UARS MLS DATA QUALITY DOCUMENT VERSION 4 DATA JANUARY 8, 1997 POINTS OF CONTACT: - General Dr. Lucien Froidevaux Information Jet Propulsion Laboratory and O3 Mail Stop 183-701 Pasadena CA, USA 91109 E-mail: lucien@mls.jpl.nasa.gov Phone: (818) 354-8301 - Temperature Dr. Dong Wu Jet Propulsion Laboratory Mail Stop 183-701 Pasadena CA, USA 91109 E-mail: dwu@mls.jpl.nasa.gov Phone: (818) 353-1954 - H2O Dr. Hugh Pumphrey Department of Meteorology The University of Edinburgh Edinburgh EH9 3JZ, Scotland E-mail: hcp@met.ed.ac.uk Phone: 0131-650-6026 (44-131-650-6026 if outside U.K.) - ClO Dr. Joe Waters (UARS MLS Principal Investigator) Jet Propulsion Laboratory Mail Stop 183-701 Pasadena CA, USA 91109 E-mail: joe@mls.jpl.nasa.gov Phone: (818) 354-3025 - SO2 Dr. William Read Jet Propulsion Laboratory Mail Stop 183-701 Pasadena CA, USA 91109 E-mail: bill@mls.jpl.nasa.gov Phone: (818) 354-6773 - HNO3 Dr. Michelle Santee Jet Propulsion Laboratory Mail Stop 183-701 Pasadena CA, USA 91109 E-mail: mls@mls.jpl.nasa.gov Phone: (818) 354-9424 - WORLD WIDE WEB site (UARS MLS Homepage): http://mls.jpl.nasa.gov This document gives a brief description of the type and quality of Version 0004 (Version 4, for short) data from the Microwave Limb Sounder (MLS) aboard the Upper Atmosphere Research Satellite (UARS). These data (produced from MLS software Version 422) are publicly available from the NASA Goddard Space Flight Center Distributed Access Archive Center (DAAC) after being produced on Goddard's UARS Central Data Handling Facility (CDHF). The seven upper atmospheric parameters retrieved by MLS in Version 4 Level 3 data files are temperature (parameter name TEMP), ozone from the 205-GHz band radiances (O3_205), ozone from the 183-GHz band radiances (O3_183), water vapour (H2O), chlorine monoxide (CLO), sulphur dioxide (SO2), and nitric acid (HNO3). HNO3 is a new product for Version 4, and SO2 files are also routinely produced for every day (but see the caveats under the HNO3 and SO2 sections). Refinements in spectroscopic data for O2 (see the temperature section) have led to systematic height-dependent decreases in the tangent-point pressure (of order 200 m). Other refinements in relative pointing of the 205- and 183-GHz radiometers (see details below) have led to additional (species-dependent) adjustments. Version 4 software changes specific to each MLS profile are described in separate sections below. Plans for the next major upgrade of MLS data processing are to use more (including non-optically-thin) radiances, include iterations, improve error covariances, increase vertical resolution, apply hydro- static balance more rigorously, and provide new products of upper tro- pospheric water vapour, geopotential height, and temperature variances at gravity-wave scales. GENERAL INFORMATION UARS MLS measurements are obtained from observations of millimeter- wavelength thermal emission (day and night) as the instrument field of view (FOV) is scanned vertically through the earth's atmospheric limb. The 63-GHz radiometer measures radiances from O2 emission (MLS band 1) to obtain tangent-point pressure and atmospheric temperature, the 205-GHz radiometer provides chlorine monoxide (ClO) (bands 2 and 3), ozone (band 4), nitric acid (band 4), and sulphur dioxide (bands 2 and 4), while the 183-GHz radiometer provides water vapour (band 5) and ozone (band 6). The MLS standard operating mode consists of full vertical scans (approximately 0 to 90 km tangent heights) every 65.536 sec (every MLS major frame or MMAF), and the calibrated limb radiances from each band are used to retrieve atmospheric profiles; pressure is the independent variable. The MLS instrument calibration is described by Jarnot et al. [JGR 101, 9957-9982, 1996]. The UARS orbital motion (7 km/s) during limb scans smears the profile measurements over about 400 km in a direction perpendicular to the MLS line of sight; the profile is also "averaged" over a similar distance along the line of sight, because of radiative transfer "smearing." The antenna fields of view (half power beamwidths) are 3.0, 3.6, and 9.5 km for the 205-GHz, 183-GHz, and 63-GHz measurements, respectively. The measurement latitudinal coverage is from 34 degrees on one side of the equator to 80 degrees on the other. UARS performs a yaw maneu- ver at roughly 36-day intervals (every "UARS month" or "yaw period"), when MLS high-latitude coverage switches between north and south. Within each UARS month the UARS orbit plane precesses slowly with respect to the Earth-Sun line. The orbit precession causes the mea- surements to sweep through essentially all local solar times during the course of a UARS month, becoming 20 min earlier each day. The MLS retrievals (for Version 4 and previous versions) are based on the sequential estimation algorithm, which follows optimal estima- tion theory. Some approximations were made in the treatment of error