., duringpolarwinter/spring),aswellasforlonger-termtrend T studies in the lower stratosphere. IWC S Table 3.10.1 summarizes the MLS HCl resolution, precision, and accuracy estimates as a function of pres- T sure. More discussion and data screening recommendations for the MLS HCl v4.2 data are provided below IWP S (but this is unchanged from the v3.3/v3.4 recommendations). Analyses describing detailed validation of the T [2008a]. Based MLS (v2.2) product and comparisons with other data sets are described in Froidevaux et al. N S 2 on the fairly small overall changes in v3.3/v3.4 and v4.2 HCl data versus v2.2, the conclusions of the latter O T reference should remain essentially unchanged. S O 3 3.10.2 Changes from v3.3/v3.4 T OH S ¿ere were no large algorithmic changes relating to HCl for the v4.2 retrievals. Small dierences in HCl abundances (see below) have occurred mainly near 147hPa, most likely mainly as a result of changes in water T RHI S vapor in the UTLS. ¿e background observed in the 640-GHz radiances includes emissions from N O. ¿ere , and H , O 2 2 2 T SO are laboratory-based and ground-based models for the continuum absorptions that are the basis for the MLS S 2 T S T Aura Microwave Limb Sounder (MLS) 76 Level 2 Version 4.2 x Quality T

83 Help 3.10. Hydrogen Chloride (HCl) Overview 0.46 hPa ; Latitudes: -90. to 90. 3.8 Table 3.7 3.6 BrO S 3.5 T CH S 3.4 HCl / ppbv 3 Cl 3.3 T CH S 3 3.2 CN T 3.1 CH 2004 2005 2006 2007 2008 2009 S 3 OH Daily zonal averages for MLS HCl at 0.46 hPa, from mid-August, 2004, through January, 2010, for the Figure 3.10.1: T originally-targeted band 13 measurements (red points), now available only on occasion (to preserve lifetime), and ClO S the band 14 data (blue points). The lines are simple linear ts through the daily data points; trend dierences are T apparent in this region of the atmosphere, where the information obtained from band 14 HCl data is not reliable S CO enough. T GPH S absorption model [ Pardo, 2001 , and references therein]. ¿ese models were tested against MLS extinction T measurementsfromthewingchannelsinthe640-GHzradiometer. ¿elatitudedependenceofthisextinction H S was found to agree better with the expected most plus dry continuum extinction values if the dry and moist 2 O continuum functions were scaled by factors close to 20%; however, this is not a change from the v3.3/v3.4 T HCl S retrievals. A comparison plot showing zonal average HCl contours and dierences between the two data versions T HCN S for a typical month (March, 2009) is provided in Figure 3.10.2. For pressures larger than or equal to 0.22hPa, the average dierences between the two data versions are typically within 0.1ppbv (or 1–2%). ¿e average T HNO changes are easily within the estimated accuracy values (see Table 3.10.1). ¿e largest percentage changes S in HCl occur for very small mixing ratio values, close to 100 and (mainly) 147hPa; v4.2 values are usually 3 T smaller than the v3.3/v3.4 values by 10 to 15% in the lower stratosphere at low latitudes or during winter at HO S polar latitudes. ¿ere is still, however, an unrealistic high bias in HCl at 147hPa in the tropics, so values at 2 ○ ○ T S to 40 N latitude range. ¿e this pressure are not recommended (reliable in absolute value) within the 40 HOCl S precisions estimated in the Level 2 les are essentially unchanged from v3.3/v3.4. T 3.10.3 Resolution IWC S Typical (rounded o) values for resolution are provided in Table 3.10.1. Based on the width of the averaging T kernels shown in Figure 3.10.3, the vertical resolution for the standard HCl stratospheric product is ~ 3km IWP S (2.7km at best in the lower stratosphere), or about double the 640-GHz radiometer vertical eld of view T width at half-maximum; the vertical resolution degrades to 4–6km in the lower mesosphere. ¿e along- N S 2 O ~ track resolution is ~ 500km in the lower mesosphere. ¿e 200 to 350km for pressures of 2hPa or more, and T cross-track resolution is set by the 3km width of the MLS 640-GHz eld of view. ¿e longitudinal separation S O ○ ○ 3 –20 over middle and lower latitudes, with much ner of MLS measurements, set by the Aura orbit, is 10 T sampling in polar regions. OH S T 3.10.4 Precision RHI S ~ 0.2 to 0.6ppbv in the ¿e estimated single-prole precision reported by the Level 2 so ware varies from T stratosphere(seeTable3.10.1),withpoorerprecisionobtainedinthelowermesosphere. ¿eseprecisionvalues SO S x have not changed signicantly for v4.2 data. ¿e Level 2 precision values are o en only slightly lower than 2 T S T Aura Microwave Limb Sounder (MLS) x 77 Level 2 Version 4.2 Quality T

84 Help 3.10. Hydrogen Chloride (HCl) Overview Table BrO S T CH S 3 Cl T CH Averages for March, 2009 S 3 CN HCl v3.3 HCl v4.2 T CH S 3 1.0 1.0 OH T ClO S 10.0 10.0 T Pressure / hPa Pressure / hPa S CO T 100.0 100.0 GPH o o o o o o o o o o o o S N 30 S S EQ 30 N 60 90 N N 60 N 60 30 EQ 90 90 N S 60 90 S 30 S S Latitude Latitude T H S 0.25 1.50 3.50 0.50 0.25 1.00 3.50 2.00 1.00 0.50 2.50 3.00 2.00 2.50 3.00 1.50 2 O HCl / ppbv HCl / ppbv T HCl S HCl Difference (v4.2-v3.3) HCl % Difference (v4.2-v3.3) T HCN S 1.0 1.0 T HNO S 10.0 10.0 3 T Pressure / hPa Pressure / hPa HO S 2 100.0 100.0 T HOCl o o o o o o o o o o o o S EQ N 90 30 N S N 60 60 N 90 90 90 N S 60 S S 30 30 S EQ 30 S N 60 Latitude Latitude T IWC 10 5 3 1 0 -1 -3 -5 -10 -0.06 0.06 0.03 0.00 -0.03 S HCl Difference / ppbv HCl Difference / % T IWP S Figure 3.10.2: Zonal averages for MLS HCl proles during March, 2009, showing the MLS v3.3 HCl mixing ratio contours (top left panel), the v4.2 contours (top right panel), and their dierences in ppbv (v4.2 minus v3.3, bottom T left panel) and percent (v4.2 minus v3.3 versus v3.3, bottom right panel). N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 78 Level 2 Version 4.2 Quality T

85 Help 3.10. Hydrogen Chloride (HCl) Overview Table Equator FWHM / km FWHM / km 800 600 400 200 0 1000 -2 0 2 4 6 8 10 12 1200 0.1 BrO S T CH S 3 Cl 1.0 T CH S 3 CN T 10.0 Pressure / hPa CH S 3 OH T ClO S 100.0 T -0.2 0.4 0.6 1.0 1.2 0.2 0.8 0.0 -4 4 2 -2 0 S CO Kernel, Integrated kernel Profile number 0 N 70 T FWHM / km FWHM / km GPH 200 1000 1200 0 600 10 8 6 12 400 800 4 -2 0 2 S 0.1 T H S 2 O T 1.0 HCl S T HCN S 10.0 T Pressure / hPa HNO S 3 T HO S 100.0 2 T HOCl 1.0 0.8 0.6 0.4 0.2 0.0 -0.2 0 -2 -4 1.2 4 2 S Kernel, Integrated kernel Profile number T IWC S Figure3.10.3: Typical two-dimensional (vertical and horizontal along-track) averaging kernels for the MLS v4.2x HCl T ◦ data at the equator (upper) and at 70 N (lower); variation in the averaging kernels is suciently small that these IWP S are representative of typical proles. Colored lines show the averaging kernels as a function of MLS retrieval level, T indicating the region of the atmosphere from which information is contributing to the measurements on the indi- N S vidual retrieval surfaces, which are denoted by plus signs in corresponding colors. The dashed black line indicates 2 O the resolution, determined from the full width at half maximum (FWHM) of the averaging kernels, approximately T scaled into kilometers (top axes). (Left) Vertical averaging kernels (integrated in the horizontal dimension for ve S O 3 along-track proles) and resolution. The solid black line shows the integrated area under each kernel (horizontally T and vertically); values near unity imply that the majority of information for that MLS data point has come from the OH S measurements, whereas lower values imply substantial contributions from a priori information. (Right) Horizon- T tal averaging kernels (integrated in the vertical dimension) and resolution. The horizontal averaging kernels are RHI S shown scaled such that a unit averaging kernel amplitude is equivalent to a factor of 10 change in pressure. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 79 Level 2 Version 4.2 Quality T

86 Help 3.10. Hydrogen Chloride (HCl) Overview the observed scatter in the data, as evaluated from a narrow latitude band centered around the equator where atmospheric variability is o en smaller than elsewhere, or as obtained from a comparison between ascending Table and descending coincident MLS proles. ¿e scatter in MLS data and in simulated MLS retrievals (using noise-free radiances) becomes smaller than the theoretical precision (given in the Level 2 les) in the upper and smoothing constraints. ¿e HCl stratosphere and mesosphere, where there is a larger impact of a priori BrO S precision values increase rapidly at pressures less than 0.2hPa, and are generally agged negative (or zero) at T pressures less than 0.1hPa; this indicates an increasing inuence from the a priori (with poorer measurement CH S 3 sensitivity and reliability). Cl T CH S 3.10.5 Accuracy 3 CN ¿e accuracy estimates in the Table came from a quantication of the combined eects of possible systematic T errors in MLS calibration, spectroscopy, etc. on the HCl retrievals; there are several error sources (mostly CH S from radiometric calibration components) that contribute signicantly to the total error. ¿ese values are 3 OH 2 σ estimates of accuracy. For more details, see the MLS validation paper by Froidevaux intended to represent T ), given the trend issues aecting the (band 14) standard HCl x and v3.4 et al. [2008a]. For v4.2 x (as for v3.3 x ClO S product in the upper stratosphere and lower mesosphere, we have recommended an accuracy estimate of no T better than 10% in this region (or about 0.3ppbv). For the lower stratosphere, given the agreement between S CO the two bands’ retrievals as well as some trend studies (for the standard product), we use the more formal T accuracy estimates (see Table 3.10.1). GPH S T 3.10.6 Data screening H S 2 Pressure range: 100–0.32hPa O T Valuesoutsidethisrangearenotrecommendedforscienticuse.WenotethattheMLSvaluesat147hPa HCl S are biased high, at least at low to mid-latitudes – and these values are not recommended (particularly ○ T equatorward of about 40 ). Also, although the vertical range at the top end is recommended up to HCN S 0.32hPa, users should note the signicant issues relating to HCl trend estimates in the upper strato- T sphereandlowermesosphere; averageprolesinthisregioncanbeusedforstudiesnotinvolvingtrends HNO S ). % (or accuracy requirements not as tight as 10 3 Estimated precision: Only use values for which the estimated precision is a positive number. T HO S information has a strong inuence are agged with negative or zero precision, a priori Values where the 2 and should not be used in scientic analyses (see Section 1.5). T HOCl S Status Status ag: Only use proles for which the eld is an even number. T Status Odd values of indicate that the prole should not be used in scientic studies. See Section 1.6 IWC S eld. Status for more information on the interpretation of the T than 1.2 should be used. greater eld Quality Quality eld: Only proles with a value of the IWP S ¿is criterion removes proles with the poorest radiance ts, typically less than 0.1% of the daily pro- T les. For HCl (and for other 640GHz MLS products), this screening correlates well with the poorly N S 2 O converged sets of proles (see below); we recommend the use of both the Convergence and Quality T elds for data screening. S O 3 less eld Convergence Convergence eld: Only proles with a value of the than 1.05 should be used. T For the vast majority of proles (99% or more for most days), this eld is less than 1.05. Nevertheless, OH S on occasion, sets of proles (typically one or more groups of ten proles, retrieved as a “chunk”) have T this Convergence eld set to larger values, and should be discarded. RHI S ¿ick clouds can add signicant artifacts (mainly in the tropics, statistically), with total systematic Clouds: T SO S errors potentially as large as 0.5 ppbv at 100 hPa and even larger at 147 hPa. Studies in this region could 2 T S T Quality x 80 Level 2 Version 4.2 Aura Microwave Limb Sounder (MLS) T

87 Help 3.10. Hydrogen Chloride (HCl) Overview Table 3.10.1: Summary for MLS hydrogen chloride Precision Resolution b Comments Table Accuracy Pressure a H × V ppbv ppbv % % hPa km Unsuitable for scientic use — — 0.22–0.001 — — — BrO S Unsuitable for trend studies 25 × 450 0.3 10 0.32 5 0.8 T Unsuitable for trend studies 5 × 400 0.5 10 0.7 20 0.3 CH S Unsuitable for trend studies × 300 0.3 1 0.5 15 4 10 3 Cl Unsuitable for trend studies 2 0.4 15 3 × 250 10 0.3 T Unsuitable for trend studies 3 0.3 200 × 0.3 10 10 5 CH S 3 0.2 10 3 × 200 0.2 10 10 CN 0.2 15 3 20 200 0.15 10 × T 46 0.2 10 to > 40 250 3 20 × 0.2 CH S 0.2 15 to > 80 3 68 300 0.2 25 × 3 OH 350 0.3 30 to > 100 3 × 40 0.2 100 T High bias at low lats. (use ClO 100 > 100 3 147 0.4 50 to 0.3 50 to 400 × > S with caution elsewhere) T Unsuitable for scientic use 1000–215 — — — — — S CO a % ) for individual proles; note that Precision ( 1 values tend to vary strongly with latitude in the lower stratosphere. σ T b σ 2 estimate from systematic uncertainty characterization tests (but see text for estimates at pressures lower than 10hPa); note GPH S that percent values tend to vary strongly with latitude and season in the lower stratosphere, due to the variability in HCl. T H S 2 O benet from additional screening that correlates any outliers in the data with the occurrence of thick T clouds (using some subset of the proles with set cloud status ags, namely status bits 16 and 32, or HCl S using the MLS retrievals of ice water content). However, the large positive HCl bias at 147 hPa for low T latitudes likely arises in part from unknown or improperly modeled systematic error sources, and not HCN S just from clouds. At other pressures, the potential impact on HCl from clouds is small or negligible. T HNO S 3.10.7 Review of comparisons with other datasets 3 Froidevaux et al. [2008a] provided results of generally good comparisons between MLS HCl and other satel- T HO S lite, balloon, and aircra measurements. Both MLS and ACE-FTS HCl values are generally larger (by about 2 10–15%) than the HCl values from HALOE, especially at upper stratospheric altitudes; this feature has not T HOCl changed, overall, with the new data version(s) from both MLS and ACE-FTS. MLS HCl at 147hPa is biased S high versus WB-57 aircra in-situ (CIMS) measurements (low to mid-latitudes); while this is still true for v4 T data, MLS data at this pressure level are potentially useful and accurate enough at high latitudes. IWC S T 3.10.8 Artifacts IWP S We do not recommend the use of the MLS HCl standard product in the upper stratosphere and lower • T mesosphere, especially in terms of trend studies, for reasons mentioned above. N S 2 O ¿e HCl values at 147hPa are biased high and generally not usable (except possibly at high latitudes). • T Please consult with the MLS team for further information. S O 3 • Users should screen out the non-converged and poorest quality HCl proles, as such proles (typically T lessthan0.1%ofthedata)tendtobehaveunlikethemajorityoftheotherMLSretrievals. Seethecriteria OH S listed above. T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 81 Level 2 Version 4.2 Quality T

88 Help Overview 3.11 Hydrogen Cyanide (HCN) Table Swath name: HCN 21 – 0.1 hPa Useful range: BrO S [email protected]k < Contact: Hugh C. Pumphrey, Email: > T CH S 3 Cl 3.11.1 Introduction T CH HCN is retrieved from bands encompassing, in the lower sideband, the 177.26 GHz spectral line of HCN. Al- S 3 CN though the target line is in an uncluttered part of the spectrum, the upper sideband contains many interfering and HNO HCN product is not recommended for general use in the lower lines of O x . As a result, the v4.2 T 3 3 CH S stratosphere. In the recommended range it is usable, but has rather poor precision and resolution. 3 OH It is possible to retrieve weekly zonal means of HCN over a greater vertical range by rst averaging the ra- T Pumphrey diances. ResultsofthisprocessandfurtherinformationontheHCNmeasurementmaybefoundin ClO S et al. [2006]. T S CO 3.11.2 Dierences between v3.x and v4.2 x No changes specic to the HCN retrieval were made between v2.x and v3.x. For v4.2 x a new phase was T GPH S introduced in which HCN is retrieved using only bands 6 and 27, rather than all of R2. ¿e v4.2 x HCN still has obvious biases in the lower stratosphere, but they are less severe than in previous versions. As a result, T H S the lowest recommended level is now 21hPa rather than the 10hPa recommended for earlier versions. 2 O x Figure 3.11.2 shows that the precisions in v4.2 are essentially unchanged from earlier versions. ¿e re- T trieved mixing ratios change very little in the region where use was recommended for earlier versions. Mixing HCl S ratios are considerably dierent between all three versions in the lower stratosphere where the data are not T x recommended for general use. However, the values in v4.2 appear less unrealistic than in either v2.x or v3.x. HCN S T 3.11.3 Vertical resolution HNO S ¿e HCN signal is rather small, so a rather strong smoothing constraint has to be applied to ensure that the 3 T retrieval is at all useful. As Figure 3.11.1 shows, the vertical resolution is about 8km at 10hPa, degrading to HO S 12km at 0.1hPa. ¿e horizontal resolution along the measurement track is between 2 and 4 prole spacings. 2 T HOCl S 3.11.4 Precision ), together with the observed L2gpPrecision Figure 3.11.2 shows the estimated precision (values of the eld T standard deviation in an equatorial latitude band where the natural variability of the atmosphere is small. ¿e IWC S observed scatter is smaller than the estimated precision due to the eects of retrieval smoothing. T IWP S 3.11.5 Accuracy T ¿e accuracy of the HCN product has not been assessed in detail because a cursory inspection revealed that N S 2 O previous versions of the product had extremely large systematic errors in the lower stratosphere. ¿ese errors T are smaller in v4.2 x , but remain present and un-quantied. For this reason the data are not considered to S O be useful at pressures greater than 21hPa (altitudes below 27km). In the upper stratosphere the values are ~ 3 T in line with current understanding of the chemistry of HCN. Comparison to historical values suggests an OH S accuracy of no worse than 50%. ¿e precision, resolution and accuracy of the HCN data are summarized in T table 3.11.1. RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 82 Level 2 Version 4.2 Quality T

89 Help 3.11. Hydrogen Cyanide (HCN) Overview Table Equator FWHM / km FWHM / km 800 600 400 200 0 1000 -2 0 2 4 6 8 10 12 1200 0.1 BrO S T CH S 3 Cl 1.0 T CH S 3 CN T Pressure / hPa CH S 10.0 3 OH T ClO S T 100.0 -0.2 0.4 0.6 1.0 1.2 0.2 0.8 0.0 -4 4 2 -2 0 S CO Kernel, Integrated kernel Profile number 0 N 70 T FWHM / km FWHM / km GPH 200 1000 1200 0 600 10 8 6 12 400 800 4 -2 0 2 S 0.1 T H S 2 O T HCl S 1.0 T HCN S T Pressure / hPa HNO 10.0 S 3 T HO S 2 100.0 T HOCl 1.0 0.8 0.6 0.4 0.2 0.0 -0.2 0 -2 -4 1.2 4 2 S Kernel, Integrated kernel Profile number T IWC S Figure 3.11.1: Typical two-dimensional (vertical and horizontal along-track) averaging kernels for the MLS v4.2x T ◦ HCN data at the equator (upper) and at 70 N (lower); variation in the averaging kernels is suciently small that IWP S these are representative of typical proles. Colored lines show the averaging kernels as a function of MLS retrieval T level, indicating the region of the atmosphere from which information is contributing to the measurements on N S the individual retrieval surfaces, which are denoted by plus signs in corresponding colors. The dashed black line 2 O indicates the resolution, determined from the full width at half maximum (FWHM) of the averaging kernels, approx- T imately scaled into kilometers (top axes). (Left) Vertical averaging kernels (integrated in the horizontal dimension S O 3 for ve along-track proles) and resolution. The solid black line shows the integrated area under each kernel (hori- T zontally and vertically); values near unity imply that the majority of information for that MLS data point has come OH S from the measurements, whereas lower values imply substantial contributions from a priori information. (Right) T Horizontal averaging kernels (integrated in the vertical dimension) and resolution. The horizontal averaging ker- RHI S nels are shown scaled such that a unit averaging kernel amplitude is equivalent to a factor of 10 change in pressure. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 83 Level 2 Version 4.2 Quality T

90 Help 3.11. Hydrogen Cyanide (HCN) Overview Table BrO S 80 80 0.01 0.01 T CH 70 70 S 3 Cl 0.1 0.1 60 60 ) ) T a a P P CH h h S e e 50 50 r r v04.20 v04.20 v04.20 v04.20 v04.20 v04.20 3 1 1 u u CN s s 2005d028 v03.30 2006d059 v03.30 v03.30 v03.30 v03.30 v03.30 s s e e r r T v02.21 v02.21 v02.21 v02.21 v02.21 v02.21 P P 40 40 ( ( CH 10 10 S g g o o l l 3 10 10 Approx Altitude / km Approx Altitude / km OH 30 30 T 20 20 Std dev Std dev Std dev Std dev Std dev Std dev ClO S Precision Precision Precision Precision Precision Precision 100 100 Mean VMR Mean VMR Mean VMR Mean VMR Mean VMR Mean VMR 10 10 T S CO −0.1 0.3 0.2 0.1 0.0 −0.1 0.2 0.0 0.3 0.1 T Precision and scatter (ppbv) Precision and scatter (ppbv) GPH S T Figure 3.11.2: and observed standard deviation for MLS v4.2x (black), v3.x L2gpPrecision Estimated precision ◦ H S (blue) and v2.x (red) HCN. The data shown are all proles within 20 of the equator for 28 January, 2005 and 28 2 O February 2006. Mean mixing ratio (VMR) proles are shown for comparison. Note that these are essentially the T same in v2.x, v3.x and v4.2x for the region recommended for use in all three versions(10 hPa - 0.1 hPa). HCl S T HCN S T HNO S 3 T HO S 2 T HOCl S T Summary of data quality for MLS v4.2x HCN. The precision shown is the estimated precision Table 3.11.1: IWC S ( L2gpPrecision ); the observed scatter is about 80% of this value. T Precision / Accuracy / Resolution IWP S Comments Pressure pptv % × V H / km T Unsuitable for scientic use < 0 1 . — — hPa — N S 2 O 500 50 50 1–0.1hPa 12 × T 300 30 50 × 21–1hPa 10 S O Unsuitable for scientic use 50 100–21hPa 10 × 300 Very poor 3 T Not Retrieved hPa > 100 OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 84 Level 2 Version 4.2 Quality T

91 Help 3.11. Hydrogen Cyanide (HCN) Overview 3.11.6 Data screening Pressure range: 21–0.1hPa Table Values outside this range are not recommended for scientic use. Estimated precision: Only use values for which the estimated precision is a positive number. BrO S a priori information has a strong inuence are agged with negative or zero precision, Values where the T and should not be used in scientic analyses (see Section 1.5). CH S 3 Cl Status ag: Only use proles for which the eld is an even number. Status T Odd values of Status indicate that the prole should not be used in scientic studies. See Section 1.6 CH S 3 Status eld. for more information on the interpretation of the CN T are suitable for use. Status Clouds: Clouds have no impact, proles with non-zero even values of CH S As HCN is only useable in the upper stratosphere, proles which have either, both or neither of the 3 OH cloud ags set may be used. T eld is greater than 0.2 should be used. Quality: Only proles whose Quality ClO S Values of Quality are usually near 1.5; occasional lower values do not seem correlated with unusual T 2 > Quality proles, but we suggest as a precaution that only proles with be used. Typically this . 0 S CO will eliminate only 1-2% of proles. T GPH S Convergence eld is less than 2.0 should be used. Convergence: Only proles whose T ¿is should eliminate any chunks which have obviously failed to converge – typically this is only 1-2% H S of the total. 2 O T HCl S 3.11.7 Artifacts ¿ere are no obvious artifacts within the recommended altitude range T HCN S 3.11.8 Desired improvements for future data version(s) T HNO Hopefully it will prove possible to retrieve HCN in the lower stratosphere. S 3 T HO S 2 T HOCl S T IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 85 Level 2 Version 4.2 Quality T

92 Help Overview 3.12 Nitric Acid (HNO ) 3 Table Swath name: HNO3 Useful range: 215 – 1.5 hPa (1.0 hPa under enhanced conditions) BrO S Email: [email protected] > Gloria Manney, Contact: < T CH S 3 Cl 3.12.1 Introduction T product is derived from the 240-GHz retrievals at pressures equal to or greater than 22 ¿e standard HNO 3 CH S 3 hPa and from the 190-GHz retrievals for lesser pressures. ¿e quality and reliability of the Aura MLS v2.2 CN [2007]. V3.3 HNO Santee et al. measurements were assessed in detail by HNO was greatly improved over 3 3 T CH that in version v2.2, especially in that a low bias through much of the stratosphere (especially evident at levels S 3 OH was with pressure greater than or equal to 100hPa) was largely eliminated. However, v3.3 240-GHz HNO 3 Livesey et al. adversely impacted by clouds [ , 2013], leading to a noisy HNO product in the UTLS. ¿e adverse T 3 ClO S cloud impacts have been substantially mitigated in v4.2 (see 1.4). Figure 3.12.1 shows an example of typical x x HNO x andv4.2 /v3.4 x . Wheneachversionisscreenedasrecommended[seebelow dierencesbetweenv3.3 3 T andv3.4 x isslightlylowerthanv3.3 x x ,2013],v4.2 Liveseyetal. and atpressuresgreaterthanorequalto215hPa S CO in middle to high latitudes, and shows an oscillatory pattern of dierences in the tropical UTLS, such that T values are slightly higher than those in v3.3 x and v3.4 x at 147hPa and slightly lower at 100hPa. A v4.2 x GPH S summary of the estimated precision, resolution (vertical and horizontal), and systematic uncertainty of the T measurements as a function of altitude is given in Table 3.12.1. x HNO v4.2 3 H S 2 O T 3.12.2 Resolution HCl S Rodgers , 2000]; the two- ¿e resolution of the retrieved data can be described using “averaging kernels” [e.g., T dimensional nature of the MLS data processing system means that the kernels describe both vertical and HCN S horizontal resolution. Smoothing, imposed on the retrieval system in both the vertical and horizontal di- rections to enhance retrieval stability and precision, reduces the inherent resolution of the measurements. T HNO S HNO x data, as determined from the full width at half maxi- Consequently, the vertical resolution of the v4.2 3 mum of the rows of the averaging kernel matrix shown in Figure 3.12.2, is 3–4km through most of the useful 3 T ~ range, degrading to 4.5km at 22hPa and some levels in the upper stratosphere (see Table 3.12.1). Note that HO S the averaging kernels for the 215 and 316-hPa retrieval surfaces overlap over most of their depth, indicating 2 T that the 316-hPa retrieval provides little independent information. Figure 3.12.2 also shows horizontal aver- HOCl S aging kernels, from which the along-track horizontal resolution is determined to be 400–500km over most T of the vertical range, improving to 250–300km between 15 and 4.6hPa, and degrading to 600–700km at IWC S 1.5 and 1hPa. ¿e cross-track resolution, set by the widths of the elds of view of the 190-GHz and 240-GHz ○ ~ radiometers, is 10km. ¿e along-track separation between adjacent retrieved proles is 1.5 great circle an- T ○ ○ IWP S –20 gle ( 165km), whereas the longitudinal separation of MLS measurements, set by the Aura orbit, is 10 ~ over low and middle latitudes, with much ner sampling in the polar regions. T N S 2 O 3.12.3 Precision T ¿e precision of the MLS HNO measurements is estimated empirically by computing the standard deviation 3 S O ○ 3 -wide latitude band centered around the equator, where natural atmospheric variabil- of the proles in the 20 T ity should be small relative to the measurement noise. Because meteorological variation is never completely OH S negligible, however, this procedure produces an upper limit on the precision-related variability. As shown in T ~ 0.5–0.6ppbvthroughouttherangefrom215to3.2hPa, Figure3.12.3,theobservedscatterinthev4.2 datais x RHI S and increases sharply at lower pressures. ¿e scatter is essentially invariant with time, as seen by comparing T the results for the dierent days shown in Figure 3.12.3. An alternative method using the dierence between SO S ascending and descending values at orbit crossings yields very similar estimates. 2 T S T Aura Microwave Limb Sounder (MLS) x 86 Level 2 Version 4.2 Quality T