covariances for certain "constrained" quantities (e.g. for temperature and tangent pressure, which are constrained during the constituent retrievals); these approximations will be avoided in Version 5. The vertical resolution for retrieved profiles is 2 UARS pressure surfaces (delta log10(P) = 1/3, or about 6 km). These profiles are represented as piecewise-linear functions with breakpoints at alternate (even-numbered) UARS pressure surfaces (e.g. 46, 22, 10 hPa). Values on the even-numbered surfaces for Level 3AT files are the retrieved breakpoint values, while those on the odd-numbered surfaces (e.g. 32, 15 hPa) are averages of the values on adjacent even-numbered surfaces. The Level 3AL profiles have an additional linear interpolation with respect to latitude to generate an evenly spaced latitude grid. The `quality' field in the Level 3A files is the retrieval's estimated uncertainty, includes random and some systematic components, and is obtained by propagating precisions of the radiance measurements, estimates of constrained parameter uncertainties, forward model inaccu- racies, and some calibration uncertainties through the retrieval soft- ware. For parameters with weak signals (ClO, HNO3, SO2), the random (noise) components will dominate the estimated quality field, but for temperature, O3, and H2O, this field can be looked at as a lower bound for the total error (rss of random and systematic components). At the conclusion of the retrieval, the estimated uncertainty is compared with the assumed a priori uncertainty, and if the ratio is greater than 0.5, the quality is set negative, indicating that the retrieved value is influenced more than 25% by the a priori. Additional quality indicators exist in the MLS Level 3 parameter files (one file/day); these files are described in the "Standard Formatted Data Units, MLS Level 3TP Parameter File" and "Standard Formatted Data Units, MLS Level 3LP Parameter File" documents. The diagnostic flag MMAF_STAT should be used to screen poor data (see more discussion below) on an individual major frame basis; possible values of this (single character) flag are described in the Table below. Also, other diagnostic parameters in the MLS Level 3 parameter files are related to radiance quality and radiance fits in each band (flags have names corresponding to the primary parameter being retrieved in each band (QUALITY_TEMP, QUALITY_CLO, QUALITY_O3_205, QUALITY_H2O, and QUALITY_O3_183 for bands 1, 2, 4, 5, and 6, respectively). These flags have a value less than 4 under (very infrequent) abnormal conditions (at least among major frames with MMAF_STAT equal to 'G', 't', or 'T'). Table describing the major frame quality indicator MMAF_STAT _____________________________________________________________________ |_____________________________________________________________________| | | | MMAF_STAT MMAF_STAT | | Value Meaning | |_____________________________________________________________________| | | | 'G' Good Data | |_____________________________________________________________________| | | | 't' Temperatures missing from NCEP data at heights below 22 hPa | |_____________________________________________________________________| | | | 'T' Temperatures missing from NCEP data at heights below 100 hPa| |_____________________________________________________________________| | | | 'M' Missing too many tangent points from scan | |_____________________________________________________________________| | | | 'P' Pointing anomaly | |_____________________________________________________________________| | | | 'S' Scan mode anomaly (i.e., not normal full scan range) | |_____________________________________________________________________| | | | 'B' Bad Data | |_____________________________________________________________________| |_____________________________________________________________________| The 183-GHz radiometer failed in April 1993. The last good full day of data was April 15, 1993 (UARS Day 582), for ozone from the 183-GHz band and for H2O. Therefore, ONLY FIELDS FOR O3_183 and H2O PRIOR TO APRIL 16 1993 SHOULD BE USED FOR SCIENTIFIC STUDIES. Occasional interference effects (induced by the switching mirror stepper motor at low spacecraft battery voltage) can perturb the radiances and retrieved parameter values. This problem occurred largely between September 1992 and June 1993, typically just before sunrise (at the satellite location) for a few minutes. Some diagnos- tics are sensitive to this effect (quality fields in the Level 3 para- meter files show a degradation for ClO and O3_205), but this is not reflected in the error bars (quality values) given in the Level 3A files. After 2.3 years in orbit (in late December 1993), the antenna- scanning mechanism began to exhibit signs of wear. March 1994 through May 1994, and July 1994 were periods of testing and significantly reduced data gathering; these months have