93 Help 3.12. Nitric Acid (HNO ) 3 Overview Table BrO S T CH S 3 Cl T CH S Averages for February, 2009 3 CN HNO v4.2 HNO v3.3 3 3 T 1.0 1.0 CH S 3 OH T 10.0 10.0 ClO S T Pressure / hPa Pressure / hPa S CO 100.0 100.0 T GPH o o o o o o o o o o o o S EQ 30 S N 60 30 N 90 S N EQ 30 S N 60 60 N 90 90 S 60 90 S 30 N S Latitude Latitude T H S 1 1 2 3 4 12 5 6 10 9 8 7 6 5 4 3 2 11 0 0 7 8 9 12 11 10 2 O HNO / ppbv HNO / ppbv 3 3 T HNO HNO % Difference (v3.3-v4.2) Difference (v3.3-v4.2) HCl 3 3 S 1.0 1.0 T HCN S 10.0 10.0 T HNO S 3 Pressure / hPa Pressure / hPa T 100.0 100.0 HO S 2 T o o o o o o o o o o o o 60 30 S EQ N S 90 60 S 90 N 90 30 30 S N 60 S N N 90 60 N S EQ 30 HOCl S Latitude Latitude 50 0.2 1.0 1.4 100 -0.2 20 10 5 1 0 -1 -5 -10 -20 -50 0.6 T HNO Difference / ppbv Difference / % HNO 3 3 IWC S T for February 2009, and dierences (v3.3 v4.1) V3.3 (top left) and v4.1 (top right) zonal mean HNO Figure 3.12.1: − IWP 3 S expressed in ppbv (bottom left) and percent (bottom right). Each version is screened with the recommended cri- T teria. N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 87 Level 2 Version 4.2 Quality T

94 Help 3.12. Nitric Acid (HNO ) 3 Overview Table Equator Equator FWHM / km FWHM / km 600 -2 0 2 4 6 8 10 12 1200 1000 800 0 200 400 0.1 BrO S T CH S 1.0 3 Cl T CH S 3 CN 10.0 T Pressure / hPa CH S 3 OH 100.0 T ClO S T 1000.0 2 0.6 4 0.8 1.0 1.2 0.2 0.0 0.4 -4 -2 0 -0.2 S CO Kernel, Integrated kernel Profile number 0 0 70 70 N N T FWHM / km FWHM / km GPH 800 1000 1200 0 2 4 6 8 10 12 0 -2 200 400 600 S 0.1 T H S 2 O 1.0 T HCl S T HCN 10.0 S T Pressure / hPa HNO S 100.0 3 T HO S 2 1000.0 T HOCl 1.2 1.0 0.8 0.6 0.4 0.2 -4 -0.2 0 2 4 -2 0.0 S Profile number Kernel, Integrated kernel T IWC S Figure 3.12.2: Typical two-dimensional (vertical and horizontal along-track) averaging kernels for the MLS v4.2x T ◦ N (lower); variation in the averaging kernels is suciently small that data at the equator (upper) and at 70 HNO 3 IWP S these are representative of typical proles. Colored lines show the averaging kernels as a function of MLS retrieval T level, indicating the region of the atmosphere from which information is contributing to the measurements on N S the individual retrieval surfaces, which are denoted by plus signs in corresponding colors. The dashed black line 2 O indicates the resolution, determined from the full width at half maximum (FWHM) of the averaging kernels, approx- T imately scaled into kilometers (top axes). (Left) Vertical averaging kernels (integrated in the horizontal dimension S O 3 for ve along-track proles) and resolution. The solid black line shows the integrated area under each kernel (hori- T zontally and vertically); values near unity imply that the majority of information for that MLS data point has come OH S from the measurements, whereas lower values imply substantial contributions from a priori information. (Right) T Horizontal averaging kernels (integrated in the vertical dimension) and resolution. The horizontal averaging ker- RHI S nels are shown scaled such that a unit averaging kernel amplitude is equivalent to a factor of 10 change in pressure. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 88 Level 2 Version 4.2 Quality T

95 Help 3.12. Nitric Acid (HNO ) 3 Overview 1 Table BrO S 10 T CH S v4 v3.3 / v3.4 v4 v4 v3.3 / v3.4 v3.3 / v3.4 v3.3 / v3.4 v4 3 Cl 2009d015 15 Jan 2009 T Pressure / hPa 15 Apr 2009 2009d105 CH S 2009d196 15 Jul 2009 3 CN 100 15 Oct 2009 2009d288 T CH S 3 OH 1 0 2 4 3 2 1 4 3 T Precision / ppbv HNO HNO Precision / ppbv 3 3 ClO S Figure 3.12.3: measurements for four representative days Precision of the (left) v3.3 and (right) v2.2 MLS HNO 3 T (see legend). Solid lines depict the observed scatter in a narrow equatorial band (see text); dotted lines depict the S CO theoretical precision estimated by the retrieval algorithm. T GPH S ¿e single-prole precision estimates cited here are, to rst order, independent of latitude and season, T H but it should be borne in mind that the large geographic variations in HNO abundances gives rise to a wide S 3 2 O abundances are range of signal to noise ratios. At some latitudes and altitudes and in some seasons, HNO 3 T smaller than the single-prole precision, necessitating the use of averages for scientic studies. In most cases, HCl S √ precision can be improved by averaging, with the precision of an average of proles being 1/ N N times the T precision of an individual prole (note that this is not the case for averages of successive along-track proles, HCN S which are not completely independent because of horizontal smearing). T ¿e observational determination of the precision is compared in Figure 3.12.3 to the theoretical precision HNO S values reported by the Level 2 data processing algorithms. Although the two estimates compare very well 3 between 215 and 15hPa, at pressures smaller than 10hPa the predicted precision substantially exceeds the T HO S observedscatter. ¿isindicatesthattheaprioriinformationandtheverticalsmoothingappliedtostabilizethe 2 retrieval are inuencing the results at the higher retrieval levels. Because the theoretical precisions take into T HOCl account occasional variations in instrument performance, the best estimate of the precision of an individual S datapointisthevaluequotedforthatpointintheL2GPles, butitshouldbeborneinmindthatthisapproach T overestimates the actual measurement noise at pressures less than 10hPa. Conversely, the observed scatter IWC S and much larger at pressures higher than 100hPa is somewhat larger than the theoretical precision in v4.2 x T than the theoretical precision in v3.3. ¿is is related to the spikes and increased noise in the UTLS in v3.3, IWP S and can be seen to be much improved in v4.2 . x . Nevertheless, some outliers in this region persist in v4.2 x T Procedures for screening outliers in the UTLS are discussed below. N S 2 O T 3.12.4 Accuracy S O ¿e eects of various sources of systematic uncertainty (e.g., instrumental issues, spectroscopic uncertainty, 3 T and approximations in the retrieval formulation and implementation) on the MLS v4.2 x HNO measure- 3 OH S ments have been quantied through a comprehensive set of retrievals of synthetic radiances; see Santee et al. [2007] for details of a similar analysis conducted on MLS v2.2 HNO data. ¿e overall systematic uncertainty, T 3 RHI S or accuracy, is calculated by combining (RSS) the contributions from both the expected biases and the addi- tional scatter each source of uncertainty may introduce into the data. In aggregate, the factors considered in T SO these simulations are estimated to give rise to a total systematic uncertainty ranging from approximately 0.1 S 2 T S T Aura Microwave Limb Sounder (MLS) x 89 Level 2 Version 4.2 Quality T

96 Help 3.12. Nitric Acid (HNO ) 3 Overview HNO data (see Table 3.12.1). to 2.4ppbv, depending on the level, in the MLS v4.2 x 3 Table 3.12.5 Comparisons with other datasets proles generally show only small dierences in magnitude from v3.3 when each data version is HNO v4.2 x 3 screened according to the criteria above. ¿erefore, the comparisons with correlative data are expected to be BrO S [2013]. Livesey et al. very similar to those described for v3.3 by T CH S 3 3.12.6 Data screening – all data Cl T Pressure range: 215–1.5hPa CH S Values outside this range are not recommended for scientic use. 3 CN Estimated precision: Only use values for which the estimated precision is a positive number. T CH S information has a strong inuence are agged with negative or zero precision, Values where the a priori 3 OH and should not be used in scientic analyses (see Section 1.5). T Status Status ag: Only use proles for which the eld is an even number. ClO S indicate that the prole should not be used in scientic studies. See Section 1.6 Odd values of Status T eld. Status for more information on the interpretation of the S CO T 3.12.7 Data screening – upper troposphere, lower stratosphere (pressures of 22 hPa or GPH S greater) T Quality and Convergence elds included in the HNO3 swath in the standard ¿e les are appro- L2GP-HNO3 H S 2 O priate for use in screening at levels at and below (that is, pressures greater than) 22hPa. For those levels: T HCl S Quality: Only proles whose eld is greater than 0.8 should be used. Quality T 1–3% of HNO ~ typically excludes Quality ¿is threshold for proles on a daily basis; it is a conser- 3 HCN S vative value that potentially discards a signicant fraction of “good” data points while not necessarily T proles in v4.2 x show a “notch”, with identifying all “bad” ones. A signicant fraction of the HNO 3 HNO S unexpectedly low (o en negative in the tropics) values at 100hPa and high values at 147hPa; many 3 threshold. Quality such proles are removed by using this T HO S Convergence Convergence: Only proles whose eld is less than 1.03 should be used. 2 T Convergence On a typical day this threshold for discards a very small fraction of the data, rarely elim- HOCl S proles. Some such proles show unphysical behavior. inating more than 0.1% of the HNO 3 T Clouds: Clouds impact HNO data in the UTLS, see discussion below and the discussion on “outliers” 3 IWC S that follows. T Nonzero but even values of Status indicate that the prole has been marked as questionable, typically IWP S because the measurements may have been aected by the presence of thick clouds. Globally ~ 1–2% of T proles are identied in this manner, with the fraction of proles possibly impacted by clouds rising to N S ~ data. 5% on average in the tropics. Clouds generally have little inuence on the stratospheric HNO 2 3 O In the lowermost stratosphere and upper troposphere, however, thick clouds can lead to spikes in the T S O mixing ratios in the equatorial regions. HNO 3 3 T ¿erefore, it is recommended that at and below 68hPa, in addition to performing the “outlier screen- OH S ing” described below, all proles with nonzero values of Status be discarded because of the potential for cloud contamination. While this will reject some proles that are probably not signicantly impacted T RHI S by cloud eects, and the number of proles rejected that are not agged by other criteria is small ( 0.5 ~ to 1% of total proles), it does eliminate some unphysical proles that are not agged by the outlier T SO screening. S 2 T S T Aura Microwave Limb Sounder (MLS) x 90 Level 2 Version 4.2 Quality T

97 Help 3.12. Nitric Acid (HNO ) 3 Overview Outliers: Alternative screening approaches in the UTLS remove outliers while reducing “false positives” x HNO While the number, and particularly the extreme values of, outliers in v4.2 at levels between 3 Table 316 and 100 (sometimes to 68) hPa is greatly reduced over that in v3.3, such outliers are still present. ¿ese typically appear as highly negative mixing ratios at the lowest several retrieval levels, o en as BrO S part of oscillatory proles with unrealistically high values at higher altitudes. A simple procedure is recommended to screen such proles based on eliminating all proles with large negative mixing ratios T CH at pressure levels between 316 and 68hPa. ¿rough extensive examination of data screened in this way, S 3 Cl ppbv at 316hPa or less than 6 − vmr less than ppbv at 0 . 2 . − 1 agging proles that have either HNO 3 T any level between 215 and 68hPa eliminates most of the troublesome outliers, including those with CH S positive vmr spikes overlying the negative ones that are directly agged by these criteria. ¿is screening 3 CN procedure is recommended for any studies focusing on the UTLS. A detailed description of the testing T Livesey et al. procedure for this screening is given by [2013], and this testing has demonstrated that it CH S 3 eectively removes most of the suspect proles that are not eliminated by requiring Status to be zero. OH Figure 3.12.4 shows an example of the results of screening proles by each of the Quality , Conver- T ClO S gence and the recommended outlier agging on a typical ‘bad’ day (i.e., one with a relatively large number of outliers) in v4.2 versus v3.3. ¿e Quality screening removes many of the proles that are x T strongly negative at the bottom, as well as many (but not all) with the “notch” structure between 147 and S CO 100hPa. Most or all of the remainder of the proles that are strongly negative at the bottom are agged T by the outlier screening; many of these proles are oscillatory, so this screening also removes most or GPH S all of the strong positive outliers (typically at 147hPa). Most of the proles agged by any of the criteria T are in the tropics, as expected; low HNO values in Antarctic winter resulting from denitrication are 3 H S 2 O rarely, if ever, aected by this outlier screening. It is clear from this example that the outliers in v4.2 x T are less frequent and much less extreme than those in v3.3 HCl S 3.12.8 Data screening – upper stratosphere (pressures of 15 hPa or less) T HCN for 15hPa and higher altitudes, as they result in ltering should not be used ¿e above screening criteria S and Convergence proles for which all quality indicators are good when the Quality values are properly T taken from the 190-GHz HNO information, and not ltering ones with indications of poor quality. ¿e HNO 3 S x and swath has been included in the standard HNO HNO3-190 , and the appropriate Quality les for v4.2 3 3 T from that swath must be used to apply the following screening values for the 190-GHz HNO Convergence 3 HO S criteria: 2 T has a non-zero even number can be used without Clouds: Proles where the Status eld for HNO3-190 HOCl S restriction. T data at these altitudes. Clouds generally have little inuence on the stratospheric HNO 3 IWC S HNO3-190 than 0.8 greater (see section 1.6) Quality eld for Quality: Only proles with a value of the T should be used in scientic study. IWP S ¿is threshold for ~ 1–3% of HNO proles on a daily basis; it is a con- typically excludes Quality 3 T N servative value that potentially discards some “good” data points while not necessarily identifying all S 2 O “bad” ones. T S O Convergence: Only proles with a value of the Convergence eld (see section 1.6) for the HNO3-190 prod- 3 than 1.4 should be used in investigations. uct less T OH S On a typical day this threshold for 0.5–1.5% of the HNO ~ proles. Many Convergence discards 3 of the proles thus agged show some unphysical behavior, especially at the highest altitudes in the T RHI S recommended range. T For levels at and above (pressures less than) 4.6hPa, especially at 2.2hPa and above, some pro- Outliers: SO S and is very low. ¿e les show vertically oscillatory behavior in conditions where HNO Quality 3 2 T S T Aura Microwave Limb Sounder (MLS) x 91 Level 2 Version 4.2 Quality T

98 Help 3.12. Nitric Acid (HNO ) 3 Overview v04-11-c01_2009d051 v04-11-c01_2009d051 All Profiles All Profiles Mean & Range Mean & Range 1 1 1 1 Q_Crit = 0.80 Q_Crit = 0.80 C_Crit = 1.03 C_Crit = 1.03 Status: 0 Status: 0 Table 10 10 10 10 BrO S Pressure / hPa Pressure / hPa Pressure / hPa Pressure / hPa 100 100 100 100 T 3381 good 718 good CH 53 failed Q 33 failed Q S 0 failed C 0 failed C 15 20 15 -5 0 5 10 5 20 -10 -5 0 5 10 15 20 -10 20 -10 10 5 0 -5 10 15 0 -10 -5 3 0 failed Q&C 0 failed Q&C HNO HNO HNO HNO Cl 3 3 3 3 13 outliers flagged 26 outliers flagged -90.0 to 90.0 -20.0 to 20.0 -90.0 to 90.0 -20.0 to 20.0 T v03-30-c01_2009d051 v03-30-c01_2009d051 CH Mean & Range Mean & Range All Profiles All Profiles S 1 1 1 1 Q_Crit = 0.50 Q_Crit = 0.50 3 C_Crit = 1.40 C_Crit = 1.40 CN Status: 0 Status: 0 T CH S 10 10 10 10 3 OH Pressure / hPa Pressure / hPa Pressure / hPa Pressure / hPa T 100 100 100 100 ClO S 546 good 3091 good 19 failed Q 28 failed Q 4 failed C 4 failed C 5 10 15 20 15 20 -10 10 15 -10 -5 0 5 10 15 20 20 -5 0 -10 -5 0 5 10 -10 -5 0 5 T 0 failed Q&C 1 failed Q&C HNO HNO HNO HNO 3 3 3 3 10 outliers flagged 14 outliers flagged -90.0 to 90.0 -90.0 to 90.0 -20.0 to 20.0 -20.0 to 20.0 S CO T Figure3.12.4: HNO proles on 20 Feb 2009 color-coded by screening. Cyan proles have Quality less than 0.8 (0.5), GPH 3 S Convergence greater less than 0.8 (0.5) and olive-green Convergence greater than 1.03 (1.4), and red both Quality T than 1.03 (1.4) for v4.2x (v3.3). Orange proles are those agged by the simple screening procedure described above H S 2 and/or Convergence (using large negative mixing ratios at high pressures) after the proles that failed tests Quality O were removed. Black proles are all those remaining (the “good” proles) after all screening. The left panels show T HCl all individual proles in the day; the right panels show the means in each category, with the standard deviation S shown as bars and the range as dotted lines. The horizontal line is at 22 hPa, above which HNO is from the 190-GHz 3 T radiometer and thus not appropriately screened by these criteria. The four pairs of panels show all v4.2x proles HCN S ◦ ◦ latitude (top right), all v3.3 proles (bottom left) and v3.3 proles and 20 20 − (top left), v4.2x proles between ◦ ◦ T − and 20 20 latitude. between HNO S 3 Convergence criteria dened above eliminate many of these proles. T HO S 2 T 3.12.9 Artifacts HOCl S ¿ereisapersistent“notch”at100hPa, notedinFigures3.12.1and3.12.4, thatismostpronouncedinthe • T x at147hPaisconsistentlyslightlyhigherthanthat HNO tropicsbutalsoapparentinmidlatitudes. v4.2 3 IWC S at100hPa. AscanbeseeninFigure3.12.4, thisfeatureisverystronginmanyprolesthatareeliminated T by one or more of the screening criteria. However, Figure 3.12.4 (right panels) and Figure 3.12.1 (lower IWP S right panel) show that this feature is more pronounced than it was in v3.3 and v3.4 x , and it is reduced, x T but by no means eliminated, by following the recommended screening procedures. ¿is feature was N S x testing, and its origin is still unclear. persistent throughout v4.2 2 O T S 3.12.10 Desired improvements for future data version(s) O 3 Diagnose and eliminate the unphysical notch (minimum) at 100hPa. • T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 92 Level 2 Version 4.2 Quality T

99 Help 3.12. Nitric Acid (HNO ) 3 Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T ClO S Summary of Aura MLS v4.2x HNO Table 3.12.1: Characteristics 3 T Systematic Resolution S CO b Pressure Precision c a Uncertainty Comments V × H T / hPa / ppbv / ppbv / km GPH S 0.68–0.001 — — — Unsuitable for scientic use T 1.2 ± 0.3 Caution, averaging recommended 1.5–1.0 4.5 × 550–750 ± H S 2 0.5 2.1 5 × 450–500 ± 0.8 ± O 3.2 4.5 × 400 ± 0.6 ± 0.1 T HCl S 250–350 ± 0.6 ± 0.5 × 15–4.6 3–4 2.4 ± 0.6 ± 500 × 4.5 22 T ± 1.5 32 4 ± 400 0.6 × HCN S 147–46 3–4 × 350–400 ± 0.6 ± 1.0 T ± 0.6 ± 1.1 × 4–4.5 215 450 HNO S 316 — — — Unsuitable for scientic use — Not retrieved 1000–464 — — 3 T HO S a Horizontal resolution in along-track direction. 2 b Precision on individual proles, determined from observed scatter in the data in a region of minimal atmospheric variability. T c HOCl estimates of the probable magnitude. σ Values should be interpreted as 2- S T IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 93 Level 2 Version 4.2 Quality T

100 Help Overview 3.13 Peroxy Radical (HO ) 2 Table HO2 Swath name: Useful range: 22 – 0.046 hPa BrO S and Shuhui Wang, [email protected] Contact: > Luis Millan, < Email: T Email: [email protected] > < CH S 3 Cl T 3.13.1 Introduction CH S 3 data quality, precision, systematic errors, and validation for an earlier version, v2.2, is A description of HO CN 2 Pickett et al. given in [2008]. An early validation using v1.5 so ware is also described in [2006b]. Pickett et al. T CH S HO are summarized below in Table 3.13.1. ¿e estimated uncertainties, precisions, and resolution for v4.2 x 2 3 OH An algorithm to retrieve daily zonal means of HO over an extended vertical range by rst averaging 2 T , 2015]. ¿is alternative dataset is the rst Millán et al. the radiances has been developed by the MLS team [ ClO S satellite record covering the stratosphere and the mesosphere. In the long-term daytime and nighttime HO 2 T near future, this dataset will be available at the GFSC DISC, in the meantime, those interested in using it are S CO advised to contact the MLS team. T GPH 3.13.2 Resolution S ○ Figure 3.13.1 shows the HO N and the Equator. ¿e latitudinal variation averaging kernel for daytime at 70 2 T H in the averaging kernel is very small. ¿e vertical resolution for pressures greater than 0.1hPa is generally S 2 O about 5km. T HCl S 3.13.3 Precision T ) are shown in Fig- x prole and the associated precisions (for both v3.3 x and v3.4 x and v4.2 A typical HO HCN 2 S ure 3.13.2. ¿e prole is shown in both volume mixing ratio (vmr) and density units. All MLS data are T 3 − 6 ) cm reported in vmr for consistency with the other retrieved molecules. However, use of density units (10 HNO S reduces the apparent steep gradient of HO vertical prole, allowing one to see the prole with more detail. 2 3 T ¿e night HO prole is expected to exhibit a narrow layer near the altitudes of the nighttime OH layer at 2 HO S , 2006a], which is not shown in Figure 3.13.2 since MLS HO ~ 82km [ Pickett et al. data is not recommended 2 2 ○ for altitudes above 0.046hPa ( ~ 70km). Precisions are such that an HO zonal average within a 10 latitude T 2 HOCl S 2000 samples) for most bin can be determined with better than 10% relative precision with 20 days of data ( ~ pressure levels over 22–0.046hPa. T IWC S 3.13.4 Accuracy T . ¿e eect of each identied source of systematic er- Table 3.13.1 summarizes the accuracy expected for HO 2 IWP S Read et al. ror on MLS measurements of radiance has been quantied and modeled [ , 2007]. ¿ese quantied T σ estimates of uncertainties in each MLS product, or an estimate of the maxi- 2 eects correspond to either N S 2 O mum reasonable uncertainty based on instrument knowledge and/or design requirements. ¿ese accuracy T x and v3.4 calculations were performed with more realistic HO x estimates. atmospheric proles than for v3.3 2 S O ¿e HO bias can be eliminated by taking day-night dierences over the entire recommended pressure range. 3 2 T ¿e overall uncertainty is the square root of the sum of squares of the precision and accuracy. OH S 3.13.5 Data screening T RHI S It is recommended that HO data values be used in scientic investigations if all the following tests are suc- 2 cessful: T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 94 Level 2 Version 4.2 Quality T

101 Help 3.13. Peroxy Radical (HO ) 2 Overview Table BrO S T CH S 3 Cl T CH S 3 CN 0 70 N Equator T FWHM / km FWHM / km 12 6 -2 4 2 12 10 8 0 -2 0 2 4 6 8 10 CH S 0.1 3 OH T ClO S T 1.0 S CO T GPH S Pressure / hPa T 10.0 H S 2 O T HCl S T 100.0 HCN S 0.4 0.0 1.2 0.8 0.6 1.2 1.0 0.8 0.6 1.0 0.2 0.0 -0.2 0.4 0.2 -0.2 Kernel, Integrated kernel Kernel, Integrated kernel T HNO S 3 ◦ T data at the equator (left) and at 70 N (right); Typical vertical averaging kernels for the MLS v4.2x HO Figure 3.13.1: 2 HO S variation in the averaging kernels is suciently small that these are representative of typical proles. Colored lines 2 show the averaging kernels as a function of MLS retrieval level, indicating the region of the atmosphere from which T HOCl information is contributing to the measurements on the individual retrieval surfaces, which are denoted by plus S signs in corresponding colors. The dashed black line indicates the vertical resolution, determined from the full T width at half maximum (FWHM) of the averaging kernels, approximately scaled into kilometers (top axes). The solid IWC S black line shows the integrated area under each kernel; values near unity imply that the majority of information for that MLS data point has come from the measurements, whereas lower values imply substantial contributions T from a priori information. The low signal to noise for this product necessitates the use of signicant averaging (e.g., IWP S monthly zonal mean), making horizontal averaging kernels largely irrelevant. T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 95 Level 2 Version 4.2 Quality T

102 Help 3.13. Peroxy Radical (HO ) 2 Overview (a) HO2 v4 (b) HO2 v4 Table 1 0.1 BrO S T CH S 1.0 3 Cl T pressure / hPa pressure / hPa 10 CH S Day Night 3 CN 10.0 T CH S 3 0.0 0.2 -1 0.3 0.4 4 3 2 1 0 0.1 OH vmr / ppb vmr / ppb T (c) HO2 v4 (d) HO2 v4 ClO S T 0.1 0.1 S CO T GPH S 1.0 1.0 T H S pressure / hPa pressure / hPa 2 O T 10.0 10.0 HCl S T HCN S 60 80 -20 100 15 10 5 0 20 25 40 20 0 6 6 -3 -3 cm density / 10 cm density / 10 T (e) HO2 v3 (f) HO2 v3 HNO S 3 T 0.1 0.1 HO S 2 T HOCl S 1.0 1.0 T pressure / hPa pressure / hPa IWC S T 10.0 10.0 IWP S T 60 80 100 25 -20 0 5 0 20 10 15 40 20 N 6 -3 6 -3 S density / 10 cm cm density / 10 2 O T Monthly zonal mean of retrieved HO Figure 3.13.2: and its estimated precision (horizontal error bars) for Septem- S 2 O ◦ ◦ 3 ber, 2005 averaged over 29 N to 39 vmr vs. pressure for day (black) and night (blue). N. Panel (a) shows v4.2x HO 2 T Panel (b) shows the same data plotted for the stratosphere as a day-night dierence (note that a day-night dier- OH S ence is required for HO for all pressure levels). Panel (c) shows the same data in (a) converted into density units. 2 T Panel (d) shows the day-night dierences for the data in panel (c). Panels (e) and (f ) are equivalent to (c) and (d) but RHI S using v3.3x and v3.4x data. The average in panels (a) – (d) using v4.2x data includes 3052 proles, while the average in panels (e) – (f ) using v3.3x and v3.4x data includes 2695 proles. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 96 Level 2 Version 4.2 Quality T