from one third of the days with bad data to almost all bad days (days with no profiles retrieved). August and September of 1994 contain mostly good data, but the months of October 1994 through January 1995 again have very few days of useful atmospheric profile data. Reverse scanning and other modifications to the operations (including short periods of "mechanism rest" every orbit) have been implemented since February 1, 1995. Very little limb data gathering (typically only a few days per month) occurred from February through July 1995, which was a period during which instrument power sharing began for UARS (in May 1995) because of poor solar array performance. Since June 1995, MLS has been in a mode of operation characterized by off periods for power savings and on periods during which typically 2 days of full (reverse) scans are obtained followed by one day of limb tracking at altitudes near 18 km. The limb tracking days do not lead to standard catalogued profile data files, although there is some information on the atmosphere from those days. The August 1995 through September 1996 time period generally contains about one half to one third catalogued good days for atmospheric profiles, and this mode of operation is expected to continue. We note that MLS has been operating without "scan slips" since January 1995. Calendars describing the various instrument modes (especially useful for time periods since March 1994, but also prior to this) can be made available for interested users; however, access of the diagnostic parameters mentioned above should generally allow users to appropria- tely screen for bad quality data and is ultimately needed for optimum use of the MLS Level 3 data. PARAMETER #1: TEMPERATURE This document supplements the MLS temperature validation paper by Fishbein et al. [JGR 101, 9983-10,016, 1996], which should be studied by any user of these data. CHANGES FROM VERSION 3 TO VERSION 4: Oxygen line broadening parameters have been revised based on Liebe's re-interpretation of laboratory absorption data. The numerical accuracy of computations used to generate the linearized forward model has been improved. Radiance error from numerical approxi- mations is estimated to be less than 0.2 K. Band 1 sideband ratios have been revised to reduce radiance residuals. The revised values were obtained from simultaneous retrievals of side- band ratios, temperature, and tangent-point pressure, and did not involve comparisons with correlative data. Channel 11 experiences intermittent interference throughout the mission from an intermediate frequency oscillator and is not used in the retrievals for Version 4. A priori temperature errors have been increased to 20 K at all levels, reducing the sensitivity to NCEP temperatures (especially important in the winter at high latitude). The useful temperature range as reported in the temperature validation paper has been extended downward from the 22 hPa lower limit in Version 3, and useful temperatures are now produced from 46 hPa to 0.46 hPa. ESTIMATED PRECISION AND ACCURACY: UARS Estimated V4 - V3 Standard Single Profile Estimated Average Index Pressure Precision (1) Accuracy (2) Differences (3) 20 0.46 hPa 0.8 K 5 K -1 K 18 1 hPa 0.4 K 5 K +4 K 16 2.2 hPa 0.4 K 4 K +1 K 14 4.6 hPa 0.6 K 5 K +4 K 12 10 hPa 0.9 K 4 K +2 K 10 22 hPa 1.1 K 4 K +1 K 8 46 hPa 1.4 K 7 K +3 K Notes: (1) Precisions (1-sigma) are derived using an error analysis as described in the temperature validation paper, with one exception: forward model noise (described on pg. 9988 of Fishbein et al., 1996) is not included in the estimate of precision. These values agree better with empirical estimates. (2) Accuracies are derived using an error analysis as described in the temperature validation paper, with a few changes: channel 11 is not included in the analysis, the contributions from sideband ratio errors are decreased to 7%, and oxygen line broadening parameter errors are reduced to 3%. (3) Average differences (V4 minus V3 data) are derived from globally- averaged profiles for selected days. SYSTEMATIC EFFECTS: Differences between MLS and NCEP/UKMO temperatures are modulated with the UARS orbital and yaw cycles. Most of these differences are believed to be associated with MLS sampling of atmospheric diurnal variability. A component of the variability, of roughly 1-2 K peak-to- peak amplitude, is believed to arise from systematic errors associated with extraneous radiances from the MLS antenna. Zonal mean differences and zonal rms differences between the MLS tempe- rature field and the NCEP analysis show biases which depend on: whether the data are obtained during the ascending or descending sides of the UARS orbit, whether MLS is looking north or south, and when during a yaw period the measurements are taken. These systematics are 1-3 K in the stratosphere, but can be as much as 10 K in the lower mesosphere. MLS global average