103 Help 3.13. Peroxy Radical (HO ) 2 Overview Summary of precisions, resolution, and uncertainties for the MLS v4.2x HO product Table 3.13.1: 2 Vertical Day–night a Table Pressure Precision Comments resolution accuracy / 6 − 3 /hPa / 10 cm 3 6 − 10 cm km BrO Unsuitable for scientic use — — — 0.03hPa < S Use day–night dierence 10 6 0.046hPa 0.01 T 0.10hPa 7 10 Use day–night dierence 0.2 CH S 1.0 hPa 0.2 5 Use day–night dierence 11 3 Cl 50 14 Use day–night dierence 10 hPa 4 T 22hPa — — — Unsuitable for scientic use > CH S 3 CN a Precision for a single prole T CH S 3 OH Pressure range: 22–0.046hPa T Values outside this range are not recommended for scientic use. ClO S Estimated precision: Only use values for which the estimated precision is a positive number. T S CO a priori Values where the information has a strong inuence are agged with negative or zero precision, and should not be used in scientic analyses (see Section 1.5). T GPH S eld is an even number. Status Status ag: Only use proles for which the T Status indicate that the prole should not be used in scientic studies. See Section 1.6 Odd values of H S 2 eld. for more information on the interpretation of the Status O T Quality: MLS v4.2 x HO data can be used irrespective of the value of the Quality eld. 2 HCl S Convergence: Only proles whose Convergence eld is less than 1.1 should be used. T HCN , In version v2.2 this test o en fails for 100 out of 3500 proles in a day. In the current version, v4.2 x S there are o en zero or very few non-converged proles. T HNO S 3.13.6 Artifacts 3 T product. ¿e primary limitation is the precision and the Currently there are no known artifacts in the HO 2 HO S altitude range. 2 T HOCl S 3.13.7 Review of comparisons with other datasets data from MLS v2.2 so ware have been validated with two balloon-borne remote-sensing instruments. HO T 2 IWC S x Details of the comparison are given in [2008]. Dierences between v2.2 and v4.2 Pickett et al. show no dierences large enough to alter results of previous validation studies. T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 97 Level 2 Version 4.2 Quality T

104 Help Overview 3.14 Hypochlorous Acid (HOCl) Table Swath name: HOCl 10 – 2.2 hPa Useful range: BrO S [email protected] Lucien Froidevaux, < Email: Contact: > T CH S 3 Cl 3.14.1 Introduction T CH ¿ere has been little change in the v4.2 HOCl retrieval results, in comparison to v3.3/v3.4. We provide be- S 3 low sample mean HOCl distributions for the two data versions, and their dierences. Otherwise, previous CN information regarding this species remains largely unchanged; the main points are mentioned here mainly T CH S for new data users. 3 OH ¿e HOCl retrieval is quite noisy for individual proles and HOCl data require some averaging (e.g., in ○ T 10 zonal means for one or more weeks) to get useful precision of better than 10pptv, in comparison to typical ClO S upper stratospheric HOCl abundances of 100–150pptv. Table 3.14.1 summarizes the MLS HOCl resolution, T precision, andaccuracyestimatesfortheupperstratosphere. Morediscussionandabriefvalidationsummary S CO aregiveninthefollowingsections,alongwithdatascreeningrecommendations,whichshouldbeofparticular interest to MLS data users. T GPH S 3.14.2 Changes from v3.3/v3.4 T H ¿ere were no algorithmic changes relating to HOCl for the v4.2 retrievals. Small dierences in HOCl abun- S 2 O dances (see below) are likely related mainly to small changes in tangent pressure and/or temperature. T O. ¿ere , and H ¿e background observed in the 640-GHz radiances includes emissions from N , O HCl 2 2 2 S are laboratory-based and ground-based models for the continuum absorptions that are the basis for the MLS T absorption model [ Pardo et al. , 2001, and references therein]. ¿ese models were tested against MLS ex- HCN S tinction measurements from the wing channels in the 640-GHz radiometer. ¿e latitude dependence of this T extinction was found to agree better with the expected most plus dry continuum extinction values if the dry HNO S and moist continuum functions were scaled by factors close to 20 ; however, this is not a change from the % 3 v3.3/v3.4 retrievals. T HO S A comparison plot showing zonal average upper stratospheric HOCl contours (from 10 to 2 hPa) and 2 dierences between the two data versions for a typical month (March, 2009) is provided in Figure 3.14.1. ¿e T HOCl S v4.2 HOCl abundances are mostly slightly larger than the v3.3 retrievals, typically by only a few pptv (a few percent). ¿e estimated precision values are essentially unchanged from v3.3. T IWC S 3.14.3 Resolution T Based on the width of the averaging kernels shown in Figure 3.14.2, the vertical resolution for upper strato- IWP S 6km (signicantly worse than the 640-GHz radiometer vertical eld of view width of spheric HOCl is ∼ T 1.4km). ¿is reects the choice of smoothing constraints for HOCl which favor precision over vertical reso- N S 2 lution. O T S O 3.14.4 Precision 3 T ¿e estimated single-prole precision reported by the Level 2 so ware is about 300 to 400pptv in the upper OH S ○ stratosphere. A more useful number of 10pptv is quoted in Table 3.14.1 for the typical precision of a 10 weekly T zonal means for this product. RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) 98 Level 2 Version 4.2 x Quality T

105 Help 3.14. Hypochlorous Acid (HOCl) Overview Table BrO S T CH S 3 Cl T Averages for March, 2009 CH S 3 CN HOCl v3.3 HOCl v4.2 T CH S 3 OH T ClO S T Pressure / hPa Pressure / hPa S CO T 10.0 10.0 GPH o o o o o o o o o o o o S 90 90 N S 60 60 S 30 S S EQ 30 N N 60 90 N S 30 N N 60 S EQ 30 90 Latitude Latitude T H S 50 125 150 75 100 125 150 75 -25 0 -25 0 25 50 25 100 2 O HOCl / pptv HOCl / pptv T HCl S HOCl Difference (v4.2-v3.3) HOCl % Difference (v4.2-v3.3) T HCN S T HNO S 3 T Pressure / hPa Pressure / hPa HO S 2 T 10.0 10.0 HOCl o o o o o o o o o o o o S N N 90 S 60 S 30 S EQ 30 60 90 N 90 S N 90 N 60 N 30 EQ S 30 S 60 Latitude Latitude T IWC S -2 0 2 -8 -6 -2 -4 -5 0 2 4 6 8 5 HOCl Difference / % HOCl Difference / pptv T IWP S Zonal averages for upper stratospheric MLS HOCl proles during March, 2009, showing the MLS v3.3 Figure 3.14.1: HOCl mixing ratio contours (top left panel), the v4.2 contours (top right panel), and their dierences in pptv (v4.2 T N minus v3.3, bottom left panel) and percent (v4.2 minus v3.3 versus v3.3, bottom right panel). S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 99 Level 2 Version 4.2 Quality T

106 Help 3.14. Hypochlorous Acid (HOCl) Overview Table BrO S T CH S 3 Cl T CH S 3 CN 0 N 70 Equator T FWHM / km FWHM / km -2 6 8 12 12 10 4 10 -2 2 0 2 4 6 8 0 CH S 1 3 OH T ClO S T S CO T 10 GPH S Pressure / hPa T H S 2 O T HCl S T 100 HCN S -0.2 1.2 1.0 0.0 0.8 0.6 0.4 0.2 0.0 1.0 0.8 0.6 1.2 0.4 0.2 -0.2 Kernel, Integrated kernel Kernel, Integrated kernel T HNO S 3 ◦ T Typical vertical averaging kernels for the MLS v4.2x HOCl data at the equator (left) and at 70 N (right); Figure 3.14.2: HO S variation in the averaging kernels is suciently small that these are representative of typical proles. Colored lines 2 show the averaging kernels as a function of MLS retrieval level, indicating the region of the atmosphere from which T HOCl information is contributing to the measurements on the individual retrieval surfaces, which are denoted by plus S signs in corresponding colors. The dashed black line indicates the vertical resolution, determined from the full T width at half maximum (FWHM) of the averaging kernels, approximately scaled into kilometers (top axes). The solid IWC S black line shows the integrated area under each kernel; values near unity imply that the majority of information for that MLS data point has come from the measurements, whereas lower values imply substantial contributions T from a priori information. The low signal to noise for this product necessitates the use of signicant averaging (e.g., IWP S monthly zonal mean), making horizontal averaging kernels largely irrelevant. T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 100 Level 2 Version 4.2 Quality T

107 Help 3.14. Hypochlorous Acid (HOCl) Overview 3.14.5 Accuracy ¿e accuracy estimates shown in Table 3.14.1 come from a formal quantication of the combined eects of Table possible systematic errors in MLS calibration, spectroscopy, etc. on the HOCl retrievals [ , 2007]. Read et al. ¿ese(updated)valuesareintendedtorepresent 2 σ estimatesofaccuracy. ¿elargestcontributorstopossible BrO S errors for HOCl are contaminant species, gain compression, and sideband ratio uncertainties. ¿e Table gives a range of error estimates (for the range of pressures). ¿e average changes for upper stratospheric HOCl T CH between v3.3 and v4.2 are well within the quoted accuracy estimates (which may be somewhat conservative). S 3 Cl ¿e HOCl signal becomes too small compared to the systematic uncertainties to allow for reliable retrievals T at pressures larger than 10 hPa. CH S 3 CN 3.14.6 Data screening T Pressure range: 10–2.2hPa CH S 3 Values outside this range are not recommended for scientic use. Artifacts (negative averages) for pres- OH sureslargerthanabout10hPamakethisproductunsuitableforuseinthelowerstratosphere. Regarding T ClO S the topmost altitude range, the sensitivity to a priori increases rapidly at pressures of 1hPa or less; we continue to recommend the use of (average) HOCl values only up to 2.2 hPa. T S CO Estimated precision: Only use values for which the estimated precision is a positive number. T information has a strong inuence are agged with negative or zero precision, Values where the a priori GPH S and should not be used in scientic analyses (see Section 1.5). T eld is an even number. Status Status ag: Only use proles for which the H S 2 O Odd values of indicate that the prole should not be used in scientic studies. See Section 1.6 Status T Status eld. for more information on the interpretation of the HCl S greater Quality eld: Only proles with a value of the Quality eld than 1.2 should be used. T HCN S ¿is criterion removes proles with the poorest radiance ts, typically less than 0.1 % of the daily pro- les. For HOCl (and for other 640-GHz MLS products), this screening correlates well with the poorly T HNO converged sets of proles (see below); we recommend the use of both the and Convergence Quality S elds for data screening. 3 T eld Convergence Convergence eld: Only proles with a value of the less than 1.05 should be used. HO S 2 For the vast majority of proles (99 % or more for most days), this eld is less than 1.05. Nevertheless, T HOCl on occasion, sets of proles (typically one or more groups of ten proles, retrieved as a ‘chunk’) have S this Convergence eld set to larger values, and should be discarded. T Clouds: Proles identied as being aected by clouds can be used with no restriction. IWC S T 3.14.7 Review of comparisons with other datasets IWP S ¿e MLS HOCl retrievals exhibit the expected morphology in monthly mean latitude / pressure contour T plots; for example, such plots for September months from MLS compare favorably, to rst-order, with re- N S 2 O sults produced by the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) for September, T von Clarmann et al. 2002 [ , 2006]. MLS HOCl averages at midlatitudes are close to the results from balloon- S O 3 borne infrared measurements. Generally favorable comparisons (within the error bars) have been made of T the diurnal changes in upper stratospheric HOCl between Aura MLS (v3.3), other satellite datasets, and a 1-D OH S Khosravi et al. , 2013]. photochemical model [ T RHI S 3.14.8 Artifacts T ¿e 640-GHz radiometer bands 10 (for ClO) and 29 (for HOCl) were turned o for a few time periods • SO S in 2006 to investigate degradation issues that might aect these channels in the future. ¿ese bands 2 T S T Aura Microwave Limb Sounder (MLS) x 101 Level 2 Version 4.2 Quality T

108 Help 3.14. Hypochlorous Acid (HOCl) Overview Summary for MLS hypochlorous acid Table 3.14.1: Vertical Table b a Comments Accuracy Pressure Resolution Precision km pptv % pptv % hPa km BrO S — Unsuitable for scientic use 1.5 or less — — — — T 150 Some averaging required 2.2 to 10 10 10 6 ~ 40–80 CH S — Unsuitable for scientic use 15 or more — — — — 3 Cl T a Precision ( ) for 1 week/10 degrees zonal means or 2 weeks/5 degrees zonal means σ 1 CH S b estimate from systematic uncertainty characterization tests σ 2 3 CN T CH S were o on April 8,9, and 10, 2006, and also for April 17, 2006 (a er 19:52 UT) through May 17, 2006. 3 OH ¿ereareessentiallynousefulHOCl(orClO)dataforthesetimeperiods. ¿ev4.2(asforthev3.3/v3.4) T Status so ware correctly ags these incidents with poor (odd) values (which should be screened out). ClO S • ¿ere are signicant artifacts in the mean values (large negative values) for HOCl in the lower strato- T sphere, where the use of this product is not recommended. S CO • Users should screen out the non-converged and poorest quality HOCl proles, as such proles (typi- T GPH cally a very small number per day) tend to behave unlike the majority of the other MLS retrievals. See S the criteria listed above. T H S 2 O T HCl S T HCN S T HNO S 3 T HO S 2 T HOCl S T IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 102 Level 2 Version 4.2 Quality T

109 Help Overview 3.15 Cloud Ice Water Content (IWC) Table Swath name: IWC 3 g/m Units: BrO S 215 – 83 hPa Useful range: T CH S < [email protected] Contact: > Alyn Lambert, Email: 3 Cl T CH S 3.15.1 Introduction 3 CN ¿e MLS IWC is retrieved from cloud-induced radiances ( ) of the 240-GHz window channel in a sepa- T cir T rate processing step a er the atmospheric state (Temperature and tangent pressure) and important gaseous CH S , HNO T O, O species (H ) have been nalized in the retrieval processing. ¿e derived are binned onto 3 cir 3 3 2 OH ○ the standard horizontal (1.5 along track) and vertical (12 surfaces per decade change in pressure) grids, and T , 2006]. ¿e standard IWC prole has converted to IWC using the modeled T –IWC relations [ Wu et al. cir ClO S a useful vertical range between 215–83hPa although the validation has been conducted for a subset of the T range of IWC values. IWC measurements beyond the value ranges specied in Table 3.15.1 are to be regarded S CO currently as giving only qualitative information on cloud ice. ¿ey require further validation for quantitative T interpretation. GPH S T 3.15.2 Resolution H S 2 In the IWC ranges specied in Table 3.15.1, each MLS measurement can be quantitatively interpreted as the O T 300km and 7km average IWC for the volume sampled. ¿is volume has a vertical extent of ~ ~ 3km, with HCl S along and cross track respectively. T HCN S 3.15.3 Precision ¿e precision values quoted in the IWC les do not represent the true precision of the data. ¿e precision T HNO S for a particular measurement must be evaluated on a daily basis using the method described in the screening section below. ¿e precision listed in Table 3.15.1 reects typical values obtained from the method described 3 T below. HO S 2 T 3.15.4 Accuracy HOCl S ¿e IWC accuracy values listed in Table 3.15.1 are estimates from comparisons of the earlier v2.2 MLS data T [2008]. Wu et al. product with CloudSat and detailed analyses on the v2.2 error budget can be found in IWC S 3.15.5 Data screening T IWP S Values outside this range are not recommended for scientic use. ¿e maxi- Pressure range (215–83hPa): 3 mum detectable IWC is ~ . 100mg/m T N S 2 ¿euserisrecommendedtoscreentheIWCdatausing Use Temperature Status, Quality and Convergence: O T , 2008]. Schwartz et al. Status eld in the collocated temperature prole to exclude bad retrievals [ the S O is an even number should be used. Status In other words, only IWC proles for which temperature 3 Similarly, the users should only use IWC proles where the corresponding Temperature proles have T OH S Convergence less than 1.03. Quality of 0.9 or larger and T ¿e IWC product derives from dierences between measured radiances and those pre- Other screening: RHI S dicted assuming cloud free conditions. Spectroscopic and calibration uncertainties give rise to tem- T porally and geographically varying biases in this dierence, and hence the IWC product. ¿ese biases SO S σ ” screening method, as described below. must be iteratively identied and removed, using a “ 2 σ – 3 2 T S T Aura Microwave Limb Sounder (MLS) 103 Level 2 Version 4.2 x Quality T

110 Help 3.15. Cloud Ice Water Content (IWC) Overview . Uncertainties in spectroscopy and atmospheric composition are manifested as residual biases in 1 the IWC elds which should be identied and removed as follows. IWC data should be averaged Table ○ latitude bins and outliers rejected iteratively by excluding measurements greater than 2 σ in a 10 standard deviation about the mean ( μ calculations a er every μ ) of the bin. Repeat the σ and is σ new set of rejections. Convergence is usually reached within 5–10 iterations, and the nal BrO S the estimated precision for the IWC measurements. T CH S μ and μ from IWC for . Interpolate the nal 2 to the latitude of each measurement, and subtract σ 3 Cl each measurement. T CH threshold to determine if an IWC measurement is statistically signicant. In σ . Finally, apply the 3 3 S 3 CN other words, it must have IWC μ + 3 σ in order to be considered as a signicant cloud hit. ¿e > T threshold is needed for cloud detection since a small percentage of clear-sky residual noise can σ 3 CH S result in a large percentage of “false alarms” in cloud detection. 3 OH T 3.15.6 Artifacts ClO S At wintertime mid-to-high latitudes, strong stratospheric gravity waves may induce large uctuations in the T σ σ screeningmethod. ¿efalsecloud retrievedtangentpressure, andcausefalseclouddetectionwiththe2 –3 S CO detection seems to aect the 100hPa pressure level most, as expected for such impact coming from the lower T stratosphere. GPH S 3.15.7 Comparisons with other datasets T H S Comparedtov2.2IWCthev4.2 x IWCvaluesarewithin5%overthepressurerange215–100hPaandgenerally 2 O x the random noise in v4.2 IWC is larger than in v2.2 (see Figure 3.15.1 and Table 3.15.1). Apart from the T HCl dierencesnotedabove, theMLSv4.2 x IWCissimilartotheMLSv2.2productdescribedandvalidatedin Wu S et al. [2008]. A revised validation paper for IWC is not planned in the near future and users are encouraged T HCN to read Wu et al. [2008] for more information. S Comparisons between v2.2 MLS and CloudSat IWC showed good agreement with PDF dierences < 50% T for the IWC ranges specied in Table 3.15.1. Comparisons with AIRS, OMI and MODIS suggest that MLS HNO S 1km than the correlative data in general. ~ cloud tops are slightly higher by 3 T HO S 2 T HOCl S T IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 104 Level 2 Version 4.2 Quality T

111 Help 3.15. Cloud Ice Water Content (IWC) Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T ClO S T S CO T GPH S T H S 2 O T HCl S T HCN S T HNO S 3 T HO S 2 T HOCl S T IWC S T IWP S T N S 2 O MLS v4, v3 and v2 IWC comparisons for July 2009 at 146 hPa and 100 hPa. (a) Left: Probability density Figure 3.15.1: T functions (PDF) (v4, red; v3, blue; and v2, green) with dashed lines showing the corresponding noise levels (ob- S O 3 tained by folding the negative IWC values about the origin) and the thin black lines representing the gaussian error T function. (b) Right: Scatter plots of IWC v4 vs. v2 (black points) with dashed red lines indicating the 1:1 line, dashed OH S uncertainties and the blue lines are linear ts to the data. σ yellow lines the 1 T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 105 Level 2 Version 4.2 Quality T

112 Help 3.15. Cloud Ice Water Content (IWC) Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T Table 3.15.1: Summary of MLS v4.2x IWC precision, accuracy, and resolution. ClO S Valid IWC Typical a c 3 T Resolution / d b Accuracy / mg/m range precision / Pressure / hPa S CO 3 3 km 3 3 10mg/m 10mg/m > < / mg/m mg/m T Unsuitable for scientic use 70 < p GPH S 0.02–50 7 × × 83 0.07 200 100% — 5 T 0.10 × 7 × 5 200 100% 150% 0.02–50 100 H S 2 O 100% 7 4 0.15 100% × 0.04–50 × 121 250 T 7 × 4 0.25–0.35 100% 100% 0.1–50 147 300 × HCl S 7 177 × 4 0.5–1.0 150% 100% 0.3–50 300 × T 7 0.6–50 215 300 100% 300% 1.2–2.1 4 × × HCN S 260 p > Unsuitable for scientic use T a HNO S ¿e along-track, cross-track and vertical extent, respectively of the atmospheric volume sampled by an individual MLS mea- surement. 3 b T precisions where the better values are for the extratropics and the poorer values for the tropics. ¿e precision σ ¿ese are typical 1 HO S for a particular measurement must be evaluated on a daily basis using the method described in the text. c 2 Estimated from comparisons with CloudSat. T d ¿is is the range where the stated precision, accuracy and resolution are applied. In this range MLS measurements can be HOCl S quantitatively interpreted as the average IWC for the volume sampled. IWC values above this range, currently giving qualitative information on cloud ice, require further validation for quantitative interpretation. T IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 106 Level 2 Version 4.2 Quality T

113 Help Overview 3.16 Cloud Ice Water Path (IWP) Table (stored as an additional swath in the IWP Swath name: L2GP-IWC le). 2 Units: g/m BrO S 6 km Useful range: MLS IWP is the ice water column above ~ T CH S > Contact: < Email: Alyn Lambert, [email protected] 3 Cl T CH S 3.16.1 Introduction 3 CN ) of the 240-GHz window channel at MLS standard IWP is retrieved from cloud-induced radiances ( T cir T 650hPa tangent pressure (see Figure 3.16.1). It represents a partial column above ~ 6 km, and is stored in CH S the v4.2 x L2GP IWC le as a separate swath. For the IWP retrieval, T is rst converted to a near horizontal 3 cir OH ○ –hIWP relation. ¿e hIWP is 3 slant path (with a elevation angle) IWP “hIWP”, using the modeled T ~ cir T then converted to the nadir IWP at the tangent point location, and interpolated to the MLS standard hori- ClO S zontal grid. T S CO 3.16.2 Resolution T In the IWP ranges specied in the summary at the end of this section, each MLS measurement can be quan- GPH S titatively interpreted as the average IWP for the volume sampled. ¿e MLS IWP volume is a vertical column T 6km, with 60km and 7km along and cross track extent respectively. ~ above H S 2 O T 3.16.3 Precision HCl S ¿e precision values quoted in the IWP swaths do not represent the true precision of the data. ¿e precision T for a particular measurement must be evaluated on a daily basis using the method described in the screening HCN 2 S for typical values precision given the summary at the end of this section reects section below. ¿e 3g/m MLS IWP measurements. T HNO S 3.16.4 Accuracy 3 T ¿eIWPaccuracyis ~ 50%,asestimatedfromcomparisonsoftheearlierv2.2MLSdataproductwithCloudSat HO S Wu et al. [2009]. and detailed analyses on the v2.2 error budget can be found in 2 T HOCl S 3.16.5 Data screening 2 T Sensitivity: ¿e standard IWP product has a useful sensitivity up to 200g/m where MLS measurements can IWC S be quantitatively interpreted as the average IWP for the volume sampled. T Use Temperature Status, Quality and Convergence: ¿euserisrecommendedtoscreentheIWPdatausing IWP S Status the eld in the collocated temperature prole to exclude bad retrievals [ Schwartz et al. , 2008]. T is an even number should be used. In other words, only IWP values for which temperature Status N S 2 Similarly, the users should only use IWP values where the corresponding Temperature proles have O T of 0.9 or larger and less than 1.03. Quality Convergence S O 3 Other screening: the user is also recommended to screen the IWP data for signicant cloud hits on a daily T 3 σ threshold is needed for basis using the “ 2 σ – 3 σ ” method described in the IWC section (3.15). ¿e OH S cloud detection since a small percentage of clear-sky residual noise can result in a large percentage of T false alarm in cloud detection. RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 107 Level 2 Version 4.2 Quality T

114 Help 3.16. Cloud Ice Water Path (IWP) Overview 3.16.6 Artifacts High-latitudehigh-landsurfacecanbemistakenlydetectedasacloudwhentheatmosphereisverydry, allow- Table ing MLS 240-GHz radiances to penetrate down to the surface. Surface emission/scattering can then reduce brightness temperature. Surface eects (e.g., over the highland over Antarctica) may introduce articial IWP 2 BrO S values as large as 10g/m . In addition, the geographical location of MLS IWP is currently registered at the tangent point, which is 2 proles away from the actual location of the IWP column as shown in Figure 3.16.1. ~ T CH ¿e user needs to correct this oset by replacing the IWP location with the one at 2 proles earlier. S 3 Cl T 3.16.7 Comparisons with other datasets CH S Comparedtov2.2IWPthev4.2 x IWPvaluesaresystematicallylargerby ~ 2%andtherandomnoiseisslightly 3 CN IWP is similar x smaller than in v2.2 (see Figure 3.16.2). Apart from the dierences noted above, the MLS v4.2 T [2009]. A revised validation paper for IWP is Wu et al. to the MLS v2.2 product described and validated in CH S 3 [2009] for more information. Wu et al. not planned in the near future and users are advised to read OH 50% < Comparisons between v2.2 MLS and CloudSat IWP showed good agreement with PDF dierences T for the IWP range specied in the summary at the end of this section. ClO S T 3.16.8 Desired improvements for future data version(s) S CO T measurement. x ¿eIWPretrievalinv4.2 isasimplerst-orderconversion, appliedindependentlytoeach cir T Plansforfutureversionsincludedevelopmentof2-Dcloudy-skyradiativetransfermodel. ¿iswillallowIWP GPH S T to be retrieved jointly with the measurements from adjacent scans. cir T H S 2 3.16.9 Summary for IWP O T IWP Column Bottom: 6km (estimated from MLS radiative transfer model calculations). HCl S ¿e calculation of the bottom height of the IWP column depends on the tropospheric water vapor T loading and on the IWP itself and is discussed in Wu et al. [2009]. HCN S 2 Typical precision: 3g/m is the typical 1 σ precision. T HNO ¿e precision for a particular measurement must be evaluated on a daily basis using the method de- S scribed in the text. 3 T Accuracy: 50% (estimated from comparisons with CloudSat) HO S 2 Resolution: 60km along track, 7km across track (the volume of air sampled by MLS) T HOCl S 2 ≤ Valid IWP range: 200g/m T ¿is is the range where the stated precision, accuracy and resolution are applied. In this range MLS IWC S measurements can be quantitatively interpreted as the average IWP for the volume sampled. IWP valuesabovethisrange, currentlygivingqualitativeinformationoncloudice, requirefurthervalidation T IWP S for quantitative interpretation. T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 108 Level 2 Version 4.2 Quality T

115 Help 3.16. Cloud Ice Water Path (IWP) Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T ClO S T S CO T GPH S T H S 2 O T HCl S T HCN S T HNO S 3 T HO S 2 T HOCl S T IWC Figure 3.16.1: Diagram to illustrate the MLS IWC and IWP measurement. The dashed lines are the MLS tangen- S tial beams. At high tangent heights, the beams penetrate through the limb and become sensitive to a volume- T averaged IWC, whereas at low tangent heights the MLS beams cannot penetrate through the limb due to strong IWP S ◦ ~ ) angle, “hIWP”. 3 gaseous absorption and become only sensitive to a partial slant column of IWP, with a shallow ( T 300 km away from the tangent point, or ~ Note that the actual volume of the air represented by hIWP is centered N S 2 proles from the location of the nominal prole. ~ 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 109 Level 2 Version 4.2 Quality T

116 Help 3.16. Cloud Ice Water Path (IWP) Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T ClO S T S CO T GPH S T H S 2 O T HCl S T HCN S T HNO S 3 T HO S 2 Figure 3.16.2: MLS v4, v3 and v2 IWP comparisons for July 2009. (a) Left: Probability density functions (PDF) (v4 T HOCl (red), v3 (blue) and v2 (green)) with dashed lines showing the corresponding noise levels (obtained by folding the S negative IWP values about the origin) and the thin black lines representing the gaussian error function. (b) Right: T v2 (black points) with a dashed red lined indicating the 1:1 line and a linear t to the data Scatter plot of IWP v4 vs ̇ IWC S shown as a blue line. T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 110 Level 2 Version 4.2 Quality T