temperatures are 3 K warmer than NCEP temperatures at 46 hPa and 0-2 K warmer at 10 hPa through 1 hPa. Mesospheric lapse rates are significantly more negative in Version 4. A notch, usually negative, is observed in many MLS temperature profiles at 0.22 hPa, which is outside the recommended range of useful informa- tion. The radiances are sensitive to smaller scale temperature perturbations than the current retrieval grid allows. When these perturbations are large, the retrieved profile, especially at 1 hPa, may not represent the smoothed true profile. CAVEATS: The retrieval for temperature uses an a priori estimate based on both the NCEP daily analysis (when available) and a month-dependent latitude- dependent climatology developed by the UARS science team. Although the profiles extend from 1000 to 0.0001 hPa, useful information is provided by MLS only from 46 hPa to 0.46 hPa. Above 0.22 hPa the profiles relax slowly to the climatology. Below 46 hPa, the profiles are linearly interpolated from NCEP daily analyses (or climatology when necessary) onto the even-numbered surfaces. TEMPERATURE VALUES OUTSIDE THE RANGE FROM 46 TO 0.46 hPa SHOULD NOT BE USED WITHOUT CONSULTING THE MLS TEAM. The MLS radiances are less sensitive to temperature at 46 hPa than to temperature at any of the other recommended levels, as indicated by the estimated accuracy, and users are advised to exercise caution in inter- preting these data. The 46 hPa temperatures seem to be more sensitive to systematic errors associated with calculated radiances. Positive notches in temperature (and biases relative to NCEP data) greater than 15 K can occur, primarily at high latitudes during winter. During a scan, the tangent point moves approximately 400 km and the radiation path length through the atmosphere is also of that order. Currently, a 1D (height) linearized forward model is used to fit the radiances in a one pass (top to bottom) retrieval through the data. These approximations are particularly poor in winter at high latitudes. SPATIALLY VARYING SYSTEMATIC ERRORS ARE LARGE AND WAVE AMPLITUDES MAY BE MISREPRESENTED DURING PERIODS OF LARGE WAVE ACTIVITY AND STRONG HORIZONTAL GRADIENTS. THE PARAMETER FILES SHOULD BE EXAMINED AND ONLY PROFILES WITH MMAF_STAT='G', 't', or 'T' AND QUALITY_TEMP=4 SHOULD BE USED FOR SCIENTIFIC STUDIES. Profiles with MMAF_STAT set to 'T' (which occurs quite infrequently) may have biases of about 1 K at 46 hPa. MOREOVER, ONLY DATA HAVING POSITIVE ESTIMATED UNCERTAINTIES (QUALITY FIELD IN LEVEL 3 DATA FILES) SHOULD BE USED. PARAMETER:#2: O3_205 (Ozone retrieved from 205-GHz radiometer radiances) This document supplements the MLS O3 validation paper by Froidevaux et al. [JGR 101, 10,017-10,060, 1996], which should be studied by any user of these data; additional useful information on MLS ozone quality can be found in the articles by Cunnold et al. [JGR 101, 10,061-10,075, 1996, and JGR 101, 10,335-10,350, 1996]. CHANGES FROM VERSION 3 TO VERSION 4: A post-launch estimate of the pointing offset between the 63- and 205-GHz radiometers is now used [this was obtained from MLS antenna scans across the Moon]; the pointing offset change from Version 3 corresponds to a tangent height change of about 430 m (with larger tangent heights in V4, relative to the 63-GHz FOV). HNO3 is now retrieved from band 4 radiances (in addition to O3_205); this has had some impact on O3_205 in the lowermost stratosphere. Radiances for tangent pressures of 100 hPa are included in Version 4 by spline interpolation of the radiances to minimize vertical sampling gaps sometimes created in Version 3 (but a criterion to neglect opti- cally thick radiances is still in effect in Version 4). O3_205 values are affected by changes which occurred in band 1 (for temperature and - mostly - tangent pressure). Changes in error-budget accounting have produced slightly reduced (and more realistic) estimates of individual profile precision (the quality field placed in the data files) in the lower stratosphere. ESTIMATED PRECISION AND ACCURACY: UARS Estimated V4 - V3 Standard Single Profile Estimated Average Index Pressure Precision (1) Accuracy (2) Differences (3) (hPa) (ppmv) (%) (ppmv) (%) (ppmv) (%) 20 0.46 0.37 20 0.3 15 -0.02 - 1 18 1 0.31 10 0.3 10 -0.1 - 3 16 2.2 0.23 4 0.3 5 -0.1 - 1 14 4.6 0.20 2 0.3 5 -0.1 - 1 12 10 0.18 2 0.3 5 -0.2 - 2 10 22 0.16 3 0.4 5 -0.3 - 4 8 46 0.22 11 0.4 20 -0.5 -22 6 100 0.55 > 50 > 50 0.5 See (3) Notes: (1) Precisions (1-sigma) are estimates obtained by propagating the radiance measurement precisions through the retrievals, and are consistent with observed variability in latitude bands where meteorological variability is small. These values are based on Version 3 but should be very close to the Version 4 values. (2) Accuracies are estimates based on various error sources and statisti- cal comparisons with other data sets (see above validation paper); they