117 Help Overview 3.17 Nitrous Oxide (N O) 2 Table N2O Swath name: 68 – 0.46 hPa Useful range: BrO S [email protected] Contact: Alyn Lambert, Email: < > T CH S 3 Cl 3.17.1 Introduction T CH Redenition of the N O standard product for v4.2 x to use band 3 (190-GHz) signals S 2 3 CN ¿e standard product for v4.2 O is taken from the 190-GHz (“Core+R2”) retrieval ( N2O-190 ) instead of x N 2 T ) as used in previous versions. N2O-640 the 640-GHz (“Core+R4B”) retrieval ( CH S Aging of the MLS “band 12” (640-GHz) signals was revealed by analysis of housekeeping data during 3 OH June-August 2013 which showed declining output in counts, increasing radiance noise (variance) and also T increasing correlation between the band 12 radiance channels. As the 640GHz N O product began to show 2 ClO S further deterioration in quality a decision was made to turn o band 12 on August 6, 2013. Prior to that, the T level-2processingstreamfortheN Ostandardproductinv3.xwasswitchedtooutputthe“band3”(190-GHz, 2 S CO N2O-190 ) retrieval on 7 June 2013 and later data for N2O-640 are not recommended for scientic use. O data show slightly worse precision and resolution compared to the 640-GHz retrievals. ¿e 190-GHz N T 2 GPH S have been compared from launch until the end of band 12 operations. ¿ere N2O-640 N2O-190 Data from and > 30% at 100 hPa (although some seasonal variations are evident) and is a persistent high bias in N2O-190 of T H N2O-190 we recommend that at this particular pressure level only be used in consultation with the MLS S 2 O science team. A smaller bias of 5–10% at 68hPa in is seen compared to N2O-640 . From 46 to 3.2hPa N2O-190 T the biases are less than 5% (see Figure 3.17.1). HCl S and N2O-640 data such as the resolution, preci- Other signicant dierences between the v4.2 x N2O-190 T sion, recommended pressure levels, quality and convergence criteria are noted below. HCN S x N O 640-GHz product is available for the period from launch until June 6, 2013 and ¿e secondary v4.2 2 T stored in the N2O-640 les in the L2GP-DGG swath. Details of the retrieval method and validation results are HNO S presented in [ , 2007]. Lambert et al. 3 T HO S 3.17.2 Resolution 2 ¿e spatial resolution reported by the averaging kernel matrices is shown in Figures 3.17.2 for the 190GHz T HOCl S les for data prior to L2GP-DGG measurements and 3.17.3 for the 640GHz measurements (available in the , the vertical resolution is 4–8km and the horizontal along-track resolution is N2O-190 June 6, 2013). For T 300–600km. For , the vertical resolution is 4–6km and the horizontal along-track resolution is N2O-640 IWC S 300–600km over most of the useful range of the retrievals. T ¿e horizontal cross-track resolution is set by the 7km width of the MLS 190-GHz eld-of-view for all IWP S pressures. Note that the higher frequency MLS 640-GHz measurements have a narrower 3km eld-of-view. T ○ ○ ¿e longitudinal separation of the MLS measurements is 10 over middle and lower latitudes, with much –20 N S 2 ner sampling in polar regions. O T S O 3.17.3 Precision 3 T σ Precision as dened here is the 1 uncertainty in the retrieved value calculated by the Level 2 algorithms OH S and has been validated against the scatter about the mean of coincident ascending/descending MLS prole T dierences. ¿e estimated precision on a single retrieved prole given in Tables 3.17.1 and 3.17.2 varies with RHI S N2O-190 and . 15–20ppbv for ~ ~ height from 12–25ppbv for N2O-640 O values at the 147hPa pressure level have a large a priori inuence and practically all precisions ¿e N T 2 SO S are agged negative at this level. 2 T S T Aura Microwave Limb Sounder (MLS) x 111 Level 2 Version 4.2 Quality T

118 Help 3.17. Nitrous Oxide (N O) 2 Overview Zonal Means for Data Over March, 2009 Table N2O-640, v04.2x N2O-190, v04.2x 68 hPa 100 hPa 46 hPa BrO S 500 350 300 T 450 CH S 250 300 3 400 Cl T 250 350 200 CH S 3 O (190) / ppbv CN 2 300 N 200 150 T 250 CH S 3 200 150 100 OH o o o o o o o o o o o o N N 45 EQ S 45 S 45 S S 90 N 90 90 45 EQ S 45 S 90 N 90 N 45 90 EQ N T 22 hPa 32 hPa 15 hPa ClO S 300 300 300 T 250 250 S CO 250 T 200 200 GPH 200 S 150 150 O (190) / ppbv T 2 N H 150 S 2 100 100 O T 50 50 100 HCl S o o o o o o o o o o o o EQ N 45 S N 90 N N 90 S N 90 EQ S 90 45 45 S 45 45 S EQ 45 S N 90 90 T 10 hPa 6.8 hPa 4.6 hPa HCN 200 120 250 S 100 T 200 150 HNO S 80 150 3 T 60 100 HO S 100 O (190) / ppbv 2 40 2 N T 50 50 HOCl 20 S 0 0 0 T o o o o o o o o o o o o 90 EQ N 45 S N 90 S N 90 90 45 45 S S EQ 45 45 N 90 90 N S EQ 45 S N IWC S 3.2 hPa 2.2 hPa 1.5 hPa 40 25 60 T IWP S 50 20 30 T 40 15 N S 2 O 10 20 30 T O (190) / ppbv 2 S 5 20 O N 3 10 T 10 0 OH S 0 0 -5 o o o o o o o o o o o o 90 90 N 90 45 EQ N S 90 N EQ 45 EQ S 45 45 S S 90 S 90 N 45 S N N 45 T Latitude / Degrees Latitude / Degrees Latitude / Degrees RHI S T SO S N2O-190 Figure 3.17.1: and N2O-640 comparison for March 2009 MLS v4.2x 2 T S T Aura Microwave Limb Sounder (MLS) 112 Level 2 Version 4.2 x Quality T

119 Help 3.17. Nitrous Oxide (N O) 2 Overview Table Equator FWHM / km FWHM / km 800 600 400 200 0 1000 -2 0 2 4 6 8 10 12 1200 0.1 BrO S T CH S 3 Cl 1.0 T CH S 3 CN T Pressure / hPa CH 10.0 S 3 OH T ClO S 100.0 T 0.0 0.8 1.0 -2 0 2 4 0.6 -4 -0.2 0.2 0.4 1.2 S CO Profile number Kernel, Integrated kernel 0 N 70 FWHM / km FWHM / km T 200 400 600 800 -2 1200 0 2 4 6 8 10 12 1000 0 GPH 0.1 S T H S 2 O T 1.0 HCl S T HCN S Pressure / hPa T 10.0 HNO S 3 T HO S 2 100.0 T -4 0.6 1.2 1.0 0.8 -2 0.4 0.2 0.0 -0.2 0 2 4 HOCl S Profile number Kernel, Integrated kernel T IWC S Typical two-dimensional (vertical and horizontal along-track) averaging kernels for the MLS v4.2x Figure 3.17.2: ◦ T O data at the equator (upper) and at 70 190 GHz N N (lower); variation in the averaging kernels is suciently small 2 IWP S that these are representative of typical proles. Colored lines show the averaging kernels as a function of MLS re- trieval level, indicating the region of the atmosphere from which information is contributing to the measurements T on the individual retrieval surfaces, which are denoted by plus signs in corresponding colors. The dashed black N S 2 O line indicates the resolution, determined from the full width at half maximum (FWHM) of the averaging kernels, T approximately scaled into kilometers (top axes). (Left) Vertical averaging kernels (integrated in the horizontal di- S O mension for ve along-track proles) and resolution. The solid black line shows the integrated area under each 3 kernel (horizontally and vertically); values near unity imply that the majority of information for that MLS data point T OH has come from the measurements, whereas lower values imply substantial contributions from a priori information. S (Right) Horizontal averaging kernels (integrated in the vertical dimension) and resolution. The horizontal averag- T ing kernels are shown scaled such that a unit averaging kernel amplitude is equivalent to a factor of 10 change in RHI S pressure. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 113 Level 2 Version 4.2 Quality T

120 Help 3.17. Nitrous Oxide (N O) 2 Overview Table Equator FWHM / km FWHM / km 800 600 400 200 0 1000 -2 0 2 4 6 8 10 12 1200 0.1 BrO S T CH S 3 Cl 1.0 T CH S 3 CN T Pressure / hPa CH 10.0 S 3 OH T ClO S 100.0 T 0.0 0.8 1.0 -2 0 2 4 0.6 -4 -0.2 0.2 0.4 1.2 S CO Profile number Kernel, Integrated kernel 0 N 70 FWHM / km FWHM / km T 200 400 600 800 -2 1200 0 2 4 6 8 10 12 1000 0 GPH 0.1 S T H S 2 O T 1.0 HCl S T HCN S Pressure / hPa T 10.0 HNO S 3 T HO S 2 100.0 T -4 0.6 1.2 1.0 0.8 -2 0.4 0.2 0.0 -0.2 0 2 4 HOCl S Profile number Kernel, Integrated kernel T IWC S Typical two-dimensional (vertical and horizontal along-track) averaging kernels for the MLS v4.2x Figure 3.17.3: ◦ T O data at the equator (upper) and at 70 640 GHz N N (lower); variation in the averaging kernels is suciently small 2 IWP S that these are representative of typical proles. Colored lines show the averaging kernels as a function of MLS re- trieval level, indicating the region of the atmosphere from which information is contributing to the measurements T on the individual retrieval surfaces, which are denoted by plus signs in corresponding colors. The dashed black N S 2 O line indicates the resolution, determined from the full width at half maximum (FWHM) of the averaging kernels, T approximately scaled into kilometers (top axes). (Left) Vertical averaging kernels (integrated in the horizontal di- S O mension for ve along-track proles) and resolution. The solid black line shows the integrated area under each 3 kernel (horizontally and vertically); values near unity imply that the majority of information for that MLS data point T OH has come from the measurements, whereas lower values imply substantial contributions from a priori information. S (Right) Horizontal averaging kernels (integrated in the vertical dimension) and resolution. The horizontal averag- T ing kernels are shown scaled such that a unit averaging kernel amplitude is equivalent to a factor of 10 change in RHI S pressure. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 114 Level 2 Version 4.2 Quality T

121 Help 3.17. Nitrous Oxide (N O) 2 Overview 3.17.4 Accuracy ¿e accuracy values given in Table 3.17.2 were obtained for both v4 N O products using the same detailed 2 Table [2007] to quantify the systematic uncertain- v2.2 data in Lambert et al. N2O-640 analyses presented for MLS ties associated with the MLS instrument calibration, spectroscopic uncertainty and approximations in the BrO retrieval formulation and implementation. S T 3.17.5 Data screening CH S 3 N2O-190 ): 68–0.46hPa Pressure range ( Cl T N2O-640 Pressure range ( ): 100–0.46hPa CH S 3 CN Values outside this range are not recommended for scientic use. In the upper stratosphere and lower x mesosphere v4.2 O requires signicant averaging for useful signals, but see the note under “Arti- N T 2 CH S facts” for issues at pressures below 0.1hPa. 3 OH Estimated precision: Only use values for which the estimated precision is a positive number. T Values where the information has a strong inuence are agged with negative or zero precision, a priori ClO S and should not be used in scientic analyses (see Section 1.5). T S CO Status ag: Only use proles for which the Status eld is an even number. Odd values of indicate that the prole should not be used in scientic studies. See Section 1.6 Status T GPH S eld. for more information on the interpretation of the Status T Quality ( eld is greater than 1.0 should be used. Quality ): Only proles whose N2O-190 H S 2 . See the discussion of this x Note that this is a change from the threshold originally suggested for v4.2 O T topic below. HCl S ): Only proles whose Convergence N2O-190 Convergence ( eld is less than 2.0 should be used. T data(typicallylessthan0.5%)atthislevelwillbediscardedviathisscreening. Afractionofthe N2O-190 HCN S eld is greater than 1.4 should be used. Quality ): Only proles whose N2O-640 Quality ( T HNO S proles (typically less than 0.5%) will be discarded via this screening. A small fraction of N2O-640 3 N2O-640 ): Only proles whose Convergence eld is less than 1.01 should be used. Convergence ( T HO S N2O-640 data(typicallylessthan0.5%)atthislevelwillbediscardedviathisscreening. Afractionofthe 2 T Clouds: Clouds have little impact on the N O products at the recommended pressure levels. Ignore status 2 HOCl S bit 16 (high cloud) or bit 32 (low cloud) indicating the presence of clouds. See artifacts for more T details. IWC S Updated Quality screening recommendation T IWP A byproduct of the aging in MLS that gives rise to the dri in the water vapor product (see section 3.9.8) and S O (see below) is that the Quality metric for the 190-GHz N O product exhibits a decreasing 190-GHz N 2 2 T 2009). ¿is decrease is not correlated with any actual decline ~ trend (most notable in the years following N S 2 O in the accuracy of the MLS N O product, which appears largely unchanged (dri issues aside). However, 2 T application of the previously recommended 1.3 threshold leads to rejection of an unreasonably large Quality S O 3 O proles in the more recent years of the Aura mission, particularly during southern winter. fraction of the N 2 T ¿e revised 1.0 Quality threshold restores these proles without erroneously identifying any obviously poor OH S proles as acceptable. T RHI S 3.17.6 Artifacts T O at 100hPa (not removed by screening), there are occasional In addition to the large bias of the 190-GHz N 2 SO S N2O-640 N2O-190 outliers at the highest presure levels in and products. Very thick clouds in the tropics 2 T S T Aura Microwave Limb Sounder (MLS) x 115 Level 2 Version 4.2 Quality T

122 Help 3.17. Nitrous Oxide (N O) 2 Overview N produce a low rate of artifacts in the v4.2 x O products since improvements in the handling of radiances 2 aected by clouds have reduced the frequency of outliers compared to previous versions. ¿e cloud bits of the Table Status eld are too blunt a tool to identify these rare cases, needlessly discarding reasonable data. Screening using the convergence and quality elds (see above) is recommended to remove the majority of these data points. BrO S 40 − ppbv (approximately three times the retrieval ¿e retrieval restricts N O values to be greater than 2 T noise level in the recommended pressure range) in order to prevent problems in the minimization search CH S 3 process. ¿e low bound is applied at all levels, but it is only evident in the data for pressures less than 0.1hPa, Cl T where the vertical smoothing is relaxed and the retrieval noise becomes comparable to the magnitude of the CH S low bound value. Accordingly, statistical averaging of the data (zonal means or longer time periods) cannot 3 CN ppbv hard limit introduces a positive 40 − be applied successfully for pressures at and less than 0.1hPa as the T bias in any average. CH S 0 per year has been observed in the MLS data product, through N2O-190 5% . A long-term trend of about − 3 OH analysis of the mean values in the equatorial lower stratosphere in the years following 2009. Since the long T year, the term secular increase in tropospheric N O observed by ground-based instruments is about + 0 . 25% ~ 2 ClO S N2O-190 trend is probably around − 0 . 75% ~ year. anomalous MLS N2O-190 dri giving rise to the observed T ¿e cause of this small but statistically signicant dri is under investigation. S CO T O versions and other datasets 3.17.7 Review of comparisons between MLS N 2 GPH S Average values for v4.2 O are up to 10% smaller than in v2.2 for the 100 and 68hPa pressure 190-GHz N x 2 T levels, and within a few percent for pressures greater than 46hPa (see Figure 3.17.4). Dierences between H S 2 v4.2 190-GHz N O and v3.3 are less than a few percent at all levels. x 2 O T Average values for v4.2 x 640-GHz N O are 20% larger than in v2.2 for the 100hPa pressure level, up 2 HCl S to 10% smaller at the 46–32hPa levels, and within 5% for pressures greater than 22hPa (see Figure 3.17.5). N2O-190 Dierences between v4.2 x 640-GHz N O and v3.3 are less than a few percent at all levels. Unlike the T 2 HCN S product, the N2O-640 product is recommended for scientic use at 100hPa. Apart from the dierences noted above, the MLS v4.2 x 640 GHz N O is similar to the MLS v2.2 prod- 2 T HNO uct described and validated in Lambert et al. [2007]. Comparisons of v2.2 640 GHz N O with coincident S 2 measuremements by ACE-FTS, Odin/SMR, and Envisat/MIPAS and balloon borne observations are shown 3 T Lambert et al. [2007]. A revised validation paper for N in O is not planned in the near future and users are 2 HO S [2007] for more information. Lambert et al. encouraged to read 2 T HOCl S 3.17.8 Desired improvements for future data version(s) may Retrievals of N O to pressures greater than 100hPa be possible in later versions from the 190-GHz ob- T 2 IWC S servations. T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 116 Level 2 Version 4.2 Quality T

123 Help 3.17. Nitrous Oxide (N O) 2 Overview Zonal Means for Data Over March, 2009 Table N2O-190, v03.3x N2O-190, v04.2x N2O-190, v02.2x 100 hPa 46 hPa 68 hPa BrO S 300 500 350 T 450 CH S 250 300 3 Cl 400 T 250 200 CH S 350 3 O (190) / ppbv CN 2 N 200 150 T 300 CH S 3 100 250 150 OH o o o o o o o o o o o o EQ N N 45 EQ EQ S S S 45 S 45 N 90 N 45 90 S 45 S 90 N 90 N 45 90 90 T 22 hPa 32 hPa 15 hPa ClO S 300 300 300 T 250 250 S CO 250 T 200 200 GPH 200 S 150 150 O (190) / ppbv T 2 N H 150 S 2 100 100 O T 100 50 50 HCl S o o o o o o o o o o o o 45 N N N 90 EQ N S 90 90 90 S 45 45 S 90 N S 45 S S EQ 45 EQ N 90 45 T 6.8 hPa 4.6 hPa 10 hPa HCN 200 250 100 S T 200 80 150 HNO S 60 150 3 T 100 HO S 40 100 O (190) / ppbv 2 2 N T 50 50 20 HOCl S 0 0 0 T o o o o o o o o o o o o N 90 EQ N 45 90 N 90 S N 90 45 S 45 S S EQ 45 45 N 90 90 N S EQ 45 S IWC S 1.5 hPa 3.2 hPa 2.2 hPa 25 40 60 T IWP S 50 20 30 T 40 N 15 S 2 O 20 30 T 10 O (190) / ppbv 2 S 20 O N 3 10 5 T 10 OH S 0 0 0 o o o o o o o o o o o o N EQ 45 EQ N S 90 N N 45 EQ S 45 45 S S 90 S 90 N 45 S 90 N 90 45 90 T Latitude / Degrees Latitude / Degrees Latitude / Degrees RHI S T SO S Figure 3.17.4: O compared to v2.2 and v3.3 for March 2009 MLS v4.2 190-GHz N 2 2 T S T Aura Microwave Limb Sounder (MLS) 117 Level 2 Version 4.2 x Quality T

124 Help 3.17. Nitrous Oxide (N O) 2 Overview Zonal Means for Data Over March, 2009 Table N2O-640, v04.2x N2O-640, v03.3x N2O-640, v02.2x 46 hPa 100 hPa 68 hPa BrO S 320 350 320 T 300 300 300 CH S 280 3 Cl 250 280 260 T CH S 240 260 200 3 O (640) / ppbv CN 2 N 220 T 150 240 200 CH S 3 220 100 180 OH o o o o o o o o o o o o N EQ 90 90 S 45 45 45 S S EQ S 90 45 S 45 N S EQ 45 N N 90 90 90 N N T 22 hPa 32 hPa 15 hPa ClO S 300 300 300 T 250 250 S CO 250 T 200 200 GPH 200 S 150 150 O (640) / ppbv T 2 N H 150 S 2 100 100 O T 50 50 100 HCl S o o o o o o o o o o o o 45 S N 90 S EQ 45 N 45 90 90 45 N 90 90 N 45 N EQ 90 N S 45 EQ S S S T 6.8 hPa 10 hPa 4.6 hPa HCN 200 250 120 S 100 T 200 150 HNO S 80 150 3 T 100 60 HO S 100 O (640) / ppbv 2 40 2 N T 50 50 HOCl 20 S 0 0 0 T o o o o o o o o o o o o S S 45 S N 90 45 N S 90 90 EQ 45 45 N 90 N N 90 90 EQ S 45 N S EQ 45 IWC S 3.2 hPa 1.5 hPa 2.2 hPa 25 40 60 T IWP S 20 30 T 40 15 N S 2 O 10 20 T O (640) / ppbv 2 S 5 20 O N 3 10 T 0 OH S 0 0 -5 o o o o o o o o o o o o 45 90 45 S EQ N N 45 90 N N 45 EQ S 90 S S N 90 90 90 S 45 S EQ 45 N T Latitude / Degrees Latitude / Degrees Latitude / Degrees RHI S T SO S Figure 3.17.5: MLS v4.2 640 GHz N O compared to v2.2 and v3.3 for March 2009 2 2 T S T Aura Microwave Limb Sounder (MLS) x 118 Level 2 Version 4.2 Quality T

125 Help 3.17. Nitrous Oxide (N O) 2 Overview Table ) product. Summary of MLS v4.2x N N2O-190 Table 3.17.1: O ( 2 Resolution a Accuracy Region Comments Precision BrO S Horiz. × Vert. hPa km ppbv % ppbv % T CH S — — — — Unsuitable for scientic use 0.33 ≤ — 3 Cl 100 655 11.0 > 100 2 > 15 × 0.46 T CH > 89 100 2 17 540 × 9.7 0.68 S 3 440 × 8.6 1.0 50 18 > 100 2 CN 2.2 7.3 × 340 19 64 4 36 T CH S × 4.6 11 6.5 17 23 285 5 3 OH 16 12 7 10 6.0 × 265 7 T 10 6 22 5.6 9 305 15 × ClO S × 30 10 7 14 385 46 4.9 55 22 5.4 × 420 16 6 68 T S CO × 3.8 7 124 44 Consult with MLS science team 100 500 20 — 147 — — — — Unsuitable for scientic use T GPH 215 — — — — Not retrieved — ≥ S T a Precision on individual proles H S 2 O T HCl S T HCN S T N2O-640 O ( Table 3.17.2: Summary of MLS v4.2x N ) product. 2 HNO S Resolution a Accuracy Region Comments Precision 3 Vert. × Horiz. T ppbv % ppbv % km hPa HO S 2 0.33 — — — — Unsuitable for scientic use ≤ — T HOCl S 8.9 × 530 12 > 100 1 88 0.46 0.68 100 2 89 > 13 430 × 7.4 T 345 14 > 100 2 50 1.0 6.2 × IWC S 15 49 2.2 4.8 × 300 3 27 T 295 6 14 × 14 18 4.6 4.1 IWP S 10 3.9 × 360 13 10 12 11 T 4.2 22 × 410 14 8 22 14 N S 2 16 46 7 40 19 495 4.4 × O T 5.5 8 70 28 × 20 68 555 S O 9 51 18 100 5.2 × 610 25 3 T 147 — — — — Unsuitable for scientic use — OH S — — — — Not retrieved ≥ 215 — T a RHI Precision on individual proles S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) 119 Level 2 Version 4.2 x Quality T

126 Help Overview 3.18 Ozone (O ) 3 Table O3 Swath name: 261 – 0.02 hPa Useful range: BrO S Contact: Lucien Froidevaux (stratosphere/mesosphere), T [email protected] < > Email: CH S Michael Schwartz (upper troposphere), > Email: < [email protected] 3 Cl T CH S 3 3.18.1 Introduction CN As is the case with previous versions, the v4.2 product is taken from a retrieval using 240-GHz x standard O T 3 CH S x radiances, providing sensitivity from the upper troposphere into the mesosphere. In v4.2 , an optimized re- 3 OH standard product has been added prior to the phase that uses 240-GHz trieval phase for production of the O 3 T radiances to produce the standard carbon monoxide and nitric acid products. Table 3.18.1 summarizes typical ClO S O resolution, precision, and systematic uncertainty estimates as a function of pressure. Papers describing 3 T detailed validation of the MLS v2.2 product and comparisons with other data sets were published in a spe- S CO , Jiang et al. , 2008b; Froidevaux et al. , see [ Journal of Geophysical Research cial Aura validation issue of the 2007; Livesey et al. , 2008]. In the stratosphere and above, v4.2 x ozone proles are very similar to the v2.2 and T GPH S x proles, so the stratospheric results from the above references will generally hold for the product. x /v3.4 v3.3 x Initial documentation of changes, improvements, and issues with v4.2 data are discussed here, including T H x data and is again data screening criteria. Recommended screening is less complex than it was for v3.3 x /v3.4 S 2 O primarily based on thresholds for , and Convergence , as was the case for v2.2. Ozone prole be- Quality T havior has been improved, with reduction of vertical oscillations in the low-latitude UTLS and reduction of HCl S sensitivity to thick clouds, through changes in the spectral form used to model cloud impacts on the MLS T radiances, through tightening of vertical smoothing at pressures greater than or equal to 68hPa (leading to an HCN S increase in averaging kernel FWHM by a factor of ~ 1.15) and through retrieval of an independent, spectrally- T at baseline over the ozone band for each limb view to better account for cloud inhomogeneity. HNO S In addition to the swath O3 , which is the O prole on 55 pressure surfaces, the L2GP-O3 les contain a 3 3 swath, O3 column , which is the integrated stratospheric column down to the thermal tropopause (WMO def- T HO S inition) calculated from MLS temperature. ¿e MLS temperature proles from which the WMO tropopause 2 is determined are not screened per the instructions of section 3.22 and users may wish to reject columns for T HOCl which standard screening rejects the corresponding temperature prole. S T x 3.18.2 Comparison of v4.2 with past data versions IWC S ; between 316hPa and 1hPa, v4.2 x /v3.4 x ¿e vertical retrieval grid has not changed from v3.3 ozone proles x T are retrieved on 12 surfaces per decade, a grid twice as ne as the 6-level-per-decade grid used in v2.2 Tight- IWP S ened vertical smoothing has degraded vertical resolution in v4.2 x from the upper troposphere to 68hPa by T ~ 15%. Table3.18.1 summarizes resolution, precision, and accuracy estimates. N S 2 Figures 3.18.1 and 3.18.2 show zonally-averaged stratospheric and mesospheric eld comparisons between O T dataversionsforthe(full)monthofMarch, 2009, forproperlyscreenedprolesonly; meandierences(ppmv S O and percent) are also shown. Similar plots focusing on the UTLS are provided in Figures 3.18.3 and 3.18.4. 3 T Average ozone mixing ratio values (e.g., for monthly means) have typically not changed by more than 1 to 2% OH S for pressures less than 100hPa. In the tropical UTLS (most notably), the new dataset is improved. Specically, ○ ○ –10 N in Figure 3.18.5. we show average UTLS proles during March, 2009 from three data versions for 0 T RHI S ¿e new version shows less oscillatory behavior in the mean proles, in comparison to the previous version (v3.3 on the same retrieval grid), although small oscillations are still present; certain months and latitude T SO regions exhibited larger eects (in the previous data version) than those shown in the above gure. S 2 T S T Aura Microwave Limb Sounder (MLS) x 120 Level 2 Version 4.2 Quality T

127 Help 3.18. Ozone (O ) 3 Overview Table BrO S T CH S 3 Cl T CH Averages for March, 2009 S 3 CN v2.2 O O v4.2 3 3 T CH S 0.1 0.1 3 OH T 1.0 1.0 ClO S T Pressure / hPa Pressure / hPa 10.0 10.0 S CO T 100.0 100.0 GPH o o o o o o o o o o o o S 60 N N S 30 90 S EQ 30 S N 60 90 60 S 60 N S 30 90 S EQ 30 90 N N Latitude Latitude T H S 1.0 2.0 3.0 0.5 4.0 5.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 6.0 0.5 1.0 7.0 8.0 9.0 2 O / ppmv O O / ppmv 3 3 T HCl S O Difference (v4.2-v2.2) % Difference (v4.2-v2.2) O 3 3 T HCN 0.1 0.1 S T HNO 1.0 1.0 S 3 T Pressure / hPa Pressure / hPa 10.0 10.0 HO S 2 T 100.0 100.0 HOCl o o o o o o o o o o o o S N S 30 90 S N 90 N 60 N EQ 30 EQ S S 30 S 90 60 S N 90 60 N 30 60 Latitude Latitude T IWC 0.05 0.03 0.01 0.00 -0.01 -0.03 -0.05 -0.10 -1 -2 -4 -6 -8 6 4 2 8 1 0 0.10 S Difference / ppmv Difference / % O O 3 3 T IWP S Figure 3.18.1: Zonal averages for stratospheric and mesospheric MLS ozone proles during March, 2009, showing the MLS v2.2 ozone mixing ratio contours (top left panel), the v4.2x contours (top right panel), and their dierences T in ppmv (v4.2x minus v2.2, bottom left panel) and percent (v4.2x minus v2.2 versus v2.2, bottom right panel). N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 121 Level 2 Version 4.2 Quality T