should be viewed as 1-sigma values. These values are based on Version 3 but should be very close to the Version 4 values (the main change is a slight reduction in estimated systematic uncertainty arising from tangent pressure knowledge). (3) V4-V3 differences are from globally-averaged profiles for selected months. Variations in these global differences across month or lati- tude typically vary only at the 1 to 2 % level for pressures in the 22 to 0.46 hPa range. At 46 hPa, differences can vary across month or latitude by roughly 8 % about the percentage differences listed above, with largest V4-V3 differences in the tropics and smallest (< 10%) differences at high latitudes. At 100 hPa, V4 values are generally larger than V3 values, especially at low latitudes, where the increase is often large (between 0.5 and 1 ppmv) and latitudinal gradients can change significantly; percentage differences are large but not as meaningful in this region of relatively low ozone mixing ratios, where MLS data are still a research product. V4-V3 differences for more limited latitude regions (e.g. the polar regions) can depart somewhat from the above tabulation and change significantly with time of year. Column ozone above 100 hPa is typically 5-10% smaller in V4 data than in V3 data. SYSTEMATIC EFFECTS: Some aspects of the 100 hPa retrievals appear to be better in V4 data (there are fewer negative biases in the tropics, and artificial oscillations tied to the 36-day UARS yaw cycle are significantly reduced, although still present). Nevertheless, Version 4 values at 100 hPa are sometimes unrealistically large, and the 46 hPa values can be too small, compared to correlative data or climatology; these problems (typically of order 0.1 to 0.3 ppmv but on occasion signifi- cantly more) appear primarily in the tropics and in the polar regions. Thus, data quality can be worse for V4 than V3 in the lowermost strato- sphere. Some significant changes in the latitudinal structure of 100 hPa data are observed, compared to V3 data. More analyses with reprocessed data are needed to draw conclusions about reliability at this level. Version 3 mid to upper stratospheric biases of order 5 % with respect to SAGE II ozone data were present; these are somewhat reduced in Version 4 (since V4 ozone values there are smaller than V3 ozone by a few percent). CAVEATS: The retrieval for ozone uses an a priori estimate based on a month- dependent latitude-dependent climatology developed by the UARS science team. While the O3_205 profiles extend from 464 hPa to 4.6e-4 hPa, the values at pressures outside the range 100 hPa to 0.46 hPa are mostly climatological. The useful vertical range for MLS O3_205 for most purposes is 46 hPa to 0.46 hPa. O3_205 values at 100 hPa are still to be considered as a research product and should be used with caution, given the systematics mentioned above. O3_205 VALUES OUTSIDE THE RANGE FROM 100 TO 0.46 hPa SHOULD NOT BE USED FOR SCIENTIFIC STUDIES, AND VALUES AT 100 hPa SHOULD NOT BE USED WITHOUT CONSULTING THE MLS TEAM. The 205-GHz radiances show better radiance residual closure than the 183-GHz radiances, and O3_205 is the recommended ozone for stratos- pheric studies. However, O3_183 is better for mesospheric studies (up to 0.05 hPa), where the 205-GHz radiances lose sensitivity. The 2 ozone retrievals give results which track very well and are within the respective error bars. We note that the 183-GHz retrievals lead to ozone values typically 0 to 6 percent smaller than the 205-GHz values, for the pressure range from 10 to 0.46 hPa; the O3_183 values are only slightly larger (1 to 2 %) than the O3_205 values at 22 hPa, and larger by about 20 % at 46 hPa. These differences are based on data from one UARS month (Aug. 14 to Sep. 20, 1992), but they should be basically the same for any time period. THE PARAMETER FILES SHOULD BE EXAMINED AND ONLY PROFILES WITH MMAF_STAT='G', 't', or 'T' AND QUALITY_O3_205=4 SHOULD BE USED FOR SCIENTIFIC STUDIES. Profiles with MMAF_STAT set to 'T' (which occurs quite infrequently) may have slight biases at 46 and 100 hPa. MOREOVER, ONLY DATA HAVING POSITIVE ESTIMATED UNCERTAINTIES (QUALITY FIELD IN LEVEL 3 DATA FILES) SHOULD BE USED. PARAMETER #3: O3_183 (Ozone retrieved from 183-GHz radiometer radiances) This document supplements the Version 3 MLS O3 validation paper by Froidevaux et al. [JGR 101, 10,017-10,060, 1996] and the mesospheric O3 validation paper by Ricaud et al. [JGR 101, 10,077-10,089], which should be studied by any user of these data. CHANGES FROM VERSION 3 TO VERSION 4: A post-launch estimate of the pointing offset between the 63- and 183- GHz radiometers is now used [this was obtained from MLS antenna scans across the Moon]; the pointing offset change from Version 3 corres- ponds to a tangent height change of about 180 m (with larger tangent heights in V4, relative to the 63-GHz FOV). Some forward model