128 Help 3.18. Ozone (O ) 3 Overview Table BrO S T CH S 3 Cl T CH Averages for March, 2009 S 3 CN v3.3 O O v4.2 3 3 T CH S 0.1 0.1 3 OH T 1.0 1.0 ClO S T Pressure / hPa Pressure / hPa 10.0 10.0 S CO T 100.0 100.0 GPH o o o o o o o o o o o o S 60 N N S 30 90 S EQ 30 S N 60 90 60 S 60 N S 30 90 S EQ 30 90 N N Latitude Latitude T H S 1.0 2.0 3.0 0.5 4.0 5.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 6.0 0.5 1.0 7.0 8.0 9.0 2 O / ppmv O O / ppmv 3 3 T HCl S O Difference (v4.2-v3.3) % Difference (v4.2-v3.3) O 3 3 T HCN 0.1 0.1 S T HNO 1.0 1.0 S 3 T Pressure / hPa Pressure / hPa 10.0 10.0 HO S 2 T 100.0 100.0 HOCl o o o o o o o o o o o o S N S 30 90 S N 90 N 60 N EQ 30 EQ S S 30 S 90 60 S N 90 60 N 30 60 Latitude Latitude T IWC 0.05 0.03 0.01 0.00 -0.01 -0.03 -0.05 -0.10 -1 -2 -4 -6 -8 6 4 2 8 1 0 0.10 S Difference / ppmv Difference / % O O 3 3 T IWP S Figure 3.18.2: Zonal averages for stratospheric and mesospheric MLS ozone proles during March, 2009, showing the MLS v3.3 ozone mixing ratio contours (top left panel), the v4.2x contours (top right panel), and their dierences T in ppmv (v4.2x minus v3.3, bottom left panel) and percent (v4.2x minus v3.3 versus v3.3, bottom right panel). N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 122 Level 2 Version 4.2 Quality T

129 Help 3.18. Ozone (O ) 3 Overview Table BrO S T CH S 3 Cl T CH Averages for March, 2009 S 3 CN v2.2 O O v4.2 3 3 T CH S 3 OH T 100.0 100.0 ClO S T Pressure / hPa Pressure / hPa S CO T GPH o o o o o o o o o o o o S 90 S EQ 30 90 N 60 N N S 30 90 S EQ 30 S N 60 90 60 S 60 N S 30 N Latitude Latitude T H S 0.00 0.02 0.04 0.06 0.08 0.10 1.00 0.15 0.20 1.50 0.60 0.40 0.20 0.15 0.10 0.08 0.06 0.04 0.80 0.00 0.40 0.60 0.80 1.50 1.00 0.02 2 O O / ppmv O / ppmv 3 3 T HCl S O O % Difference (v4.2-v2.2) Difference (v4.2-v2.2) 3 3 T HCN S T 100.0 100.0 HNO S 3 T Pressure / hPa Pressure / hPa HO S 2 T HOCl o o o o o o o o o o o o S 30 S N 90 60 S 60 30 S 30 N 90 90 S EQ 30 S N 60 60 N 90 N N S EQ Latitude Latitude T IWC -0.03 -0.01 0.00 0.01 0.03 0.05 0.10 -0.10 -0.05 -100 -50 -25 -10 -5 0 5 100 50 25 10 S O Difference / % O Difference / ppmv 3 3 T IWP S Figure 3.18.3: Zonal averages for UTLS MLS ozone proles during March, 2009, showing the MLS v2.2 ozone mixing ratio contours (top left panel), the v4.2x contours (top right panel), and their dierences in ppmv (v4.2x minus v2.2, T bottom left panel) and percent (v4.2x minus v2.2 versus v2.2, bottom right panel). N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) 123 Level 2 Version 4.2 x Quality T

130 Help 3.18. Ozone (O ) 3 Overview Table BrO S T CH S 3 Cl T CH Averages for March, 2009 S 3 CN v3.3 O O v4.2 3 3 T CH S 3 OH T 100.0 100.0 ClO S T Pressure / hPa Pressure / hPa S CO T GPH o o o o o o o o o o o o S 90 S EQ 30 90 N 60 N N S 30 90 S EQ 30 S N 60 90 60 S 60 N S 30 N Latitude Latitude T H S 0.00 0.02 0.04 0.06 0.08 0.10 1.00 0.15 0.20 1.50 0.60 0.40 0.20 0.15 0.10 0.08 0.06 0.04 0.80 0.00 0.40 0.60 0.80 1.50 1.00 0.02 2 O O / ppmv O / ppmv 3 3 T HCl S O O % Difference (v4.2-v3.3) Difference (v4.2-v3.3) 3 3 T HCN S T 100.0 100.0 HNO S 3 T Pressure / hPa Pressure / hPa HO S 2 T HOCl o o o o o o o o o o o o S 30 S N 90 60 S 60 30 S 30 N 90 90 S EQ 30 S N 60 60 N 90 N N S EQ Latitude Latitude T IWC -0.03 -0.01 0.00 0.01 0.03 0.05 0.10 -0.10 -0.05 -100 -50 -25 -10 -5 0 5 100 50 25 10 S O Difference / % O Difference / ppmv 3 3 T IWP S Figure 3.18.4: Zonal averages for UTLS MLS ozone proles during March, 2009, showing the MLS v3.3 ozone mixing ratio contours (top left panel), the v4.2x contours (top right panel), and their dierences in ppmv (v4.2x minus v3.3, T bottom left panel) and percent (v4.2x minus v3.3 versus v3.3, bottom right panel). N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) 124 Level 2 Version 4.2 x Quality T

131 Help 3.18. Ozone (O ) 3 Overview Aura MLS Ozone for 0-10N, March 2009 40 Table version 2.2 version 3.3 70 BrO S version 4.2 T 100 CH S 3 Cl T CH S Pressure / hPa 3 CN 200 T CH S 3 300 OH T 400 ClO S 0.01 0.10 1.00 O3 / ppmv T S CO ◦ Zonal means from equator to 10 N during March, 2009, showing the average behavior of ozone Figure 3.18.5: T proles for MLS v2.2, v3.3, and v4.2x, as indicated by the color-coded legend. GPH S T H Figure 3.18.6 shows histograms of tropical UTLS v04.20 and v03.30 ozone for all of 2005. V04.20 ozone S 2 O unscreened (blue) and screened (green) histograms are quite similar to one another, with only small reduc- T tions in scatter in the screened case. V3.3 has signicantly broader distributions in the troposphere than HCl S v04.20, both in the widths of the central peaks and in the tails of the distributions. While screened v3.3 (cyan) T has reduced tails of cloud-impacted outliers compared to unscreened (red), these tails are clearly reduced in HCN S v4.2 x even before screening. ¿e reduced tropical vertical oscillation seen in Figures 3.18.4 and 3.18.5 is re- T values at 121hPa and lower ected in the shi s of the histogram peaks between versions, with higher v4.2 x HNO S values at 82, 100, 147, 178 and 316hPa. ¿e 316-hPa level of v4.2 x , though not currently recommended for 3 general scientic use, is under evaluation. T HO S 2 Ozone Columns T HOCl Changes in the MLS stratospheric ozone columns between v3.3 are quite small; typical x and v4.2 x /v3.4 x S daily zonal averages are within one percent, with a tendency towards slightly lower values (but within about T 1%) in the new data version. Only one stratospheric column swath, using a tropopause derived from MLS IWC S retrieval. ¿e column based upon the GEOS-5 tropopause is no temperature, is calculated by the v4.2 x O 3 T longer routinely produced. IWP S T 3.18.3 Resolution N S 2 Typical resolution values are provided in the summary Table 3.18.1. ¿e cross-track resolution is set by the O T 6km width of the MLS 240GHz eld of view. Daily longitudinal separation of MLS measurements, set by ○ ○ S O the Aura orbit, is 10 over middle and lower latitudes, with much ner sampling at the highest latitudes –20 3 sampled in each hemisphere. T OH S 3.18.4 Precision T RHI S As found previously, the Level 2 precision values are o en slightly lower than the scatter observed in a narrow latitude band centered around the equator (where atmospheric variability is expected to be small) or obtained T from a comparison between ascending and descending coincident MLS proles. SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 125 Level 2 Version 4.2 Quality T

132 Help 3.18. Ozone (O ) 3 Overview 121hPa 215hPa 68.1hPa 4 10 80 127 ppbv ± 455 ± 30 ppbv ± 54 38 ppbv 127 ppbv 37 ppbv ± 80 455 ± 56 ± 29 ppbv 3 40 ± 70 ppbv 149 ppbv 450 85 ppbv ± 75 ± 10 ± 68 130 ppbv ± 41 ppbv ± 52 51 ppbv 444 Table 2 10 1 10 BrO S 0 10 Observations per ppbv T 261hPa 147hPa 82.5hPa CH 4 S 10 37 ppbv 29 ppbv 82 ppbv 64 ± ± 52 274 ± 3 53 ± 35 ppbv ± 273 ± 82 ppbv 67 29 ppbv Cl 3 59 ± 97 ppbv ± 317 81 65 ppbv ± 66 ppbv 10 ± 314 78 ± 67 79 ppbv 43 ppbv 41 ppbv ± T 2 CH 10 S 3 1 CN 10 0 T 10 CH Observations per ppbv S 3 178hPa 100hPa 316hPa OH 4 10 91 ± 51 ppbv 50 78 ± 54 ppbv ± 28 ppbv ± 81 27 ppbv ± 52 ± 91 51 ppbv 50 ppbv T 3 91 ppbv 50 ± 66 ppbv ± 88 ppbv 90 114 ± 10 ± 112 ± 62 ppbv 39 ppbv ± 62 103 67 ppbv ClO S 2 10 T 1 10 S CO 0 10 Observations per ppbv T GPH 0 −200 0 200 400 600 −200 600 400 600 400 200 0 200 −200 S O (ppbv) (ppbv) O (ppbv) O 3 3 3 T screened v03.30 unscreened v03.30 screened v04.20 unscreened v04.20 H S 2 O Histograms of screened and unscreened v04.20 and v03.30 ozone for the 2005 tropical average Figure 3.18.6: T ◦ ◦ (20 S–20 N). HCl S T HCN S Negative precision values for ozone occur for almost every data point at pressures smaller than 0.01hPa, a priori indicating increasing inuence from the , although some MLS information exists (e.g., regarding av- T HNO erage day/night dierences) into the uppermost mesosphere and lower thermosphere. Generally, however, S we recommend that scientic studies be restricted to pressures of 0.02hPa or larger. 3 T HO S Column values 2 T ¿e estimated precisions for the v4.2 MLS column ozone abundances down to pressures of 100 to 215hPa x HOCl S ) variability in the tropics is 2 to σ are 2% or less. ¿e typical empirical precision in the columns based on (1- 3%. T IWC S 3.18.5 Accuracy T ¿e accuracy estimates shown in Table 3.18.1 are from an analysis which propagated estimated systematic IWP S ) measurement system. ¿e values shown errors in MLS calibration, spectroscopy, etc., through the (v4.2 x T σ estimates of accuracy. Comparison of MLS O here are intended to represent 2 with well-established data 3 N S 2 O sets shows no evidence of signicant biases overall, although the oscillations in the UTLS at low latitudes lead T to some systematic eects which can reach 20-40%, depending on the pressure level. For more details, see S O 3 [2008b], Froidevaux et al. the MLS validation papers by [2008], as well as Livesey et al. [2007], and Jiang et al. T referencestherein; somemorerecentreferencesrelevanttoMLSozoneareavailableontheMLSwebsiteunder OH S [2016] (more related information Hubert et al. “Publications”. In particular, see the comprehensive work by T further below). In recent years, validation studies have focused more on longer-term changes or dri s (with RHI S good stability being observed versus other reliable measurements) and on the UTLS region. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) 126 Level 2 Version 4.2 x Quality T

133 Help 3.18. Ozone (O ) 3 Overview Table FWHM / km BrO S 2 6 0 4 -2 10 8 12 0.001 T CH S 0.010 3 Cl T CH 0.100 S 3 CN T 1.000 CH S Pressure / hPa 3 OH 10.000 T ClO S 100.000 T S 0.0 1.2 1.0 -0.2 0.8 0.2 0.4 0.6 CO Kernel, Integrated kernel T FWHM / km GPH S 0 400 800 1200 600 1000 200 0.001 T H S 2 0.010 O T HCl S 0.100 T HCN S 1.000 T Pressure / hPa HNO S 10.000 3 T 100.000 HO S 2 T 2 -2 4 -4 0 HOCl S Profile number T Typical two-dimensional (vertical and horizontal along-track) averaging kernels for the MLS v4.2x Figure 3.18.7: IWC S ◦ N; variation in the averaging kernels is suciently small that these are representative of typical proles. O at 35 3 T Colored lines show the averaging kernels as a function of MLS retrieval level, indicating the region of the atmo- IWP S sphere from which information is contributing to the measurements on the individual retrieval surfaces, which are denoted by plus signs in corresponding colors. The dashed black line indicates the resolution, determined from T the full width at half maximum (FWHM) of the averaging kernels, approximately scaled into kilometers (top axes). N S 2 O (Upper) Vertical averaging kernels (integrated in the horizontal dimension for ve along-track proles) and resolu- T tion. The solid black line shows the integrated area under each kernel (horizontally and vertically); values near unity S O imply that the majority of information for that MLS data point has come from the measurements, whereas lower 3 values imply substantial contributions from a priori information. (Lower) Horizontal averaging kernels (integrated T in the vertical dimension) and resolution. The averaging kernels are scaled such that a unit change is equivalent to OH S one decade in pressure. T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 127 Level 2 Version 4.2 Quality T

134 Help 3.18. Ozone (O ) 3 Overview Table BrO S T CH S 3 Cl T FWHM / km CH S 0 4 6 8 10 12 -2 2 3 CN 10 T CH S 3 OH T ClO S 100 T S CO Pressure / hPa T GPH S T 1000 H S 2 O 0.2 -0.2 0.0 0.4 1.2 1.0 0.8 0.6 T Kernel, Integrated kernel HCl S FWHM / km 0 200 400 800 1000 1200 600 T 10 HCN S T HNO S 3 T HO S 100 2 T Pressure / hPa HOCl S T IWC S 1000 T -4 2 4 0 -2 IWP S Profile number T N S Figure 3.18.8: As for 3.18.7 but zooming in on the upper troposphere and lower stratosphere region. 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 128 Level 2 Version 4.2 Quality T

135 Help 3.18. Ozone (O ) 3 Overview Column values Sensitivity tests using systematic changes in various parameters that could aect the accuracy of the MLS Table estimates) of about 4%, as an estimated accuracy for the MLS column σ 2 retrievals lead to possible biases ( values (from integrated MLS ozone proles down to 100, 147, and 215hPa). See also the (v2.2) validation pa- pers (and subsequent ozone-related publications, e.g., available from the MLS website) for results on column BrO S ozone comparisons versus satellite, sonde, and lidar data. T CH S 3 3.18.6 Data screening Cl T Pressure range: 261–0.02hPa. CH S Values outside this range are not recommended for scientic use. 3 CN Estimated precision: Only use values for which the estimated precision is a positive number. T CH S Values where the a priori information has a strong inuence are agged with negative or zero precision, 3 OH and should not be used in scientic analyses (see Section 1.5). T Status ag: Only use proles for which the Status eld is an even number. ClO S Odd values of Status indicate that the prole should not be used in scientic studies. See Section 1.6 T for more information on the interpretation of the Status eld. S CO Quality: Only proles whose Quality eld is greater than 1.0 should be used. T GPH S eld is less than 1.03 should be used. Convergence Convergence: Only proles whose T Clouds: Scattering from thick clouds can lead to more systematic eects in the UTLS. H S 2 O Quality and Convergence screening recommenda- Most of the aected proles are removed by the T issues occur only rarely). tions (although Convergence HCl S Status or even Status proles with Convergence above the con- One should reject proles with odd T Quality or below the quality threshold. Conversely, one should keep prole values vergence threshold HCN S Convergence . ¿ese criteria typically remove 1 to 2 % Quality good and with even status good and T of global daily data, with tropical latitudes showing somewhat larger data removal fractions of about HNO S 5%. ¿is screening generally maintains sucient coverage for a near-complete daily map (for any given 3 T day), even in the UTLS. HO S Compared to data screening recommendations for past data versions, the screening of v4.2 data gen- x 2 T erally removes somewhat fewer ozone proles on a typical day. HOCl S 3.18.7 Review of comparisons with other datasets T IWC comparisonshaveindicatedgeneralagreementatthe5–10%levelwithstratosphericprolesfromanum- O S 3 ber of comparisons using satellite, balloon, aircra , and ground-based data. A high MLS v2.2 bias at 215hPa T had been observed in some comparisons versus certain ozonesonde and satellite datasets. Such high biases IWP S were reduced in versions v3.3 x and v3.4 x , with additional smaller reductions in the ozone values in v4.2 x . We T have found that latitudinal and temporal changes observed in various correlative datasets are well reproduced N S 2 O by the MLS ozone product. Intercomparisons of a large variety of ozone measurements by satellite instru- T Tegtmeier et al. ments have been documented by [2013], as part of the analyses produced by the SPARC Data S O 3 Initiative; the Aura MLS ozone values compare quite favorably to the multi-instrument mean values as well as T to SAGE II ozone. ¿e temporal stability of the Aura MLS ozone dataset has been shown to be very good, in OH S [2016] includes both Hubert et al. [2012]). ¿e more recent work by Nair et al. comparison to lidar datasets ( T the lidar network and the ozonesonde network as references for a comprehensive satellite data intercompari- RHI S son study. ¿e latter authors show that average biases between MLS and latitudinally-binned data from lidar T and ozonesonde sites across the globe is typically within 5% or better, with poorer behavior (known verti- SO S cal oscillations) at low latitudes in the UTLS. In terms of dri s, the Aura MLS ozone dataset is shown (in the 2 T S T Aura Microwave Limb Sounder (MLS) x 129 Level 2 Version 4.2 Quality T

136 Help 3.18. Ozone (O ) 3 Overview Summary for MLS ozone Table 3.18.1: a b Precision Accuracy Pressure / Resolution Table Comments ppmv % ppmv % hPa × Vert. Horiz. ≤ — Unsuitable for scientic use — — — 0.01 — BrO S 5.5 50 1.2 300 0.2 200 × 0.02 × 200 0.8 150 0.2 30 0.05 5.5 T 0.1 60 0.2 4 × 400 20 0.5 CH S 3 3 0.2 450 0.4 × 30 0.15 10 Cl 20 0.2 10 3.5 × 0.5 0.3 550 T CH 10 1 3 × 500 0.2 7 0.3 S 3 3.5 2 0.15 × 0.3 3 450 7 CN 5 0.4 2 0.15 450 × 7 3 T CH 0.1 2 0.4 6 3 10 500 × S 3 × 400 0.1 2 2.5 5 22 0.25 OH 350 0.06 3 0.2 8 46 2.5 × T 0.04 4 0.1 7 2.5 350 68 × ClO S + 300 × 3 ] 7% 100 005 . 0 + [ 0.03 15–25 T 3 7% ] 0 005 + [ 0.02 5–70 400 × . 150 + S CO 215 3.5 × 350 0.02 5–100 [ + 0 . 01 + 10% ] c [ 261 3.5 × 400 0.03 5–100 02 + 0 . See note + 10% ] T GPH 316 2.5 × 550 0.03 — — — Not recommended S — — Not retrieved 1000–464 — — — T H S a 2 Precision on individual proles O b Primarily as estimated from systematic uncertainty characterization tests. Stratospheric values are expressed in ppmv with a T typical equivalent percentage value quoted. 261–100hPa accuracies are the sum of the ppmv and percentage uncertainties; the ppmv HCl S terms arise from observed average positive biases in the MLS values relative to tropical sonde data (from 2005 through 2015). c T Positive bias in the UT, but the mean annual variation is nevertheless well behaved versus tropical sonde data. HCN S T above work) to be very stable, a term also applied only to the SAGE II dataset; MLS exhibits dri s with respect HNO S to the ground-based networks that fall within 1.5 to 2% per decade, with zero dri encompassed by the error 3 T bars (i.e. non-signicant dri s), at least in the middle stratosphere. ¿e Aura MLS data, in combination with HO S older datasets, provides a critical tool for the study of global O in a changing climate and into the expected 3 2 T recovery period, as anthropogenic ozone depleting substances decline in concentration. O 3 HOCl S 3.18.8 Artifacts T IWC proles (especially at x ¿ere has been some improvement in the v4.2 Oscillations in tropical UTLS ozone: S low latitudes), in terms of reducing (but not eliminating) systematic vertical oscillations. Further char- T acterization of MLS tropical UTLS data, in particular, is warranted in order to more fully understand IWP S any data limitations in this region. T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 130 Level 2 Version 4.2 Quality T

137 Help Overview 3.19 Hydroxyl Radical (OH) Table OH Swath name: Useful range: 32 – 0.0032 hPa BrO S and Shuhui Wang, Contact: > [email protected] Luis Millan, < Email: T > Email: < [email protected] CH S 3 Cl T 3.19.1 Introduction CH S ¿e MLS THz radiometer is dedicated to measuring OH in the 2.5THz spectral region. A description of OH 3 CN [2008]. ¿e Pickett et al. data quality, precision and systematic errors for an earlier version, v2.2, is given in T Pickett et al. [2008]. While the OH data quality is [2008] and Wang et al. validation studies are described in CH S , there are signicant improvements in the current version x and v3.4 x generally similar between v2.2 and v3.3 3 OH ~ x . In previous versions, OH data near the mesospheric density peak, 0.032 hPa, o en show considerable v4.2 T amount of data agged with negative precision, indicating strong inuence of a priori information, particu- ClO S larly in the summer hemisphere (or tropics when near the equinox) and sometimes causes gaps in zonal mean T x so ware resolves this issue by xing the overly tight time series. ¿e v4.2 a priori constraints. ¿e result- S CO ing mesospheric OH data are less noisy and generally have somewhat larger values than previous versions T for the problematic seasons/latitudes. Another improvement is the smaller bias at 10–15hPa. ¿erefore, the GPH S day-night correction for bias is only required for pressure levels at 21 and 32hPa. Note that users may notice T more zig-zags in v4.2 x nighttime mesospheric OH vertical proles. ¿is is a side eect of xing the possible H S 2 positive bias introduced by tight lower limits set in previous version retrievals. O T ¿e estimated uncertainties, precisions, and resolution for v4.2 x OH are summarized in Table 3.19.1. HCl S 3.19.2 Resolution T ○ Figure 3.19.1 shows the OH averaging kernel for daytime and nighttime at 35 N. ¿e reason to separate day- HCN S time and nighttime is that the largest natural variability in OH is diurnal. ¿e vertical resolution is slightly T dierent between day and night. ¿e nighttime resolution is sucient to allow the study of (for example) HNO S the “nighttime OH layer” around 82km. ¿e vertical width of the averaging kernel for pressures greater ○ 3 than 0.01hPa is2.5 km. ¿e horizontal width of the averaging kernel is equivalent to a width of 1.5 (165km T HO distance) along the orbit. ¿e changes in vertical resolution above 0.01hPa are due mainly to use of a faster in- S 2 strument vertical scan rate for tangent heights above 70km. ¿e horizontal resolution across track is 2.5km. T HOCl ¿e averaging kernel and resolution for high and low latitudes are very similar to Figure 3.19.1 for most pres- S sure levels. At the topmost two pressure levels, 0.0046hPa and 0.0032hPa, the vertical resolution is slightly T ○ N. better at the equator than at 70 IWC S 3.19.3 Precision T IWP and v3.4 x A typical OH prole and the associated precisions (for both v3.3 ) are shown in Fig- x and v4.2 x S ure 3.19.2. ¿e prole is shown in both volume mixing ratio (vmr) and density units. All MLS data are T 6 − 3 ) cm reported in vmr for consistency with the other retrieved molecules. However, use of density units (10 N S 2 O reduces the apparent steep gradient of OH vertical prole, allowing one to see the prole with more detail, T especially in the stratosphere where most atmospheric OH is present. Additionally, at THz frequencies the S O 3 collisional line-width is approximately equal to the Doppler width at 1hPa. Above 1hPa, Doppler broadening T is dominant and the peak intensity of OH spectral absorption is proportional to density, while below 1hPa OH S the peak intensity is proportional to vmr. ¿e daytime OH density prole shows two peaks at 45km and ~ T Pickett et al. , 2006a]. Precisions are such ~ 75km. ¿e night OH prole exhibits the narrow layer at ~ 82 km [ RHI S ○ latitude bin can be determined with better than 10% relative precision that an OH zonal average within a 10 T ~ with one day of data ( 100 samples) over 21–0.01hPa. With 4 days of data, the 10% precision limits can be SO S extended to 32–0.0046hPa. 2 T S T Aura Microwave Limb Sounder (MLS) x 131 Level 2 Version 4.2 Quality T

138 Help 3.19. Hydroxyl Radical (OH) Overview Table BrO S Day FWHM / km FWHM / km T 800 600 400 200 0 -2 0 2 4 6 8 10 12 1000 1200 0.001 CH S 3 Cl T 0.010 CH S 3 CN 0.100 T CH S 3 1.000 OH Pressure / hPa T ClO 10.000 S T 100.000 S CO 2 4 0.4 1.2 0.2 -4 0 0.6 -2 0.8 1.0 0.0 -0.2 Kernel, Integrated kernel Profile number T Night FWHM / km FWHM / km GPH S 200 12 600 800 1000 1200 400 -2 0 2 4 6 8 10 0 0.001 T H S 2 0.010 O T HCl S 0.100 T HCN S 1.000 Pressure / hPa T HNO S 10.000 3 T HO 100.000 S 0.8 2 0 1.2 -2 4 0.6 0.4 0.2 0.0 -0.2 -4 1.0 2 Kernel, Integrated kernel Profile number T HOCl S Typical two-dimensional (vertical and horizontal along-track) averaging kernels for the MLS v4.2x Figure 3.19.1: ◦ T N for daytime (upper) and nighttime (lower); variation in the averaging kernels is suciently small OH data at 35 IWC S that these are representative of typical proles. Colored lines show the averaging kernels as a function of MLS re- trieval level, indicating the region of the atmosphere from which information is contributing to the measurements T IWP S on the individual retrieval surfaces, which are denoted by plus signs in corresponding colors. The dashed black line indicates the resolution, determined from the full width at half maximum (FWHM) of the averaging kernels, T approximately scaled into kilometers (top axes). (Left) Vertical averaging kernels (integrated in the horizontal di- N S 2 mension for ve along-track proles) and resolution. The solid black line shows the integrated area under each O T kernel (horizontally and vertically); values near unity imply that the majority of information for that MLS data point S O has come from the measurements, whereas lower values imply substantial contributions from a priori information. 3 (Right) Horizontal averaging kernels (integrated in the vertical dimension) and resolution. The averaging kernels T are scaled such that a unit change is equivalent to one decade in pressure. OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 132 Level 2 Version 4.2 Quality T

139 Help 3.19. Hydroxyl Radical (OH) Overview (a) OH v4 (b) OH v4 Table 1 0.001 0.010 BrO S T 0.100 CH S 3 Cl 10 T pressure / hPa pressure / hPa 1.000 CH S Night Day 3 Night Day CN 10.000 T CH S 3 0.6 0.8 0 15 -0.2 0.0 0.2 5 -5 0.4 10 OH vmr / ppb vmr / ppb T (d) OH v4 (c) OH v4 ClO S 0.001 0.001 T S CO 0.010 0.010 T GPH S 0.100 0.100 T H S pressure / hPa pressure / hPa 1.000 1.000 2 O T HCl S 10.000 10.000 T HCN S 10 -5 25 5 20 15 10 5 0 -5 15 0 25 20 -3 -3 6 6 density / 10 cm density / 10 cm T (f) OH v3 (e) OH v3 HNO S 0.001 0.001 3 T HO 0.010 0.010 S 2 T HOCl 0.100 0.100 S T pressure / hPa pressure / hPa 1.000 1.000 IWC S T 10.000 10.000 IWP S T 0 25 20 15 10 5 -5 -5 25 20 15 10 5 0 N S 6 -3 6 -3 2 cm density / 10 density / 10 cm O T Figure 3.19.2: Zonal mean of retrieved OH and its estimated precision (horizontal error bars) for September 20, S O ◦ ◦ 3 N. The average includes 98 proles. Panel (a) shows v4.2x OH vmr vs. pressure for N to 39 2005 averaged over 29 T day (black) and night (blue). Panel (b) shows the same data plotted for the stratosphere. The retrieved night OH OH S concentration is near zero for altitudes below 1 hPa. Panel (c) shows the same data in (a) converted into density T units. Panel (d) shows the day-night dierences for the data in panel (c). Note that the day-night dierence is RHI S required for altitudes below 20 hPa. Panels (e) and (f ) are equivalent to (c) and (d) but using v3.3x and v3.4x OH data. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 133 Level 2 Version 4.2 Quality T

140 Help 3.19. Hydroxyl Radical (OH) Overview 3.19.4 Accuracy Table 3.19.1 summarizes the accuracy expected for OH. ¿e eect of each identied source of systematic er- Table Read et al. ror on MLS measurements of radiance has been quantied and modeled [ , 2007]. ¿ese quantied σ eects correspond to either estimates of uncertainties in each MLS product, or an estimate of the max- 2 BrO S imum reasonable uncertainty based on instrument knowledge and/or design requirements.¿ese accuracy and v3.4 x estimates. calculations were performed with more realistic OH atmospheric proles than for v3.3 x T CH Biases can be eliminated by taking day-night dierences from 32–21hPa. For 15–0.1hPa, the observed night S 3 Cl OH concentration is small and day-night dierencing is not ordinarily needed. ¿e overall uncertainty is the T square root of the sum of squares of the precision and accuracy. CH S 3 CN 3.19.5 Data screening T It is recommended that OH data values be used in scientic investigations if all the following tests are suc- CH S 3 cessful: OH T Pressure range: 32–0.0032hPa. ClO S Values outside this range are not recommended for scientic use. T S CO Estimated precision: Only use values for which the estimated precision is a positive number. T Values where the a priori information has a strong inuence are agged with negative or zero precision, GPH S and should not be used in scientic analyses (see Section 1.5). T eld is an even number. Status Status ag: Only use proles for which the H S 2 O indicate that the prole should not be used in scientic studies. See Section 1.6 Status Odd values of T for more information on the interpretation of the Status eld. HCl S Quality: MLS v4.2 Quality eld. x OH data can be used irrespective of the value of the T HCN S Convergence eld is less than 1.1 should be used. Convergence: Only proles whose T 3.19.6 Artifacts HNO S For some seasons, the Gas Laser Local Oscillator (GLLO) for the THz receiver is automatically “relocked” as 3 T 257 and the prole many as 5 times during a day, leading to data gaps. In these cases the Status ag is set to HO S is ignored. ¿is can present a problem when compiling maps, because the missing data may appear at the 2 T same latitude and longitude on successive days. HOCl S 3.19.7 Review of comparisons with other datasets T IWC S Data from MLS v2.2 so ware have been validated with two balloon-borne remote-sensing instruments and with ground-based column measurements. Details of the comparison are given in Pickett et al. [2008] and T IWP and v3.4 show no signicant dierences Wang et al. [2008]. ¿e comparison among v2.2, v3.3 x , and v4.2 x x S OH daytime values are somewhat larger and less noisy in the summer x except for the mesosphere where v4.2 T hemisphere (or tropics when near equinox). N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 134 Level 2 Version 4.2 Quality T

141 Help 3.19. Hydroxyl Radical (OH) Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T ClO S T S CO Summary of precisions, resolution, and uncertainties for the MLS OH product Table 3.19.1: T a GPH S Precision Accuracy / Resolution Comments Pressure (day/night) 6 − 3 T 10 cm × H /km V 6 − 3 / 10 cm H S 2 O — — 0.003hPa < — Unsuitable for scientic use T 5.0 × 220 0.5 / 0.5 0.5 0.003hPa HCl S 180 2 1.1 /1.1 × 2.5 0.01 hPa 3.3 / 0.6 2.5 1.0 165 × 0.1hPa T HCN S 1.0hPa 1.0 2.5 165 1.9 / 0.4 × 165 10 hPa 2.5 × 2.3 / 1.4 0.9 T 1.3 6–10 / 4–8 165 × 2.5 32–14hPa Use day–night dierence HNO S — 32hPa — Unsuitable for scientic use — > 3 T a Precision on an individual prole HO S 2 T HOCl S T IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 135 Level 2 Version 4.2 Quality T

142 Help Overview 3.20 Relative Humidity with respect to Ice (RHI) Table RHI Swath name: Useful range: ≤ UTRHI, mean layer value for P 383 hPa, Prole from 316 – 0.002 hPa. BrO S Contact: William Read, Email: > [email protected] < T CH S 3 Cl 3.20.1 Introduction T RHi is relative humidity with respect to ice. ¿e vertical grid for RHi is 12 levels per decade change in pressure CH S for 1000–1.0 hPa thinning to 6 levels per decade for 1.0–0.1hPa and nally 3 levels per decade for 0.1– 3 CN 5 − 10 hPa. ¿e RHi product is a fusion of information from two separate retrievals. From 1000–383hPa, T RHi is retrieved directly from optically thick radiances using measurement and retrieval principles similar CH S 3 to nadir sounding humidity receivers (e.g., TOVS). All grid levels between 1000–383hPa are lled with a OH uniform “UTRHI” (upper tropospheric relative humidity with respect to ice) value representing the mean T ClO value of a broad layer ( ~ 4–6km) that peaks between ~ 650hPa (typical ~ 350hPa (in the moist tropics) and S a priori for the vertically for dry high latitudes). ¿is humidity is used as a lower altitude constraint and T resolved humidity product that begins at 316hPa. From 316–0.002hPa, RHi is derived from the standard S CO products of water and temperature using the Go-Gratch ice humidity saturation formula. RHi validation is T [2007]. Table 3.20.1 is summary of precision, resolution, and accuracy. Read et al. presented in GPH S T 3.20.2 Changes from v3 H S initial ¿emaindierencesbetweenv4.2andv3.3relatetocloudscreeningandimprovementsinthe InitUTH 2 O T upper tropospheric and lower stratospheric humidity estimation phase that produces the rst guess and sets HCl S smoothing constraints on the H O prole retrieved in the nal retrieval phase that produces the standard 2 product. ¿ere have been no spectroscopy changes either in linewidth or continuum charaterization between T HCN S v4 and v3. An improved cloud detection methodology has been developed for v4 that does a better job of rejecting cloudy radiances that are likely to cause poor ts and corrupted proles. ¿e number of erroneous T HNO phase retrieves H InitUTH spikes in the upper troposphere has been reduced in v4 relative to v3. ¿e O on a S 2 more coarse 6 levels per decade grid. ¿e H O is used for the initial guess for the nal 12 levels per InitUTH 2 3 T decade retrieval and sets smoothing constraints for the prole shape instead of smoothing to the shape of a HO S phasehasbeenexpandedtousemorechannelsandbetterforward InitUTH climatological prole. ¿e apriori 2 T model representations to improve its retrieval accuracy. In addition, the retrieval range has been expanded HOCl S to have its top moved from 100hPa (in v3) to 10hPa (in v4). ¿is change makes a more seamless transition from the troposphere to the stratosphere. Simulation studies show that these changes have improved the T IWC S agreement between the “truth” proles used to produce the simulated radiances and the retrieved proles in the 316–215hPa levels. T IWP Figure 3.20.1 compares MLS v4.2 to v3.3. RHI shows bigger dierences than that for H O because it also S 2 includes changes between the two versions for temperature. ¿e dierences sometimes exceed 20% and can T be either a low or high bias. N S 2 O Relative humidity data at pressures greater than 316hPa are derived from a broad layer relative humidity T retrieval (using low limb viewing MLS wing channel radiances) similar to that obtained from NOAA opera- S O 3 , 2007], the v2.2 retrieval at these pressures Read et al. tional humidity sounders such as TOVS. As noted in [ T 30% too high based on comparisons with AIRS. ¿e accuracy of this retrieval is highly sen- was likely to be ~ OH S sitive to the transmission eciency of the MLS optics system. In v3.3 this was adjusted empirically (within T the uncertainty range established from MLS calibration) to give better agreement with AIRS in the tropics. RHI S continuum was adjusted only for this phase to minimize the clear sky cloud induced radiance In v4 the N 2 T bias. ¿is retrieval is used as an a priori and prole constraint for the humidity prole at pressures greater SO S than 316hPa which are not retrieved in the standard H O product retrieval. 2 2 T S T Aura Microwave Limb Sounder (MLS) x 136 Level 2 Version 4.2 Quality T

143 Help 3.20. Relative Humidity with respect to Ice (RHI) Overview Table Profile Precision Differences 180W--180E, 60N--90N 12 v03.30 v02.21 17 v04.20 BrO S 26 38 56 T 82 CH S 121 3 Cl Pressure (hPa) 177 261 T 383 CH S 0 20 80 40 40 20 0 60 -40 0.0 5 10 15 20 -20 Profile Differences Precision 3 CN RHi (%) (%) RHi (%) 180W--180E, 30N--60N 12 v03.30 v02.21 T 17 v04.20 26 CH S 38 3 OH 56 82 T 121 ClO Pressure (hPa) S 177 261 383 T 20 80 20 0 15 10 5 20 40 -40 -20 0 40 60 0.0 Precision Profile Differences S CO (%) RHi (%) RHi (%) 180W--180E, 30S--30N 12 v03.30 v02.21 T 17 v04.20 GPH 26 S 38 56 T 82 H S 121 2 O Pressure (hPa) 177 T 261 383 HCl S 40 0 80 20 0.0 5 10 15 20 40 -40 -20 0 20 60 Differences Profile Precision RHi (%) (%) RHi (%) T 180W--180E, 60S--30S 12 v03.30 v02.21 HCN 17 S v04.20 26 38 T 56 HNO S 82 121 3 Pressure (hPa) 177 T 261 HO S 383 2 -20 -40 15 0 40 60 80 20 10 5 0.0 40 20 20 0 Precision Profile Differences T (%) RHi (%) RHi (%) 180W--180E, 90S--60S 12 HOCl v03.30 S v02.21 17 v04.20 26 T 38 IWC 56 S 82 121 T Pressure (hPa) 177 IWP S 261 383 T 0.0 20 0 -20 -40 0 40 20 15 10 5 20 60 40 80 N RHi (%) RHi (%) (%) S 2 O T S O 3 Figure 3.20.1: A comparison of v3.3 (blue), v2.2(green), and v4.2 (red) rhi for Jan-Feb-Mar 2005 in 5 latitude bands. T Other time periods are similar. The left panel compares mean proles, the center shows the mean dierence OH S (red and green diamonds) surrounded by each versions’ estimated precision, and the right panel shows the es- timated retrieval precision (solid and bullets) and measured variability (dotted) which includes atmospheric vari- T ability about the mean prole. RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 137 Level 2 Version 4.2 Quality T

144 Help 3.20. Relative Humidity with respect to Ice (RHI) Overview ¿e third panel in Figure 3.20.1 shows the mean estimated single prole precision and the measured variability (which includes instrument noise and atmospheric variability). ¿e precisions for the two versions Table are nearly identical except for pressures greater than 68hPa, where v4 produces lower values. 3.20.3 Resolution BrO S RHi for pressures of 316hPa and smaller is a derived product and therefore a retrieval averaging kernel is T not directly available. An estimate for the spatial resolution (vertical X along track) of this product is a con- CH S 3 O O resolutions. Since temperature has lower spatial resolution than H volution of the temperature and H Cl 2 2 T in the troposphere and lower stratosphere it is assumed that the spatial resolution of temperature shown in CH S Figure 3.22.4 best represents the resolution of the RHi product. ¿e cross track resolution is probably 12km, 3 CN the larger of temperature and H O cross track resolutions. ¿ese resolutions are only true in the limit that 2 T the mean log ( H O) doesn’t change appreciably over the broader temperature measurement volume. ¿e 2 CH S ○ ○ longitudinal separation of the MLS measurements, set by the Aura orbit, is 10 –20 over middle and lower 3 OH latitudes, with much ner sampling in polar regions. T ¿e RHi for pressures greater than 316hPa, represents a mean value in a broad layer (4–6km) whose ClO S ~ 650hPa (typical for dry high latitudes). ~ 350hPa (in the moist tropics) and sensitivity peaks between T S CO 3.20.4 Precision T O and temperature propagated ¿e values for precision are the root sum square (RSS) precisions for H 2 GPH S through the Go-Gratch relationship, see sections 3.9 and 3.22 for more details. ¿e precisions are set to T negative values (or zero in some cases) in situations when the retrieved precision is larger than 50% of the H S 2 a priori precision for either temperature or H O — an indication that the data is biased toward the a priori 2 O T value. HCl S 3.20.5 Accuracy T HCN S O and temperature scaled into %RHi units. see sec- ¿e values for accuracy are the RSS accuracies for H 2 tions 3.9 and 3.22 for more details. T HNO S 3.20.6 Data screening 3 T Pressure range: Prole from 316–0.002hPa, larger pressures represent mid/upper troposphere column. HO S Values outside this range are not recommended for scientic use. 2 T HOCl Estimated precision: Only use values for which the estimated precision is a positive number. S Values where the information has a strong inuence are agged with negative or zero precision, a priori T and should not be used in scientic analyses (see Section 1.5). IWC S Status eld is an even number. Status ag: Only use proles for which the T IWP S Status indicate that the prole should not be used in scientic studies. See Section 1.6 Odd values of for more information on the interpretation of the eld. Status T N S 2 Ignore status bits 16 (high clouds) or 32 (low clouds) set indicating the presence of clouds. See Clouds: O T artifacts for more details. S O 3 Quality Quality eld: ¿e swaths need to be con- and L2GP-Temperature elds for both the L2GP-RHI T sidered, as described below: OH S greater For pressures of 83hPa and smaller: Use only proles with RHi Quality than 1.45 and Tem- T RHI S greater perature than 0.2 Quality T greater than 1.45 and Tem- For pressures of 100hPa and larger: Use only proles with RHi Quality SO S greater Quality perature than 0.9. 2 T S T Aura Microwave Limb Sounder (MLS) x 138 Level 2 Version 4.2 Quality T

145 Help 3.20. Relative Humidity with respect to Ice (RHI) Overview than 2.0 and Temperature Convergence eld: Only proles with a value of the RHi less Convergence Convergence less than 1.03 should be used in scientic studies. Table L2GP-Temperature eldinthe L2gpPrecision ¿e Temperature precision: lecanbeusedtofurtherelim- inate outliers that are believed to be the result of thick clouds, primarily in the tropics. If careful screen- BrO S ing of the troposphere is required, levels 261–178hPa should be avoided if any of the following criteria are met: T CH S ○ 1 > − 60 K and 1 . latitude > L2gpPrecision At 316hPa: 3 Cl T 0 At 261hPa: L2gpPrecision 7 > . K CH S 3 > 0 . 825 At 215hPa: L2gpPrecision K CN T ¿e last four proles of each day show greatly increased rates of large temperature End of day (v4.20 only): CH S . Accordingly the last four proles of the RHI product each day should not be a priori departure from 3 OH used in v4.20. ¿e issue is xed in v4.22. T ClO S 3.20.7 Artifacts T O and 3.22 for temperature for specic issues related to these parent products. Eects See sections 3.9 for H 2 S CO ~ 1–2K) must be considered if one wishes to use MLS RHi to study supersat- of MLS temperature precision ( T uration. In simulation studies, systematic errors (such as tangent pressure retrieval and errors), in addition GPH S to introducing biases, also increase variability in dierences with respect to a “truth” data set particularly T for pressures greater than 200hPa. ¿is will add to the frequency of supersaturation in the tail of MLS RHi H S distribution functions. ¿erefore, MLS RHI is not recommended for studying statistics of supersaturation at 2 O T pressures greater than 178hPa. For smaller pressures, one must remove the contribution from temperature HCl S noise as part of the analysis. Measurements taken in the presence of clouds signicantly degrade the preci- sion, that is increases the scatter about the mean, but the mean bias as compared to AIRS changes by less than T HCN S 10%. ¿e last four temperature proles of each day show increased rates of outliers relative to GEOS-5 tem- T HNO perature, compared to other prole positions in the day. ¿ese outliers can be as large as 50K at some levels. S Attempts to address this end of day boundary problem by creating daily overlap periods that can be discarded 3 T have not yet satisfactorily addressed this problem and the magnitude of the outliers has increased somewhat HO S . A recommendation to discard the last 4 proles of each day for temperature, GPH, and RHI has in v4.2 x 2 T been added to recommended screening. HOCl S Also, users should note the possibility of a potential dri in the MLS lower stratospheric water vapor product, discussed in section 3.9.8. Such a dri will also impact the RHi product. T IWC S 3.20.8 Review of comparisons with other datasets T IWP Figures 3.20.2 and 3.20.3 show comparisons between AIRS v6 and MLS v2, 3, and 4. Mapped features tend to S agree well but MLS produces higher relative humidities in the moist regions of the tropics relative to AIRS. T Version 4 at 300hPa in high northern latitudes during January and February show marked improvement over N S 2 O the earlier versions by comparison to AIRS. Except as noted previously, the three MLS versions show little T change amongst themselves. S O 3 T 3.20.9 Desired improvements OH S We are aiming to improve high latitude performance of 261 and 215hPa H O in future work. 2 T RHI S T SO S 2 T S T Quality x 139 Level 2 Version 4.2 Aura Microwave Limb Sounder (MLS) T

146 Help 3.20. Relative Humidity with respect to Ice (RHI) Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T ClO S T S CO T GPH S T H S 2 O T HCl S T HCN S T HNO S 3 T HO S 2 T HOCl S T IWC S T IWP S T N S 2 O T S O 3 T OH S T Figure 3.20.2: Mapped elds from AIRS v6 (left), MLS v2.2 (center-left), MLS v3.3 (center-right), and MLS v4.2 (right) RHI S pressures between 300–150 hPa for January and February 2005. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 140 Level 2 Version 4.2 Quality T

147 Help 3.20. Relative Humidity with respect to Ice (RHI) Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T ClO S T S CO T GPH S T H S 2 O T HCl S T HCN S T HNO S 3 T HO S 2 T HOCl S T IWC S T IWP S T N S 2 O T S O 3 T OH S T Figure 3.20.3: Mapped elds from AIRS v6 (left), MLS v2.2 (center-left), MLS v3.3 (center-right), and MLS v4.2 (right) RHI S pressures between 300–150 hPa for June–August 2005. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 141 Level 2 Version 4.2 Quality T

148 Help 3.20. Relative Humidity with respect to Ice (RHI) Overview Table BrO S T CH S 3 Cl Summary of MLS v3.3 UTLS RHi product. Table 3.20.1: T CH S Single prole Pressure / Resolution b 3 Comments Accuracy / % CN a precision / % hPa H km × V T CH — — — Unsuitable for scientic use 0.001 S 3 × 230 190 100 0.002 13 OH × 13 0.004 100 100 260 T ClO S 12 × 590 50 100 0.010 100 40 × 12 0.022 750 T 30 400 × 16 0.046 100 S CO 100 0.10 14 × 420 30 T 90 20 370 × 8 0.22 GPH S 8 × 320 15 75 0.46 T × 15 8 280 60 1.00 H S 2 O × 35 250 15 8 2.15 T 220 15 × 6 4.64 15 HCl S × 210 15 15 10 4 T 22 × 210 15 20 4 HCN S 210 25 46 15 4 × T 68 4 × 200 15 25 HNO S 20 200 × 4 83 25 3 100 × 200 20 20 4 T 121 × 200 25 20 4 HO S 2 147 4 × 200 25 20 T 35 × 178 4 200 30 HOCl S 215 4 × 200 45 35 see Table 3.9.1 T 261 see Table 3.9.1 30 45 200 × 4 IWC S 316 × 200 70 20 see Table 3.9.1 6 T 10(est) 40(est) 316 6 > 150 UTRHI, measurement height depends × IWP S on atmospheric humidity T a N Absolute error in percent S 2 b O Fractional error ( [error in RHi] / RHi) in percent T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 142 Level 2 Version 4.2 Quality T

149 Help Overview 3.21 Sulfur Dioxide (SO ) 2 Table Swath name: SO2 215 – 10.0 hPa Useful range: BrO S < [email protected] > Contact: William Read, Email: T CH S 3 Cl 3.21.1 Introduction T CH product is taken from the 240-GHz retrieval ( ). MLS can only measure signi- CorePlusR3 ¿e standard SO S 2 3 CN cantly enhanced concentrations above nominal background such as that from volcanic injections. Validation Pumphrey et al. is published in [2015]. of SO T 2 CH S 3 OH 3.21.2 Changes from v3 T ¿e v4 SO retrieval has two signicant changes from v3. ¿e rst is a dierent method of modeling the 2 ClO S background radiance signal from unknown or unaccountable sources. ¿e method is specically designed T to be more immune to clouds. ¿e other change is an improved cloud detection and radiance rejection algo- S CO rithm. ¿ese changes have signicantly improved the convergence and quality performance of the retrieval system and eliminated much of the poor in-cloud behavior present in v3. ¿ere have been no changes to the T GPH S . spectroscopic characteriztion of the major species in the 240 GHz bands used to retrieve SO 2 , which is far Figure 3.21.1 compares MLS v4.2 and v3.3. ¿e atmosphere typically has ~ 0.1ppbv SO 2 T H smaller than the MLS SO accuracy estimate. Both versions (erroneously) report typical abundance of order S 2 2 O of a few ppbv, due to systematic errors in the MLS measurement system. ¿e − ppbv reported at 215hPa is 20 T also a systematic artifact. HCl S T 3.21.3 Resolution HCN S is ~ Based on Figure 3.21.2, the vertical resolution for SO 3km and the horizontal resolution is 170km. ¿e 2 T horizontal resolution perpendicular to the orbit track is 6km, the full width at half maximum of the MLS HNO S antenna, for all pressures. 3 T HO S 3.21.4 Precision 2 T ~ 4ppbv for all heights between 215–10hPa. ¿e precisions are set to ¿e estimated precision for SO is 2 HOCl S negative values (or zero in some cases) in situations when the retrieved precision is larger than 50% of the a priori precision – an indication that the data is biased toward the a priori value. T IWC S 3.21.5 Accuracy T Accuracy is estimated to be 10–15ppbv for pressures less than 147hPa increasing to ~ 20ppbv at 215hPa. IWP S ¿ese may be dierent for the v4 SO product. 2 T N S 2 O 3.21.6 Data screening T Pressure range: 215–10.0hPa S O 3 Values outside this range are not recommended for scientic use. T Estimated Precision: Values with negative precision can be used, though with caution. OH S is an Although it is generally recommended to not use values where precision is agged negative, SO T 2 RHI S exception and it is appropriate to use values with negatively agged precision (provided that the entire at the larger pressures (e.g., 215 and 147hPa) will prole is not so agged). High retrieved values of SO 2 T also have larger (i.e., poorer) precision values which are sometimes large enough to trigger the “too SO S 2 T S T Quality x 143 Level 2 Version 4.2 Aura Microwave Limb Sounder (MLS) T

150 Help 3.21. Sulfur Dioxide (SO ) 2 Overview Table Differences Profile Precision 180W--180E, 60N--90N 14 BrO S v03.30 v02.21 31 T v04.20 CH S 68 3 Cl 146 Pressure (hPa) T CH S 316 3 0.0 10 -25 -50 -10 -5 25 10 20 30 40 0 0 5 Profile Precision Differences CN SO2 (ppbv) (ppbv) SO2 (ppbv) T 180W--180E, 30N--60N 14 v03.30 CH S v02.21 3 31 OH v04.20 68 T ClO S 146 Pressure (hPa) T 316 S CO -10 0 5 25 0 -25 0.0 10 20 30 40 -5 10 -50 Differences Profile Precision SO2 (ppbv) (ppbv) SO2 (ppbv) T 180W--180E, 30S--30N 14 GPH v03.30 S v02.21 31 v04.20 T 68 H S 2 O 146 Pressure (hPa) T HCl S 316 0 5 10 -25 0 25 30 40 -50 10 -10 20 0.0 -5 Precision Profile Differences T SO2 (ppbv) (ppbv) SO2 (ppbv) HCN 180W--180E, 60S--30S S 14 v03.30 v02.21 T 31 v04.20 HNO S 68 3 T 146 Pressure (hPa) HO S 316 2 T 5 10 -5 -50 -25 0 0.0 10 20 30 40 25 -10 0 Differences Profile Precision HOCl (ppbv) SO2 (ppbv) SO2 (ppbv) S 180W--180E, 90S--60S 14 v03.30 T v02.21 31 IWC S v04.20 68 T IWP S 146 Pressure (hPa) T 316 -10 10 0.0 25 40 30 20 10 0 -25 -50 5 0 -5 N S 2 SO2 (ppbv) SO2 (ppbv) (ppbv) O T S O 3 Figure 3.21.1: A comparison of v3.3 (blue) and v4.2 (red) SO for Jan-Feb-Mar 2005 in 5 latitude bands. Other time 2 T periods are similar. The left panel compares mean proles, the center shows the mean dierence (red diamonds) OH S surrounded by each versions’ estimated precision, and the right panel shows the estimated retrieval precision (solid T and bullets) and measured variability (dotted) which includes atmospheric variability about the mean prole. RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 144 Level 2 Version 4.2 Quality T

151 Help 3.21. Sulfur Dioxide (SO ) 2 Overview Table Equator FWHM / km FWHM / km 600 -2 0 2 4 6 8 10 12 1200 1000 800 0 200 400 1 BrO S T CH S 3 Cl 10 T CH S 3 CN T Pressure / hPa CH S 100 3 OH T ClO S T 1000 0 0.6 2 4 0.4 0.8 0.2 1.0 1.2 0.0 -4 -2 -0.2 S CO Kernel, Integrated kernel Profile number 0 70 N T FWHM / km FWHM / km GPH 400 600 800 -2 1200 0 2 4 6 8 10 12 1000 0 200 S 1 T H S 2 O T HCl S 10 T HCN S T Pressure / hPa HNO 100 S 3 T HO S 2 1000 T HOCl 0.4 1.0 0.8 0.6 4 0.2 0.0 -0.2 2 0 -2 -4 1.2 S Kernel, Integrated kernel Profile number T IWC S Typical two-dimensional (vertical and horizontal along-track) averaging kernels for the MLS v4.2x SO Figure3.21.2: 2 T ◦ N (lower); variation in the averaging kernels is suciently small that these data at the equator (upper) and at 70 IWP S are representative of typical proles. Colored lines show the averaging kernels as a function of MLS retrieval level, T indicating the region of the atmosphere from which information is contributing to the measurements on the indi- N S vidual retrieval surfaces, which are denoted by plus signs in corresponding colors. The dashed black line indicates 2 O the resolution, determined from the full width at half maximum (FWHM) of the averaging kernels, approximately T scaled into kilometers (top axes). (Left) Vertical averaging kernels (integrated in the horizontal dimension for ve S O 3 along-track proles) and resolution. The solid black line shows the integrated area under each kernel (horizontally T and vertically); values near unity imply that the majority of information for that MLS data point has come from the OH S measurements, whereas lower values imply substantial contributions from a priori information. (Right) Horizon- T tal averaging kernels (integrated in the vertical dimension) and resolution. The horizontal averaging kernels are RHI S shown scaled such that a unit averaging kernel amplitude is equivalent to a factor of 10 change in pressure. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 145 Level 2 Version 4.2 Quality T