approximations related to frequency averaging have been eliminated. O3_183 values are affected by changes which occurred in band 1 (for temperature and - mostly - tangent pressure). Changes in error-budget accounting have produced reduced (and more realistic) estimates of individual profile precision (the quality field placed in the data files) in the lower stratosphere. ESTIMATED PRECISION AND ACCURACY: UARS Estimated V4 - V3 Standard Single Profile Estimated Average Index Pressure Precision (1) Accuracy (2) Differences (3) (hPa) (ppmv) (%) (ppmv) (%) (ppmv) (%) 26 0.05 0.30 55 0.5 > 50 -0.06 - 9 24 0.1 0.26 30 0.3 35 0.0 0 22 0.22 0.17 17 0.2 20 -0.1 - 7 20 0.46 0.10 6 0.2 10 -0.1 - 7 18 1 0.10 3 0.4 15 -0.1 - 4 16 2 0.10 2 0.8 15 -0.5 -10 14 4.6 0.10 1 0.9 10 -0.3 - 4 12 10 0.13 2 0.9 10 0.0 0 10 22 0.20 3 0.8 15 0.0 0 8 46 0.10 6 0.8 45 0.15 8 Notes: (1) Precisions (1-sigma) are estimates obtained by propagating the radiance measurement precisions through the retrievals, and are consistent with observed variability in latitude bands where meteorological variability is small. These values are based on Version 3 but should be very close to the Version 4 values. (2) Accuracies are still dominated by pre-launch calibration uncertainties for the sideband ratios and are based on V3 values; V4 values are probably slightly improved over these tabulations because of some reduc- tion in estimated systematic uncertainties arising from tangent pressure knowledge. Estimates should be viewed as 1-sigma values. (3) V4-V3 differences are from globally-averaged profiles for selected months. V4-V3 differences for smaller latitude regions (e.g. the polar regions) can depart somewhat from the above tabulation. SYSTEMATIC EFFECTS: More analyses with reprocessed data are needed to draw final conclu- sions regarding Version 4 data, but there are no obvious significant issues. CAVEATS: O3_183 is better than O3_205 for mesospheric studies (up to 0.05 hPa) where the 205-GHz radiances are much smaller. The 2 ozone retrievals give results which track very well and are within the respective error bars. We note that the 183-GHz retrievals lead to ozone values typically 0 to 6 percent smaller than the 205-GHz values, for the pressure range from 10 to 0.46 hPa; the O3_183 values are only slightly larger (1 to 2 %) than the O3_205 values at 22 hPa, and larger by about 20 % at 46 hPa. These differences are based on data from one UARS month (Aug. 14 to Sep. 20, 1992), but they should be basically the same for any time period. Although good consistency exists between the 2 ozone fields, detailed studies of possible systematic effects using the 18 months of available O3_183 V4 data have not been completed. The retrieval for ozone uses an a priori estimate based on a month- dependent latitude-dependent climatology developed by the UARS science team. While the O3_183 profiles extend from 464 hPa to 4.6e-4 hPa, the values at pressures outside the range 46 hPa to 0.046 hPa are mostly climatological. The useful vertical range for MLS O3_183 for most purposes is 46 hPa to 0.046 hPa. O3_183 values at 100 hPa are not to be used because of line saturation effects not treated fully in Version 4 (this may have a small impact at other heights...). O3_183 VALUES OUTSIDE THE RANGE FROM 46 TO 0.046 hPa SHOULD NOT BE USED FOR SCIENTIFIC STUDIES. THE PARAMETER FILES SHOULD BE EXAMINED AND ONLY PROFILES WITH MMAF_STAT='G', 't', or 'T' AND QUALITY_O3_183=4 SHOULD BE USED FOR SCIENTIFIC STUDIES. MOREOVER, ONLY DATA HAVING POSITIVE ESTIMATED UNCERTAINTIES (QUALITY FIELD IN LEVEL 3 DATA FILES) SHOULD BE USED. PARAMETER #4: H2O This document supplements the MLS H2O validation paper by Lahoz et al. [JGR 101, 10,129-10,149, 1996], which should be studied by any user of these data. CHANGES FROM VERSION 3 TO VERSION 4: A post-launch estimate of the pointing offset between the 63- and 183- GHz radiometers is now used [this was obtained from MLS antenna scans across the Moon]; the pointing offset change from Version 3 corres- ponds to a tangent height change of about 180 m (with larger tangent heights in V4, relative to the 63-GHz FOV). The sideband ratios for band 5 have been revised, based on analyses of the saturated portion of the radiance growth curves. Some forward model approximations related to frequency averaging have been eliminated. A new climatology is used, based on SAGE II and HALOE data. H2O values are affected by changes which occurred in band 1 (for tem- perature and - mostly - tangent pressure). Changes in error-budget accounting have produced slightly reduced (and more realistic) estimates of individual profile precision (the quality field placed in the data files) in the lower stratosphere. The above changes have generally eliminated the Version 3 problems of poor data quality in the winter polar regions. ESTIMATED PRECISION AND ACCURACY: UARS Estimated V4 - V3 Standard Single Profile