152 Help 3.21. Sulfur Dioxide (SO ) 2 Overview pecentage of good data screened by quality 100 Table 80 BrO S 60 T CH S 40 3 Cl Data Percentage T CH S 20 3 CN T 0 CH S 1.5 2.0 0.0 0.5 1.0 3 OH Quality Value T ClO S Figure 3.21.3: Percentage of data having a quality greater than that indicated on the x-axis for v4.2 data processed T for June–August 2005. S CO T GPH S inuence, and a priori inuence” negative-precision ag. In these cases, there will be an a priori much the retrieved value will probably be smaller than reality because the retrieval is being pulled towards T H S a priori the value of zero ppbv. Nonetheless, this does not detract from the fact that greatly enhanced 2 O is being observed, reecting the detection of a plume. Proles where the precision is set to zero, SO 2 T however, should be omitted from analyses. HCl S Status eld is an even number. Status ag: Only use proles for which the T HCN S Status Odd values of indicate that the prole should not be used in scientic studies. See Section 1.6 T for more information on the interpretation of the eld. Status HNO S Quality eld: Only proles having Quality greater than 0.95 should be used. 3 T As with water vapor, this value represents where the quality versus yield sharply drops indicating the HO S transition from well t proles toward more poorly t ones as shown in gure 3.21.3 2 T Convergence eld: Only proles having Convergence less than 1.03 should be used. HOCl S 3.21.7 Artifacts T IWC S accompanied with negative errors are likely to be underestimated due to a priori inuence High values of SO 2 ~ ( which biases the value toward zero. ¿e atmospheric background SO 0.1ppbv) cannot be measured by 2 T IWP MLS even with extensive data averaging because there are systematic errors producing biases of a few ppbv. S 20 has a background bias of − ppbv. ¿e 215-hPa SO ∼ 2 T N S 2 O 3.21.8 Review of comparisons with other datasets T from 23 eruptions since launch. ¿ese include Manam (Papua, New MLS has successfully detected SO 2 S O 3 Guinea – 3 events), Anatahan (Mariana Islands), Sierra Negra (Galapogos Island), Soufriere Hills (Montser- T rat, West Indies – 2 events), Tunguraua (Ecuador), Rabaul (Papua New Guinea), Piton de la Fournaise (Re- OH S union Island), Jebel al-Tair (Yemen), Okmok (Alaska), Kasatochi (Alaska), Dalalla (Ethiopia), Redoubt T (Alaska), Sarychev (Kuril Islands, Russia), Pacaya (Guatamala), and Merapi (Indonesia), Grimsvotn (ice- RHI S land), Puyehue-Cordon Caulle, (Chile), Nabro (Eritrea), Paluweh (Indonesia), Sangeang Api (Indonesia). T measured by OMI and the same calculated by Figure 3.21.4 shows an overlay comparison of column SO 2 SO S MLS for two days following the Sarychev eruption. It is clear that MLS detects the main plume dispersal fea- 2 T S T Aura Microwave Limb Sounder (MLS) 146 Level 2 Version 4.2 x Quality T

153 Help 3.21. Sulfur Dioxide (SO ) 2 Overview Table MLS SO2 OMI SO2 column map of sarychev 2009.06.14 50 BrO S 100 T CH S 150 3 Pressure (hPa) Cl T 200 CH S 250 300 200 150 100 50 0 3 ppbv CN T MLS HCl CH S 50 3 OH T 100 ClO S 150 T Pressure (hPa) S CO 200 13.0 4.0 0.0 1.0 2.0 5.0 6.0 14.0 12.0 11.0 10.0 9.0 8.0 7.0 3.0 Dobson Units T 4 6 8 -2 0 2 ppbv GPH S T H S MLS SO2 OMI SO2 column map of sarychev 2009.06.16 2 O 50 T HCl S 100 T HCN S 150 Pressure (hPa) T HNO S 200 100 150 250 300 0 50 200 3 ppbv T HO S MLS HCl 2 50 T HOCl S 100 T IWC S 150 Pressure (hPa) T 200 IWP 21.4 17.9 14.3 10.7 7.1 3.6 0.0 39.3 42.9 46.4 28.6 32.1 35.7 25.0 50.0 S Dobson Units -2 0 2 4 6 8 ppbv T N S 2 O T measurement on 14 and 16 June 2009 (separate Figure3.21.4: An overlay of MLS measurement tracks on an OMI SO S O 2 3 maps) showing the dispersial of SO from the Sarychev eruption (13 June 2009, black triangle). The color scale 2 T column measured by OMI. The daytime MLS tracks are small open circles and the nighttime indicates the SO 2 OH S tracks are lled black. When the calculated column from MLS exceeds 1 DU, that measurement is indicated by a T larger open circle lled with the color of the column measurement as indicated by the color scale below (same as RHI S for OMI). The panels at right show all the measured proles covering the area shown in the maps for SO and HCl. 2 Proles where the MLS column calculation exceeds 1 DU are highlighted in red. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 147 Level 2 Version 4.2 Quality T

154 Help 3.21. Sulfur Dioxide (SO ) 2 Overview 68.13 hPa 100 80 Table 60 40 20 0 -20 -40 BrO S 100.00 hPa 600 T 500 400 CH S 300 200 3 Cl 100 0 T 146.78 hPa CH S 300 250 3 CN 200 SO2 ppbv 150 100 T 50 CH 0 S -50 3 215.44 hPa OH 150 100 T 50 ClO S 0 -50 T S 2145 2120 CO data record T GPH S Figure 3.21.5: Time series overlay where MLS measures more than 500 ppbv at 100 hPa for the Sarychev erruption. T H S Blue is version 3, and red is version 4. 2 O T HCl S tures. It also appears that MLS columns are o en smaller than those from OMI. Interpreting the signicance of this is not straightforward given that OMI has to make assumptions regarding the prole shape and can T HCN S down to the boundary layer. ¿e MLS column begins at 215hPa and integrates upward neglect- observe SO 2 ing the tropospheric contributions. Another limitation is that OMI can only make measurements during the T HNO S day whereas MLS can make them day and night. Since the plume is moving relatively quickly over the 12- hour measurement separation time, MLS nighttime measurements o en miss and/or detect plume features 3 T dierently than OMI. HO S measured Figure 3.21.5 shows a comparison between v4 and v3 for a track showing the maximum SO 2 2 T on 16 June 2009. Although both version detect large amounts of SO at 100hPa, v4 shows this being more 2 HOCl S concentrated in the 100-hPa layer than v3, as the newer version shows less SO at 147hPa and none at 215hPa 2 at 68hPa. compared to v3. Neither version shows any SO T 2 IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 148 Level 2 Version 4.2 Quality T

155 Help 3.21. Sulfur Dioxide (SO ) 2 Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T ClO S T S product. Summary of MLS v3.3 SO Table 3.21.1: CO 2 Accuracy / Single prole T Resolution Comments Pressure / hPa GPH S ppbv precision ppbv × H km V T — Unsuitable for scientic use — — 10 < H S 2 16 4.3 165 3.1 × 10 O T 12 × 171 4.3 15 3.1 HCl S 3.9 171 × 3.1 22 17 14 32 3.0 × 176 3.4 T HCN S 46 2.9 × 12 3.2 176 10 68 3.2 183 × 2.9 T HNO 100 2.8 × 180 3.1 10 S 3.2 180 × 2.8 147 10 3 T 215 20 3.8 186 × 3.5 HO S — Unsuitable for scientic use 215 — — > 2 T HOCl S T IWC S T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 149 Level 2 Version 4.2 Quality T

156 Help Overview 3.22 Temperature (T) Table Swath name: Temperature Useful range: 261 – 0.001 hPa BrO S [email protected] Email: Michael J. Schwartz, Contact: > < T CH S 3 Cl 3.22.1 Introduction T CH x ¿e MLS v4.2 temperature product is similar to both the v3.3 product and to the v2.2 product that is de- S 3 Schwartz et al. [2008]. MLS temperature is retrieved primarily from bands near O spectral lines scribed in CN 2 at 118GHz and 239GHz that are measured with MLS radiometers R1A/B and R3, respectively. ¿e isotopic T CH S 239-GHz line is the primary source of temperature information in the troposphere, while the 118-GHz line is 3 OH x 1 ~ − K bias the primary source of temperature in the stratosphere and above. MLS v4.2 temperature has a T with respect to correlative measurements in the troposphere and stratosphere, with 2–3K peak-to-peak ad- ClO S ditional systematic vertical structure. Table 3.22.1 summarizes the measurement precision, resolution, and T modeled and observed biases. ¿e following sections provide details. S CO x /v3.4 and v3.3 x 3.22.2 Dierences between v4.2 x T GPH S /v3.4 temperature and v3.3 x temperature are typically less than 0.5K and ex- x Mean dierences between v4.2 x ceed 1K only in the tropical troposphere and, in some instances, at the highest retrieval levels, 0.01–0.001hPa T H and v3.4 x reaches 2K, and x relative to v3.3 (see Figure 3.22.1). In the tropics, a cold bias in unscreened v4.2 x S 2 O is reduced to 1.5K by the recommended screening, discussed below. T FinalPtan , a preliminary “phase” Versions v3.x and v2.x take the standard temperature product from HCl S bands and that focuses on retrieval of temperature and tangent- of the MLS retrieval that uses only the O 2 T temperature product is taken from the phase, which adds x CorePlusR3 point pressure. ¿e standard v4.2 HCN S radiances from the 240-GHz radiometer and which simultaneously retrieves standard products for a number T temperature and GPH along with the accompanying tangent- of atmospheric constituents. ¿e CorePlusR3 HNO S point pressures, are used for all subsequent constituent retrievals, so they are more internally consistent with 3 the suite of constituent retrievals than were the standard temperature and GPH products of previous versions. T HO S MLS is generally insensitive to the presence of thin clouds, but scattering in the cores of convective storms 2 can produce signicant perturbations in MLS retrieved quantities. Temperature is particularly susceptible to T HOCl S “signicant” perturbations, because knowledge from analysis is generally good enough in much of the a priori lower and middle atmosphere that perturbations of a few percent (a level negligible for trace-gas constituent T knowledge may a priori retrievals) are considered “signicant” for temperature. In the troposphere where IWC S be on the order of 1–2K, large dierences between MLS and GEOS-5 are believed to typically result from T problems in the MLS measurements. Plotting dierences between the MLS retrieved temperatures and the IWP S a priori in the UT and stratosphere) is o en useful in identication of GEOS-5 temperatures (the retrieval T retrieval artifacts. Cloud-induced perturbations generally occur in the tropics and mid-latitudes (where con- N S 2 vective storms occur) in both the v3 and v4 retrieval versions, but perturbations of retrieved temperature and O T the recommended screening for their removal have changed between v3.3 x x and v4.2 x . In unscreened /v3.4 S O x and negative pertur- data at retrieval levels 316–178hPa, clouds produce primarily positive outliers in v4.2 3 T bationsinv03.30. ¿etoppanelsofFigure3.22.2showthisbehaviorat215hPa. Conversely, at121hPa, outliers OH S are primarily negative in v03.x and positive in v4.2 x . ○ ○ Unscreened and screened histograms of tropical (20 N) v03.30 and v04.20 temperature from 2005 S–20 T RHI S are shown in Figure 3.22.3. Tails of low, unscreened v04.20 outliers (dark red) are particularly evident from 316–147hPa,butaregreatlyreducedbyrecommendedscreening(magenta). Changesat316–215hPabetween T SO inmeantropicaltemperaturedierencesfromGEOS-5.9, givennumericallyintheupperle v03.30andv4.2 x S 2 T S T Aura Microwave Limb Sounder (MLS) x 150 Level 2 Version 4.2 Quality T

157 Help 3.22. Temperature (T) Overview Table BrO S T CH S 3 Cl ( Unscreened Temperature v04−20 ) minus ( Unscreened Temperature v03−30) T CH S ( Screened Temperature v04−20 ) minus ( Screened Temperature v03−30) 3 CN ° ° ° ° ° ° ° ° ° ° −20 to −50 20 to −20 lat −90 lat 50 −50 to 90 lat to 50 lat to 20 lat T 0.001 CH S 3 0.01 OH T 0.1 ClO S 1 T 10 S CO 100 T 1000 GPH S 1 2 2 −2 −1 0 1 2 −2 −1 0 0 1 0 −1 −2 −1 −2 1 2 −2 −1 0 1 2 VMR VMR VMR VMR VMR T scatter (solid) scatter (solid) scatter (solid) scatter (solid) scatter (solid) H precision (dashed) precision (dashed) precision (dashed) precision (dashed) precision (dashed) S 2 0.001 O T 0.01 HCl S 0.1 T HCN 1 S 10 T HNO S 100 3 1000 T 4 1 2 3 4 5 3 3 5 3 1 0 2 4 5 5 4 3 2 4 1 0 2 1 2 0 0 5 1 0 HO VMR VMR VMR VMR VMR S 0.001 2 0.01 0.1 T 1 HOCl S 10 100 1000 T 2 2 0 2 4 4 4 2 4 0 2 4 0 0 0 5 5 5 5 5 Good Points Good Points Good Points Good Points Good Points IWC S x 10 x 10 x 10 x 10 x 10 T Figure 3.22.1: Temperature dierences between v4.2x and v03.30 standard products, screened per recommenda- IWP S tions of this document (green) and unscreened (red). 2005 data are shown. Top row is mean dierence between versions, binned by latitude. Screening has a signicant impact on mean values only at low latitudes in the tro- T posphere. The middle panel show considerably more scatter in unscreened data at the bottom of the retrieval, N S 2 O particularly at low latitudes. Bottom panels’ dierence between red and green lines is the number of points that T are screened out S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 151 Level 2 Version 4.2 Quality T

158 Help 3.22. Temperature (T) Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH T ClO S T S CO T GPH S Figure 3.22.2: The top row shows histograms of unscreened 215-hPa v03.30 (left) and v04.20 (right) temperature ◦ T dierences from GEOS-5.9 temperature from 2005 data. The horizontal bin spacing is latitudinal spacing of 1.5 ~ H S 2 the repeating MLS scan pattern. The bottom panels show histograms of analogous screened dierences on a ner O 20 K have been almost completely eliminated and are not shown. ± vertical scale. Outliers beyond T HCl S T corners of the panels, can be seen to reect shi s in the peaks of the distributions as well as the contributions HCN S of the outlier tails. T a priori ¿e tropospheric and stratospheric are from GEOS-5.9, but are x temperature proles used in v4.2 HNO S quite similar to the GEOS-5.7 and GEOS-5.2 proles used in previous version, departing most signicantly 3 nearthestratopause. ¿ischangein apriori temperatureisbelievedtohaveonlyminimalimpactonretrievals. T HO S 3.22.3 Resolution 2 T ¿e vertical and horizontal resolution of the MLS temperature measurement is shown by averaging kernels HOCl S in Figure 3.22.4. Vertical resolution, shown on the le panel, is ~ 4.5km from 261hPa to 100hPa, improves to T 3.6km at 31.6hPa and then degrades to 4.3km at 10hPa, 5.5km at 3.16hPa, 6km at 0.01hPa, and 8–10m at IWC S ~ 165km from 261hPa to 0.1hPa and degrades to 280km at 0.001hPa. ¿e 0.01hPa. Along track resolution is cross-track resolution is set by the 6-km width of the MLS 240-GHz eld of view in the troposphere and by T IWP S the 12-km width of the MLS 118-GHz eld of view in the stratosphere and above. ¿e longitudinal separation ○ ○ over middle and of MLS measurements from a given day, which is determined by the Aura orbit, is 10 –20 T N low latitudes, and ner in polar regions. S 2 O T 3.22.4 Precision S O 3 temperature measurement is summarized in Table 3.22.1. Precision is the ¿e precision of the MLS v4.2 x T random component of measurements which would average down if the measurement were repeated. ¿e re- OH S trieval so ware returns an estimate of precision based upon the propagation of radiometric noise and a priori T uncertainties through the measurement system. ¿ese values, which range from 0.5K in the lower strato- RHI S sphere to 2.5K in the mesosphere, are given, for selected levels, in column 2. Column 3 gives the rms of ~ T dierences of values from successive orbits (divided by the square-root of two as we are looking at the dier- SO S ence of two noisy signals) for latitudes and seasons where longitudinal variability is small and/or is a function 2 T S T Aura Microwave Limb Sounder (MLS) x 152 Level 2 Version 4.2 Quality T

159 Help 3.22. Temperature (T) Overview Table BrO S T CH S 3 Cl T 4 68.1hPa 215hPa 121hPa CH S 10 −0.3 ± −2.0 −1.9 1.6 K (100%) 1.3 K (100%) ± 2.3 K (100%) ± 3 CN ± 1.2 K (83%) 1.7 K (83%) ± −1.7 −2.0 1.3 K (98%) ± −0.3 3 10 ± 1.1 K (100%) ± −1.9 ± −0.5 −3.1 3.1 K (100%) 1.5 K (100%) T −2.3 ± −0.7 1.6 K (82%) 1.2 K (83%) 1.1 K (95%) ± −1.9 ± 2 CH 10 S 3 OH 1 10 T 0 10 ClO S Observations per ppbv T 4 147hPa 82.5hPa 261hPa 10 S CO −0.5 −1.8 ± ± 1.7 K (100%) 1.5 K (100%) −1.1 2.8 K (100%) ± 1.4 K (83%) 1.8 K (83%) ± −1.7 ± −1.0 ± 1.5 K (98%) 0.0 3 10 T −1.7 1.2 K (100%) 3.8 K (100%) ± ± 1.7 K (100%) −2.5 ± −1.2 ± −1.2 ± 1.5 K (83%) 1.2 K (95%) ± −1.7 −1.5 1.9 K (82%) GPH S 2 10 T 1 10 H S 2 O 0 10 T Observations per ppbv HCl S 4 100hPa 178hPa 316hPa 10 T −0.6 ± 1.4 K (100%) −2.0 ± 2.0 K (100%) 4.5 K (100%) ± 1.9 HCN 1.7 K (83%) 2.7 ± 3.1 K (83%) 1.3 K (98%) ± ± −1.8 −0.6 S 3 10 −0.8 1.4 K (100%) −0.5 ± 2.5 K (100%) ± 5.5 K (100%) −2.4 ± ± −2.0 −1.0 ± 1.1 K (83%) 0.5 ± 1.8 K (83%) 3.8 K (82%) T 2 10 HNO S 1 10 3 T 0 10 HO S Observations per ppbv 2 T −15 −20 −25 15 0 −5 10 −20 −15 −10 −5 0 5 10 15 5 −40 −30 −10 20 10 0 −10 −20 HOCl S T − T (K) (K) (K) T − T T − T GEOS−5 GEOS−5 GEOS−5 screened v03.30 screened v04.20 unscreened v03.30 unscreened v04.20 T IWC S ◦ ◦ N) dierences between MLS retrieved temperatures S–20 Panels show histograms of tropical (20 Figure 3.22.3: T and GEOS-5.9 temperature for all of 2005. Blue and green lines are unscreened and screened v03.30, respectively. IWP S Red and magenta are unscreened and screened v04.20, respectively. Distribution means and standard deviations T are given in the upper left of each panel along with the percentage of points included after screening. The 316-hPa N S level is shown for reference although this level is not recommended to be used for scientic work. 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 153 Level 2 Version 4.2 Quality T

160 Help 3.22. Temperature (T) Overview Table Equator FWHM / km FWHM / km 800 600 400 200 0 1000 1200 12 10 -2 0 2 4 6 8 0.001 BrO S 0.010 T CH S 3 Cl 0.100 T CH S 1.000 3 CN T Pressure / hPa CH 10.000 S 3 OH 100.000 T ClO S 1000.000 T 0.8 1.0 1.2 0.2 -4 -0.2 0.0 -2 0.4 0 0.6 2 4 S CO Profile number Kernel, Integrated kernel 0 N 70 FWHM / km FWHM / km T 12 200 0 -2 400 600 800 1000 1200 0 2 4 6 8 10 GPH 0.001 S T 0.010 H S 2 O T 0.100 HCl S T 1.000 HCN S Pressure / hPa T 10.000 HNO S 3 100.000 T HO S 2 1000.000 T 1.2 0 -2 0.6 0.4 0.2 0.0 -0.2 -4 0.8 4 2 1.0 HOCl S Profile number Kernel, Integrated kernel T IWC S Typical two-dimensional (vertical and horizontal along-track) averaging kernels for the MLS v4.2x Figure 3.22.4: ◦ T Temperature data at the equator (upper) and at 70 N (lower); variation in the averaging kernels is suciently small IWP S that these are representative of typical proles. Colored lines show the averaging kernels as a function of MLS re- trieval level, indicating the region of the atmosphere from which information is contributing to the measurements T on the individual retrieval surfaces, which are denoted by plus signs in corresponding colors. The dashed black N S 2 O line indicates the resolution, determined from the full width at half maximum (FWHM) of the averaging kernels, T approximately scaled into kilometers (top axes). (Left) Vertical averaging kernels (integrated in the horizontal di- S O mension for ve along-track proles) and resolution. The solid black line shows the integrated area under each 3 kernel (horizontally and vertically); values near unity imply that the majority of information for that MLS data point T OH has come from the measurements, whereas lower values imply substantial contributions from a priori information. S (Right) Horizontal averaging kernels (integrated in the vertical dimension) and resolution. The horizontal averag- T ing kernels are shown scaled such that a unit averaging kernel amplitude is equivalent to a factor of 10 change in RHI S pressure. T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 154 Level 2 Version 4.2 Quality T

161 Help 3.22. Temperature (T) Overview only of local solar time. ¿e smallest values found, which are in the tropics in the troposphere and in high- latitude summer in the stratosphere and mesosphere, are taken to be those least impacted by atmospheric Table variability and are what is reported in column 3. ¿ese values are 0.5–0.8K larger than those estimated by the ~ measurement system in the troposphere and lower stratosphere, and a factor of 1.4 larger from the middle stratosphere through the mesosphere. BrO S T 3.22.5 Accuracy CH S x measurements, similar to that A substantial study has been made of sources of systematic error in MLS v4.2 3 Cl which was done as a part of the validation of v2.2. ¿e accuracy of the v4.2 temperature measurements is x T CH S estimated both by modeling the impact of uncertainties in measurement and retrieval parameters that could 3 CN lead to systematic errors, and through comparisons with correlative data sets. Column 5 of Table 3.22.1 gives T estimates from the propagation of parameter uncertainties, as discussed in Schwartz et al. [2008]. ¿is es- CH S timate is broken into two pieces. ¿e rst term was modeled as amplier non-linearity, referred to as “gain 3 OH compression,” and was believed to have a known sign, as gain is known to drop at high background signal lev- T els. Correction of these nonlinearities was a goal of v4.2 , but closer examination of the simple non-linearity x ClO S model found that it did not close forward model and measured radiances as expected. It had been hoped T that better radiance closure would permit the use of more radiances in the middle of the 118-GHz O band, 2 S CO giving better resolution, precision and accuracy in the upper stratosphere and better accuracy everywhere. T ¿is work is still ongoing, and it is hoped that advances will manifest in improvements in a future version. GPH S σ estimates of other sources of systematic uncertainty, such as 2 ¿e second term of column 5 combines spectroscopic parameters, retrieval numerics and pointing, for which the sign of resulting bias is unknown. T H S K to Gain compression terms range from − 1 . 4 + 5 5 K, and predicted vertical structure is similar to observed . 2 O biases relative to correlative data in the troposphere and lower stratosphere. ¿e terms of unknown sign are T HCl S ~ 2K magnitude over most of the retrieval range, increasing to 5K at 261hPa and to 3K at 0.001hPa. of Column 6 contains estimates of bias based upon comparisons with analyses and with other previously- T HCN validated satellite-based measurements. In the troposphere and lower stratosphere, the observed biases be- S tween MLS and most correlative data sets are consistent to within 1.5K, and have vertical oscillation with ~ T an amplitude of 2–3K and a vertical frequency of about 1.5 cycles per decade of pressure. A global average HNO S of correlative measurements is shown in Figure 3.22.5. 3 T HO S 3.22.6 Data screening 2 Pressure range: 261–0.001hPa T HOCl S Values outside this range are not recommended for scientic use. T Estimated precision: Only use values for which the estimated precision is a positive number. IWC S a priori Values where the information has a strong inuence are agged with negative or zero precision, T and should not be used in scientic analyses (see Section 1.5). IWP S eld is an even number. Status Status ag: Only use proles for which the T indicate that the prole should not be used in scientic studies. See Section 1.6 Status Odd values of N S 2 O for more information on the interpretation of the eld. Status T Quality Quality: Only use proles with greater than 0.2 for the 83hPa level and smaller pressures, and S O 3 Quality proles with greater than 0.9 at larger pressures of 100hPa and larger. T OH S Quality less than or equal to 0.9 comprise 1.4% of all data, and 4.8% of proles in the Proles with tropics. T RHI S eld is less than 1.03 should be used. Convergence Convergence: Only proles whose T Convergence criterion rejects < 0 . 1% of proles and chunks with convergence slightly over this ¿is SO S target do not contain manifestly pathological proles. ¿e primary purpose of this criterion is to reject 2 T S T Aura Microwave Limb Sounder (MLS) x 155 Level 2 Version 4.2 Quality T

162 Help 3.22. Temperature (T) Overview proles with extremely poor convergence that may be expected to reect poor retrieval behavior. ¿e “low cloud” bit of the temperature eld does not contain useful information Cloud Screening: Status Table in v4.2 x , so it cannot be used to screen temperature as it was in previous versions. However, cloud impacts can largely be removed using the ice water content (IWC) product, rejecting proles between 3 BrO S . Implementation of this 261–100hPa for which the 215hPa value of IWC is greater than 0.005g/m IWC swath, but is responsible for most of the rejection of x criteria requires the loading of the v4.2 T CH S cloud-induced outliers seen in Figure 3.22.3. ¿is criteria removes 4.7% of UT proles globally and 3 ○ ○ Cl S). N–20 15% of proles in the tropics (20 T CH ¿e eld can be used to further eliminate outliers that are believed to be the L2gpPrecision Precision: S 3 CN result of thick clouds, primarily in the tropics. If careful screening of the troposphere is required, levels T 261–178hPa should be avoided if any of the following criteria are met: CH S ○ 3 L2gpPrecision > 1 . K and latitude > − 60 At 316hPa: 1 OH T 0 > At 261hPa: . 7 K L2gpPrecision ClO S . 0 > L2gpPrecision At 215hPa: K 825 T End of day (v4.20 only): ¿e last four proles of each day show greatly increased rates of large departure S CO a priori from and should not be used. ¿is issue was xed in v4.22. T GPH S 3.22.7 Artifacts T MLS temperature has persistent, vertically oscillating biases with respect to analysis and correlative measure- H S 2 O ments in the troposphere and stratosphere that are an area of continued research. ¿e impact of clouds is T generally limited to tropospheric levels and to the tropics and, to a lesser extent, mid-latitudes. Cloud im- HCl S and v3.4 x and pacts in unscreened data at the lowest retrieved levels are larger in v4.2 x than they were in v3.3 x T are harder to screen due to the lack of a useful “low cloud” status bit. However, a er recommended screen- HCN S of greater than 10K (believed almost always to cloud-induced ing, tropical negative departures from a priori T artifacts) are seen in only 0.2% of tropical proles. Flagging of clouds is discussed above. Further discussion HNO S [2008]. Schwartz et al. of artifacts may be found in 3 ¿e last four proles of each day show increased rates of outliers relative to GEOS-5 temperature, com- T pared to other prole positions in the day in version 4.20 (the problem is xed in version 4.22). ¿ese outliers HO S 2 can be as large as 50K at at some levels. A recommendation to discard the last 4 proles of each day has been T added to recommended screening for v4.20 (not needed for v4.22 and later versions). HOCl S T 3.22.8 Review of comparisons with other datasets IWC S Schwartz et al. [2008] describes detailed comparisons of MLS v2.2 temperature with products from the God- T Rienecker et al. , 2007] (GEOS-5), the European Center for Medium- dard Earth Observing System, version 5 [ IWP S Range Weather Forecast [e.g., Simmons et al. , 2005] (ECMWF), the CHAllenging Minisatellite Payload T ,2001], thecombinedAtmosphericInfraredSounder/AdvancedMicrowaveSound- (CHAMP)[ Wickertetal. N S ing Unit (AIRS/AMSU), the Sounding of the Atmosphere using Broadband Radiometry (SABER) [ Mlynczak 2 O , 1995], the Halogen Occultation Experiment [ and Russell , 1996] (HALOE) and the Atmospheric Hervig et al. T S Chemistry Experiment [ Bernath et al. , 2004] (ACE), as well as to radiosondes from the global network. From O 3 ~ 10hPa there is generally agreement to ~ 1K between the assimilations (ECMWF and GEOS-5) 261hPa to T ~ and AIRS, radiosondes and CHAMP, with SABER and ACE having generally warm biases of 2K relative to OH S this group. Figure 3.22.5 shows the global mean biases in the le panel and the 1 scatter about the mean σ T in the right panel for these eight comparisons. Between 1hPa and 0.001hPa, MLS has biases with respect to RHI S 3K between 0.1K and 0.01K and increasing in 5K between 1hPa and 0.1hPa, of 0K to SABER of − 1K to + − T − magnitude to 10K at 0.001hPa. Estimates of systematic error in the MLS temperature are shown in black, SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 156 Level 2 Version 4.2 Quality T

163 Help 3.22. Temperature (T) Overview Table BrO S T CH S MLS−ECMWF MLS−ECMWF 3 MLS−GEOS5 MLS−GEOS5 Cl MLS−CHAMP MLS−CHAMP T CH MLS−AIRS MLS−AIRS S 3 MLS−SABER MLS−SABER CN MLS−HALOE MLS−HALOE T MLS−ACE MLS−ACE CH S MLS−SONDE MLS−SONDE 3 OH Modeled Bias Modeled Bias T 1 1 ClO S T S CO T 3.16 3.16 GPH S T H S 2 O T 10 10 HCl S T HCN S T Pressure (hPa) Pressure (hPa) HNO 31.6 31.6 S 3 T HO S 2 T 100 100 HOCl S T IWC S T 316 316 IWP S 5 6 0 4 2 −5 0 Std. Dev. [K] Mean [K] T N S 2 O The left panel shows globally-averaged mean dierences between MLS temperature and eight cor- Figure 3.22.5: T relative data sets. Criteria for coincidences are described in detain in Schwartz et al. [2008]. The right panel shows S O the global standard deviations about the means. 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 157 Level 2 Version 4.2 Quality T

164 Help 3.22. Temperature (T) Overview Mean v04.20 Temperature minus GEOS−5.9 Temperature 1 4 Table 3.16 2 10 BrO S 0 (K) 31.6 T CH S −2 100 3 Cl T 316 −4 80 60 40 −20 20 −80 −60 −40 0 CH S 3 CN Std Dev of v04.20 Temperature minus GEOS−5.9 Temperature T 1 8 CH S 3 OH 3.16 6 T 10 ClO S 4 (K) 31.6 T S CO 2 100 T 316 0 GPH S 0 40 20 −40 −20 80 −60 −80 60 Latitude T H S 2 O Zonal mean of the dierence between MLS v4.2x temperature and GEOS-5.9 temperature (upper), Figure 3.22.6: T and variability about that mean (lower), averaged for 2005. HCl S T σ 2 with uncertainty shown with gray shading. ¿e black line is the modeled contribution of “gain compres- HCN S sion,” which was hoped would explain much of the vertical structure of MLS biases in the upper troposphere T and lower stratosphere. As discussed above, the gain-compression model used in this study does not ade- HNO S quately close the retrieval’s radiance residuals, so further study is needed to understand the forward-model 3 T inadequacies. HO S temperature and GEOS- x ¿e upper panel of Figure 3.22.6 shows zonal-mean dierences between v4.2 2 σ 1 5.9 temperature, averaged over 2005; the lower panel is the variability about that mean. Persistent vertical T HOCl S structure in the troposphere and lower stratosphere is evident, with the oscillations somewhat stronger at the equator and poles than at mid-latitudes. In the upper stratosphere, MLS has a general warm bias relative to T GEOS-5 at mid and high latitude that increases to more than 10K in the poles at 1hPa. ¿e bias at 1hPa is IWC S much smaller in polar summer, but persists in polar winter. T IWP S T N S 2 O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 158 Level 2 Version 4.2 Quality T

165 Help 3.22. Temperature (T) Overview Table BrO S T CH S 3 Cl T CH S 3 CN T CH S 3 OH Comments Unsuitable for scientic use Unsuitable for scientic use T ClO S 1.5 4 1 1 5 − − + + + T 1 9 — — + − / K 1to0 2to0 2to0 2to0 2to0 8to0 S CO 7to − 0to 0to 0to − − − − − − 2.5to accuracy Observed − T GPH S T H S 1.4 1 1 1.2 0.7 0.9 0.8 2 1.2 2.4 1.2 0.8 0.5 0.5 O ± ± ± ± ± ± ± ± ± ± ± ± ± — — 2 2 T / K 3 2 1 1 2 + + 1.8 1.5 0.8 2.7 0.7 + + + − HCl + S 0.4 + accuracy Modeled − − + + T HCN S 290 260 T 165 H 165 165 165 167 165 165 165 165 170 170 × × HNO × × × × × × × Summary of MLS Temperature product × × × — — S × × / km V 8 6 4.1 5.5 3.7 3.9 3.6 4.7 4.2 6.8 4.2 3 Resolution 10–11 T HO S b 2 T 3 Table 3.22.1: 1.3 1.5 1.5 1.2 1.0 3.5 13–14 2.7 0.7 0.7 0.7 0.8 0.6 — — ± HOCl / K S ± ± ± ± ± ± ± ± ± ± ± ± Scatter Observed T IWC a S T 1.3 1.2 3.3 3.6 0.7 0.7 2.4 0.8 0.8 0.6 0.6 0.6 0.6 — — / K ± ± IWP ± ± ± ± ± ± ± ± ± ± ± S Precision T N S 2 O T S O 3 1hPa 10hPa 0.1hPa 215hPa 261hPa 100hPa 31.6hPa 0.001hPa 3.16hPa 14.7hPa 0.01hPa T 56.2hPa Pressure 0.316hPa 0.001hPa < OH S 1000–316hPa T RHI S Precision on individual proles Precision inferred from dierences of individual proles from successive orbits (v2.2 results shown) T b a SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 159 Level 2 Version 4.2 Quality T

166 Help Overview Table Bibliography Bernath, P. F., et al., Atmospheric chemistry experiment (ACE): mission overview, Proceedings of SPIE , 5542 , BrO S 146–156, 2004. T CH S Coeld, R. E., and P. C. Stek, EOS Microwave Limb Sounder GHz optics design and eld-of-view calibration, 3 Cl (5), 1166–1181, 2006. IEEE Trans. Geosci. Remote Sens. 44 , T CH S Craig, C., K. Stone, D. Cuddy, S. Lewicki, P. Veefkind, P. Leonard, A. Fleig, and P. Wagner, HDF-EOS Aura 3 CN Tech. rep. , National Center For Atmospheric Research, 2003. le format guidelines, T CH S Cuddy, D. T., M. Echeverri, P. A. Wagner, A. Hanzel, and R. A. Fuller, EOS MLS science data processing 3 OH 44 , system: A description of architecture and capabilities, IEEE Trans. Geosci. Remote Sens. (5), 1192–1198, T 2006. ClO S Filipiak, M. J., N. J. Livesey, and W. G. Read, Precision estimates for the geophysical parameters measured by T S CO EOS MLS, , University of Edinburgh, Department of Meteorology, 2004. Tech. rep. T Froidevaux, L., et al., Validation of Aura Microwave Limb Sounder HCl measurements, J. Geophys. Res. , GPH S (D15), D15S25, doi:10.1029/2007JD009025, 2008a. 113 T H S Froidevaux, L., et al., Validation of Aura Microwave Limb Sounder stratospheric and mesospheric ozone 2 O , D15S20, doi:10.1029/2007JD008771, 2008b. measurements, J. Geophys. Res. , 113 T HCl S Harrison,J.J.,andP.F.Bernath,ACE-FTSobservationsofacetonitrileinthelowerstratosphere, Atmos.Chem. T Phys. , 7405–7413, 2013. 13 , HCN S J. Hervig, M. E., et al., Validation of temperature measurements from the Halogen Occultation Experiment, T HNO S (10), 10277–10,286, 1996. 101 , Geophys. Res. 3 T Hubert, D., et al., Ground-based assessment of the bias and long-term stability of 14 limb and occultation HO S , 9 Atmos. Meas. Tech. ozone prole data records, (6), 2497–2534, doi:10.5194/amt-9-2497-2016, 2016. 2 T HOCl Hurst, D. F., W. G. Read, H. Vömel, H. B. Selkirk, K. H. Rosenlof, S. M. Davis, E. G. Hall, A. F. Jordan, and S. J. S Oltmans, Recent divergences in stratospheric water vapor measurements by frost point hygrometers and T 9 , Atmos. Meas. Tech. the Aura Microwave Limb Sounder, (9), 4447–4457, doi:10.5194/amt-9-4447-2016, IWC S 2016. T IWP S Jarnot, R. F., V. S. Perun, and M. J. Schwartz, Radiometric and spectral performance and calibration of the (5), 1131–1143, 2006. GHz bands of EOS MLS, , IEEE Trans. Geosci. Remote Sens. 44 T N S 2 O Jiang, Y. B., et al., Validation of the Aura Microwave Limb Sounder ozone by ozonesonde and lidar measure- T , D24S34, doi:10.1029/2007JD008776, 2007. ments, J. Geophys. Res. , 112 S O 3 at the Khosravi, M., et al., Diurnal variation of stratospheric and lower mesospheric HOCl, ClO and HO T 2 OH S equator: comparison of 1-D model calculations with measurements by satellite instruments, Atmos. Chem. (15), 7587–7606, doi:10.5194/acp-13-7587-2013, 2013. 13 , Phys. T RHI S Kleinböhl, A., G. C. Toon, B. Sen, J.-F. L. Blavier, D. K. Weisenstein, and P. O. Wennberg, Infrared measure- T Geophys. Res. Lett. cn, ments of atmospheric ch , 32 , L23807, doi:10.1029/2005GL024283, 2005. 3 SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 160 Level 2 Version 4.2 Quality T

167 Help Bibliography Overview J. Kovalenko, L. J., et al., Validation of Aura Microwave Limb Sounder BrO observations in the stratosphere, , D24S41, doi:10.1029/2007JD008817, 2007. 112 , Geophys. Res. Table Lambert, A., et al., Validation of the Aura Microwave Limb Sounder stratospheric water vapor and nitrous J. Geophys. Res. , 112 (D24), D24S36, doi:10.1029/2007JD008724, 2007. oxide measurements, BrO S , JetPropul- Livesey, N.J., andW.V.Snyder, EOSMLSretrievalprocessesalgorithmtheoreticalbasis, Tech.rep. T CH S , 2004. http://mls.jpl.nasa.gov sion Laboratory, D-16159, available on the MLS web site 3 Cl T Livesey, N. J., J. W. Waters, R. Khosravi, G. P. Brasseur, G. S. Tyndall, and W. G. Read, Stratospheric CH CN 3 CH S Geophys. Res. Lett. , 28 (5), 779–782, 2001. from the UARS Microwave Limb Sounder, 3 CN Livesey, N. J., W. V. Snyder, W. G. Read, and P. A. Wagner, Retrieval algorithms for the EOS Microwave Limb T CH S (5), 1144–1155, doi:10.1109/TGRS.2006.872327, 2006. IEEE Trans. Geosci. Remote Sens. Sounder (MLS), 44 , 3 OH Tech. rep. , Jet Propul- Livesey, N. J., et al., EOS MLS version 1.5 Level 2 data quality and description document, T ClO sion Laboratory, D-32381, 2005. S T Livesey, N. J., et al., Validation of Aura Microwave Limb Sounder O and CO observations in the upper 3 S CO troposphere and lower stratosphere, J. Geophys. Res. , 113 , D15S02, doi:10.1029/2007JD008805, 2008. T GPH S Livesey, N. J., et al., EOS MLS version 3.3 and 3.4 Level 2 data quality and description document, Tech. rep. , Jet Propulsion Laboratory, available from http://mls.jpl.nasa.gov/, 2013. T H S 2 O Mahieu, E., et al., Recent Northern Hemisphere stratospheric HCl increase due to atmospheric circulation T (7), 104–107, doi:10.1038/nature13857, 2014. 515 , Nature changes, HCl S Millán, L., N. Livesey, W. Read, L. Froidevaux, D. Kinnison, R. Harwood, I. A. Mackenzie, and M. P. Chip- T HCN pereld, New Aura Microwave Limb Sounder observations of BrO and implications for Bry, Atmos. Meas. S , Tech. (7), 1741–1751, doi:10.5194/amt-5-1741-2012, 2012. 5 T HNO S Millán, L., S. Wang, N. Livesey, D. Kinnison, H. Sagawa, and Y. Kasai, Stratospheric and mesospheric HO 2 3 , 15 (5), 2,889–2,902, doi:10.5194/ Atmos.Chem.Phys. observationsfromtheAuraMicrowaveLimbSounder, T HO S acp-15-2889-2015, 2015. 2 T NASA Mlynczak, M., and J. M. Russell, III, An overview of the SABER experiment for the TIMED mission, HOCl S , Langley Research Center, Optical Remote Sensing of the Atmosphere , 1995. 2 T IWC Nair, P. J., et al., Relative dri s and stability of satellite and ground-based stratospheric ozone proles at S (6), 1301–1318, doi:10.5194/amt-5-1301-2012, 2012. , Atmos. Meas. Tech. NDACC lidar stations, 5 T IWP S Pardo, J. R., E. Serabyn, and J. Cernicharo, Submillimeter atmospheric transmission measurements on mauna T J. Quant. kea during extremely dry el niño conditions: implications for broadband opacity contributions, N S 2 Spectrosc. Radiat. Transfer , 68 (4), 419–433, doi:10.1016/S0022-4073(00)00034-0, 2001. O T Pickett, H. M., Microwave Limb Sounder THz Module on Aura, 44 (5), IEEE Trans. Geosci. Remote Sens. , S O 3 1122–1130, 2006. T OH S Geophys. Pickett, H. M., W. G. Read, K. K. Lee, and Y. L. Yung, Observation of night OH in the mesosphere, T Res. Lett. , p. L19808, doi:10.1029/2006GL026910, 2006a. RHI S measurements with remote-sensing balloon instruments, Pickett, H. M., et al., Validation of Aura MLS HO x T 33 , (1), L01808, doi:10.1029/2005GL024442, 2006b. Geophys. Res. Lett. SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 161 Level 2 Version 4.2 Quality T

168 Help Bibliography Overview J. Geophys. Pickett, H. M., et al., Validation of Aura Microwave Limb Sounder OH and HO measurements, 2 Res. 113 , (D16), D16S30, doi:10.1029/2007JD008775, 2008. Table Pumphrey, H. C., C. J. Jimenez, and J. W. Waters, Measurement of HCN in the middle atmosphere by EOS (8), L08804, doi:10.1029/2005GL025656, 2006. MLS, Geophys. Res. Lett. , 33 BrO S from MLS on Aura, Pumphrey, H. C., W. G. Read, N. J. Livesey, and K. Yang, Observations of volcanic SO 2 T CH 8 , Atmos. Meas. Tech. (1), 195–209, doi:10.1002/2014JD021823, 2015. S 3 Cl Pumphrey, H. C., et al., Validation of middle-atmosphere carbon monoxide retrievals from the Microwave T CH S Limb Sounder on Aura, , 112 J. Geophys. Res. , D24S38, doi:10.1029/2007JD008723, 2007. 3 CN Read, W. G., Z. Shippony, and W. V. Snyder, Microwave Limb Sounder forward model algorithm theoretical T CH , Jet Propulsion Laboratory, JPL D-18130, 2004. basis document, Tech. rep. S 3 OH Read, W. G., Z. Shippony, M. J. Schwartz, N. J. Livesey, and W. V. Snyder, ¿e clear-sky unpolarized forward T IEEE Trans. Geosci. Remote Sens. 44 , (5), 1367–1379, model for the EOS Microwave Limb Sounder (MLS), ClO S doi:10.1109/TGRS.2006.873233, 2006. T S CO Read, W. G., et al., EOS Aura Microwave Limb Sounder upper tropospheric and lower stratospheric humidity validation, J. Geophys. Res. , 112 , D24S35, doi:10.1029/2007JD008752, 2007. T GPH S Rienecker, M. M., et al., ¿e GEOS-5 data assimilation system: A documentation of GEOS-5.0, Tech. rep. , T NASA, TM-104606, Tehcnical report series on Global Modeling and Data Assimilation, 2007. H S 2 O Rodgers,C.D.,Retrievalofatmospherictemperatureandcompositionfromremotemeasurementsofthermal T radiation, 14 (4), 609–624, 1976. Rev. Geophys. , HCl S T , 238 pp., World Scientic, 2000. Inverse Methods for Atmospheric Science, ¿eory and Practice Rodgers, C. D., HCN S 108 (D3), , J. Geophys. Res. Rodgers, C. D., and B. J. Connor, Intercomparison of remote sounding instruments, T 4116, doi:10.1029/2002JD002299, 2003. HNO S 3 Santee, M. L., N. J. Livesey, G. L. Manney, and W. G. Read, Methyl chloride from the Aura Microwave Limb T HO Sounder: First global climatology and assessment of variability in the upper troposphere and stratosphere, S 2 118 , 13,532–13,560, doi:10.1002/2013/JD020235, 2013. J. Geophys. Res. , T HOCl S , J. Geophys. Res. measurements, Santee, M. L., et al., Validation of the Aura Microwave Limb Sounder HNO 3 112 , D24S40, doi:10.1029/2007JD008721, 2007. T IWC S Santee, M. L., et al., Validation of the Aura Microwave Limb Sounder ClO measurements, , J. Geophys. Res. T , D15S22, doi:10.1029/2007JD008762, 2008. 113 IWP S Schwartz, M. J., W. V. Synder, and W. G. Read, MLS mesosphere-specic forward model algorithm theoretical T N S basis document, Tech. rep. , Jet Propulsion Laboratory, JPL D-28534, 2004. 2 O T Schwartz, M. J., W. G. Read, and W. V. Snyder, Polarized radiative transfer for Zeeman-split oxygen lines in S O 44 IEEE Trans. Geosci. Remote Sens. (5), 1182–1190, 2006. , the EOS MLS forward model, 3 T Schwartz, M. J., et al., Validation of the Aura Microwave Limb Sounder temperature and geopotential height OH S , D15S11, doi:10.1029/2007JD008783, 2008. measurements, J. Geophys. Res. , 113 T RHI S Simmons, A., M. Hortal, G. Kelly, A. McNally, A. Untch, and S. Uppala, ECMWF analyses and forecasts of T stratospheric winter polar vortex breakup: September 2002 in the southern hemisphere and related events., SO S J. Atmos. Sci. 62 , (3), 668–689, 2005. 2 T S T Aura Microwave Limb Sounder (MLS) x 162 Level 2 Version 4.2 Quality T

169 Help Bibliography Overview CN over the Pacic Ocean: Source, sinks and Singh, H. B., et al., In situ measurements of HCN and CH 3 J. Geophys. Res. , 108 (D20), 8795, doi:10.1029/2002HD003006, 2003. budgets, Table Tegtmeier, S., et al., SPARC Data Initiative: a comparison of ozone climatologies from international satellite , limb sounders, J. Geophys. Res. 118 (2), 12,229–12,247, doi:10.1002/2013JD019877, 2013. BrO S Toohey, M., and T. von Clarmann, Climatologies from satellite measurements: the impact of orbital sampling T CH S (4), 937–948, doi:10.5194/amt-6-937-2013, 2013. Atmos. Meas. Tech. on the standard error of the mean, , 6 3 Cl T von Clarmann, T., et al., Global stratospheric HOCl distributions retrieved from infrared limb emission spec- CH S J. Geophys. Res. , tra recorded by the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS), 3 CN , D05311, doi:10.1029/2005JD005939, 2006. 111 T CH S Wang, S., et al., Validation of Aura Microwave Limb Sounder OH measurements with Fourier Transform 3 OH , Ultra-Violet Spectrometer total OH column measurements at Table Mountain, California, J. Geophys. Res. T , D22301, doi:10.1029/2008JD009883, 2008. 113 ClO S Proc. Natl. Acad. Wang, S., et al., Midlatitude atmospheric OH response to the most recent 11-y solar cycle, in T Sci. , pp. 2023–2028, doi:10.1073/pnas.1117790110, 2013. S CO T Waters, J. W., et al., ¿e UARS and EOS Microwave Limb Sounder (MLS) experiments, , 56 , J. Atmos. Sci. GPH S 194–217, 1999. T H Waters, J. W., et al., An overview of the EOS MLS experiment, Tech. rep. , Jet Propulsion Laboratory, D-15745, S 2 O 2004. T HCl S Waters, J. W., et al., ¿e Earth Observing System Microwave Limb Sounder (EOS MLS) on the Aura satellite, T 44 , IEEE Trans. Geosci. Remote Sens. (5), 1075–1092, doi:10.1109/TGRS.2006.873771, 2006. HCN S Wickert, J., et al., Atmosphere sounding by GPS radio occultation: First results from CHAMP, Geophys. Res. T HNO (17), 3263–3266, 2001. 28 , Lett. S 3 , JetPropul- Wu, D.L., andJ.H.Jiang, EOSMLSalgorithmtheoreticalbasisforcloudmeasurements, Tech.rep. T HO S sion Laboratory, jPL D-19299, 2004. 2 T Wu, D. L., J. H. Jiang, and C. P. Davis, Aura MLS cloud ice measurements and cloudy-sky radiative transfer HOCl S , IEEE Trans. Geosci. Remote Sens. 44 (5), 1156–1165, 2006. model, T IWC Wu, D. L., J. H. Jiang, W. G. Read, R. T. Austin, C. P. David, A. Lambert, G. L. Stephens, D. G. Vane, and S 113 J. W. Waters, Validation of Aura MLS cloud Ice Water Content (IWC) measurements, J. Geophys. Res. , , T D15S10, doi:10.1029/2007LD008931, 2008. IWP S T J. Geophys. Wu, D. L., et al., Comparison of global cloud ice from MLS, CloudSat and correlative data sets, N S 2 113 , , D00A24, doi:10.1029/2008JD009946, 2009. Res. O T S O 3 T OH S T RHI S T SO S 2 T S T Aura Microwave Limb Sounder (MLS) x 163 Level 2 Version 4.2 Quality T

Related documents

288390

288390

S MS SITE IMPROVEMENT THROUGH MONITORING (SIMS) TOOL SITE ASSESSMENT Version 4.0, November 30 , 2018

More info »
What We Really Value  Beyond Rubrics in Teaching and Assessing Wr

What We Really Value Beyond Rubrics in Teaching and Assessing Wr

U t a t e U niv e rsit y a h St it lC o mm o [email protected] SU g Di a   - " . . 0 ( &  / &+* .    - " . . ( (   h W e R e al l y V alue: B e y ond Rubr ic s in t W a in a g a nd As ses sin g W...

More info »
SIMATIC NET PROFIBUS, Optical Link Module

SIMATIC NET PROFIBUS, Optical Link Module

1 Preface 2 Introduction 3 Network Topologies SIMATIC NET PROFIBUS Optical Link Module 4 Product Characteristics OLM / P11 V4.0 OLM / P12 V4.0 5 Installation and Maintenance OLM / P22 V4.0 OLM / P11 V...

More info »
TIBCO End of Support Information for TIBCO Product Versions

TIBCO End of Support Information for TIBCO Product Versions

End of Support Information Global Headquarters 3307 Hillview Avenue for TIBCO Product Versions Palo Alto, CA 94304 - 846 1000 Tel: +1 650 - 8450 Toll Free: 1 800 - 420 - - 846 Fax: +1 650 - 1005 March...

More info »
Corporate Actions and Events Guide

Corporate Actions and Events Guide

Corporate Actions and Events Guide for Market Capitalisation Weighted Indexes v4.1 This document applies to any index series where the guide is methodology documents . y referenced in the index specif...

More info »
keytruda pi.pdf

keytruda pi.pdf

HIGHLIGHTS OF PRESCRIBING INFORM ATION colorectal cancer that has progressed following treatment o 1 These highlights do not include all the information needed to use with a fluoropyrimidine, oxalipla...

More info »
Microsoft PowerPoint   Differences between ICC profile versions.pptx

Microsoft PowerPoint Differences between ICC profile versions.pptx

NTNU Phil Green versions Differences between ICC profile

More info »
HANDBOOK of METAL ETCHANTS

HANDBOOK of METAL ETCHANTS

HANDBOOK of METAL ETCHANTS Editors Perrin Walker William H. Tarn CRC Press Boca Raton Boston London New York Washington, D.C. © 1991 by CRC Press LLC

More info »
e ocvm3 vol 1 022010

e ocvm3 vol 1 022010

E-OCVM Version 3.0 Volume I European Operational Concept Validation Methodology EUROPEAN COMMISSION EUROCONTROL

More info »
Copy of SITESv2 Scorecard Summary

Copy of SITESv2 Scorecard Summary

SITES v2 Rating System For Sustainable Land Design and Development

More info »
UNSCEAR 2008 Report Vol.I

UNSCEAR 2008 Report Vol.I

This publication contains: VOLUME I: SOURCES SOURCES AND EFFECTS Report of the United Nations Scientific Committee on the Effects of Atomic Radiation to the General Assembly OF IONIZING RADIATION Scie...

More info »
D:

D:JOXALER 2ALER053.DVI

Law versus Morality as Regulators of Conduct Harvard Law School Steven Shavell, It is evident that both law and morality serve to channel our behavior. Law accom- plishes this primarily through the th...

More info »
Dissertation Master.odt

Dissertation Master.odt

PHONETIC CATEGORY LEARNING DISSERTATION Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University by Grant Leese McGu...

More info »
S 8264A/B/C Series  BATTERY PROTECTION IC

S 8264A/B/C Series  BATTERY PROTECTION IC

S-8264A/B/C Series BATTERY PROTECTION IC FOR 2-SERIAL www.ablic.com TO 4-SERIAL-CELL PACK (SECONDARY PROTECTION) www.ablicinc.com © ABLIC Inc., 2005-2018 Rev.4.4 _00 lithium-ion rechargeable batteries...

More info »
StateoftheClimate2015 lowres

StateoftheClimate2015 lowres

STATE OF THE CLIMATE I N 2015 Special Supplement to the Bulletin of the American Meteorological Society Vol. 97, No. 8, August 2016

More info »
pi opdivo

pi opdivo

HIGHLIGHTS OF PRESCRIBING INFORMATION • Recurrent or metastatic squamous cell carcinoma of the head and neck These highlights do not include all the information needed to use OPDIVO safely • 240 mg ev...

More info »
Standard Document

Standard Document

® ® Microsoft Windows 8.x and 10 ® Compatibility with Xerox Equipment December 2016

More info »
Leveraging Long and Short term Information in Content aware Movie Recommendation

Leveraging Long and Short term Information in Content aware Movie Recommendation

Leveraging Long and Short-term Information in Content-aware Movie Recommendation Wei Zhao Haixia Chai, Benyou Wang Jianbo Ye SIAT, Chinese Academy of Sciences Tencent Pennsylvania State University wei...

More info »
2016 bams eee low res

2016 bams eee low res

EXPLAINING EXTREME EVENTS OF 2016 From A Climate Perspective Special Supplement to the Bulleti n of the American Meteorological Society Vol. 99, No. 1, January 2018

More info »
Deep image reconstruction from human brain activity

Deep image reconstruction from human brain activity

bioRxiv preprint first posted online Dec. 28, 2017; http://dx.doi.org/10.1101/240317 The copyright holder for this preprint . doi: (which was not peer-reviewed) is the author/funder, who has granted b...

More info »