Estimated Average Index Pressure Precision (1) Accuracy (2) Differences (3) (hPa) (ppmv) (%) (ppmv) (%) (ppmv) (%) 22 0.22 0.4 6 0.9 14 -0.13 - 2 20 0.46 0.4 7 0.7 11 -1.0 -17 18 1 0.3 5 0.7 10 -0.91 -14 16 2.2 0.2 4 0.6 9 -0.65 -11 14 4.6 0.2 4 0.5 8 -0.51 - 9 12 10 0.1 2 0.5 9 -0.02 0 10 22 0.2 5 0.8 19 -0.19 - 4 8 46 0.2 6 1.2 25 -0.09 - 2 Notes: (1) Precisions (1-sigma) are based on observed variability in latitude bands where meteorological variability is small, hence the true precisions may be somewhat better than these estimates. These values are based on Version 3 data as insufficient Version 4 data was avail- able at the time of writing, but the values are not expected to change much in the future. (2) Accuracies are estimates based on various error sources and statisti- cal comparisons with other data sets (see above validation paper); they should be viewed as 1-sigma values. These values are based on Version 3 but should be very close to the Version 4 values (the main change is a slight reduction in estimated systematic uncertainty arising from tangent pressure knowledge). (3) V4-V3 differences are from globally-averaged profiles for several UARS months. The difference is larger at 46 hPa in the dehydrated southern polar vortex (UARS days 338-375) because (a) the a-priori has less effect on the retrieved product in v4 and (b) v4 uses a different (and drier) a priori. SYSTEMATIC EFFECTS: A 'notch' of high H2O values appears in the lower mesosphere (in the region near 0.1 hPa); this is believed to be an artifact and needs further study [it occurs outside the currently recommended range of useful information]. Comparisons with other UARS data and/or correlative data suggest that MLS Version 4 H2O may be ~0-20 % too high in the range 46 hPa to 0.2 hPa. This is better than Version 3, but a preliminary conclusion. This estimate is probably conservative in the mid-stratosphere and will be updated when more of the Version 4 data becomes available. CAVEATS: The retrieval for water vapour uses an a priori estimate based on a month-dependent latitude-dependent climatology developed by the MLS science team (based on SAGE II and HALOE data). The useful vertical range for MLS H2O for most purposes is 46 hPa to 0.2 hPa. Above this, and particularly at 0.1hPa, the absolute values of the data are unrea- listic, although the seasonal variation is similar to that seen in other data. H2O VALUES AT PRESSURES LARGER THAN 46 hPa SHOULD NOT BE USED FOR SCIENTIFIC STUDIES AND VALUES AT PRESSURES SMALLER THAN 0.2 hPa SHOULD NOT BE USED FOR SCIENTIFIC STUDIES WITHOUT CONSULTING THE MLS TEAM. THE PARAMETER FILES SHOULD BE EXAMINED AND ONLY PROFILES WITH MMAF_STAT='G', 't', or 'T' AND QUALITY_H2O=4 SHOULD BE USED FOR SCIENTIFIC STUDIES. MOREOVER, ONLY DATA HAVING POSITIVE ESTIMATED UNCERTAINTIES (QUALITY FIELD IN LEVEL 3 DATA FILES) SHOULD BE USED. NOTE: Retrieved H2O values for pressures 0.2 to 1 hPa can be unreliable for the first hour of data following rare periods when the instrument was temporarily turned off. PARAMETER #5: ClO This document supplements the MLS ClO validation paper by Waters et al. [JGR 101, 10,091-10,127, 1996], which should be studied by any user of these data. CHANGES FROM VERSION 3 TO VERSION 4: The "old" line strength file used inadvertently in Version 3 processing has been corrected. This results in Version 4 ClO values being gene- rally 8% lower than Version 3. Using MLS-retrieved HNO3 to account for its effects on ClO has reduced values of enhanced lower stratospheric ClO by approximately 0.2 ppbv under conditions of depleted lower stratospheric HNO3 (such as frequen- tly occur in the Antarctic winter vortex). A post-launch estimate of the pointing offset between the 63- and 205-GHz radiometers is now used [this was obtained from MLS antenna scans across the Moon]; the pointing offset change from Version 3 corresponds to a tangent height change of about 430 m (with larger tangent heights in V4, relative to the 63-GHz FOV). The programming error has been corrected which caused some V3 L3AL ClO files for data prior to 15 April 1993 to have values too large by a factor of 1.32 (isotopic correction was inadvertently applied twice). The Version 3 L3AT files were not affected. Radiances for tangent pressures of 100 hPa are included in Version 4 by spline interpolation of the radiances to minimize vertical sampling gaps sometimes created in Version 3 (but a criterion to neglect opti- cally thick radiances is still in effect in Version 4). Additional minor changes in ClO occur due to changes in other parameters retrieved from MLS measurements. Changes in error-budget accounting have produced slightly reduced (and more realistic) estimates of individual profile precision (the quality field placed in the data files) at 100 hPa. ESTIMATED PRECISION AND ACCURACY: UARS Estimated Standard Single Profile Estimated Index Pressure Precision (1) Accuracy (2) 20 0.46 hPa 1.5 ppbv 0.2 ppbv and 20 % 18 1.0 hPa 1.1 ppbv 0.2 ppbv and 15 % 16 2.2 hPa 0.7 ppbv 0.15 ppbv and 15 % 14 4.6 hPa 0.5 ppbv 0.15 ppbv and 15 % 12 10 hPa 0.4 ppbv 0.15 ppbv and 15 % 10 22 hPa 0.3 ppbv 0.2 ppbv and 15 % 8 46 hPa 0.4 ppbv 0.2 ppbv and 15 % 6 100 hPa 1.0 ppbv .4-.8 ppbv and 25 % Notes: (1) Precisions (1-sigma) are typical values obtained from the retrieval algorithm and are consistent with the observed rms variability in situations where ClO is below the instrument noise level. The quality field in the Level 3A files gives the estimated precision of individual profiles. Precision can be improved by averaging together individual profiles. (2) The overall estimates of accuracy are the root sum square of the bias error (given above in ppbv) and the scaling error (the product of the percentage accuracy given above and the retrieved mixing ratio). These values are believed to represent envelopes which are not often exceeded, roughly 90% confidence (2-sigma) values. These accuracies do not include the random noise which, for a single profile, is the estimated precision. COMPARISON OF V4 AND V3 CLO: The following tables compare MLS Version 3 and Version 4 ClO. Because of the ClO variability between day and night, and its enhancement in the lower stratosphere polar winter vortices, separate comparisons are shown for day, night, day-night and for (A) "Antarctic ozone hole" (polar winter vortex) and (B) low-mid latitude conditions. The compa- risons are made between the V4 data, and V3 data multiplied by 0.92 to account for the V3 scaling error due to inadvertent use of an old ClO line strength file. Comparison of V4 and V3* (=V3x0.92) ClO. Values are in ppbv; "day" is sza <90 deg; "night" is sza >90 deg. ------------------------------------------------------------------ (A) 70S-90S: 14 Aug - 19 Sep 92 (avg of ~2000 "day", ~200 "night") ------------------------------------------------------------------ ----- day ------ ---- night ----- --- day-night -- hPa V3* V4 V4-V3* V3* V4 V4-V3* V3* V4 V4-V3* 0.46 0.03 0.00 -0.03 0.06 0.13 0.07 -0.03 -0.13 -0.10 1.0 0.31 0.27 -0.04 0.13 0.11 -0.02 0.18 0.16 -0.02 2.2 0.38 0.40 0.02 0.38 0.43 0.05 0.00 -0.03 -0.03 4.6 0.54 0.46 -0.08 0.15 0.10 -0.05 0.39 0.36 -0.03 10 -0.09 -0.05 0.04 -0.05 -0.04 0.01 -0.04 -0.01 0.03 22 1.22 1.23 0.01 0.06 0.04 -0.02 1.16 1.19 0.03 46 1.94 1.65 -0.29 0.32 0.20 -0.12 1.62 1.45 -0.17 100 -0.48 -0.50 -0.02 0.29 0.23 -0.06 -0.77 -0.73 0.04 ------------------------------------------------------------------ ------------------------------------------------------------------ (B) 45S-35N: 14 Aug - 19 Sep 92 (avg of ~10,000 "day"; ~12,000 "night") ------------------------------------------------------------------ ----- day ------ ---- night ----- --- day-night -- hPa V3* V4 V4-V3* V3* V4 V4-V3* V3* V4 V4-V3* 0.46 0.15 0.16 0.01 0.27 0.25 -0.02 -0.12 -0.09 0.03 1.0 0.22 0.18 -0.04 0.19 0.15 -0.04 0.03 0.03 0.00 2.2 0.38 0.41 0.03 0.17 0.20 0.03 0.19 0.21 0.02 4.6 0.36 0.33 -0.03 -0.05 -0.09 -0.04 0.41 0.42 0.01 10 0.14 0.13 -0.01 -0.10 -0.11 -0.01 0.24 0.24 0.00 22 -0.02 0.00 0.02 -0.15 -0.13 0.02 0.13 0.13 0.00 46 -0.04 -0.03 0.01 -0.11 -0.08 0.03 0.07 0.05 -0.02 100 -0.04 0.05 0.09 0.03 0.11 0.08 -0.07 -0.06 0.01 ------------------------------------------------------------------ As expected, the only significant difference (other that the factor of 0.92 for V3) is for enhanced lower stratospheric ClO in the polar vortex where V4 values are lower by ~0.2 ppbv, which is expected due to HNO3 effects. Problems with relatively large and unrealistic negative 100 hPa ClO values when 46 hPa ClO is enhanced (including day-night differences), as discussed in the validation paper for V3 data, still exist for the V4 data. The unrealistic negative night-time values of ~ -0.1 ppbv for low-mid latitude ClO at 4.6 to 46 hPa, also still exist in V4 data. SYSTEMATIC EFFECTS: There are known bias errors in the retrieved profiles of approximately -0.1 to -0.2 ppbv at 10-46 hPa, and approximately -0.05 at 4.6 hPa. These errors become evident when taking averages of large amounts of nighttime data (negative values are obtained, when ClO should be zero), and can be reduced if day-night differences can be taken. Non-linearities with respect to temperature can cause retrieved ClO values to be up to approximately 5-10% too large in the cold winter polar (especially Antarctic) vortex. CAVEATS: The retrieval for chlorine monoxide uses an a priori estimate based on a month-dependent latitude-dependent (and day/night) climatology developed by the UARS science team (2-D modeling results). Retrieved ClO values at 100 and 0.46 hPa are not considered sufficiently reliable for general use in scientific analyses, although there may be specific investigations where these can provide useful information. ClO VALUES AT 100 and 0.46 hPa SHOULD NOT BE USED FOR SCIENTIFIC STUDIES WITHOUT CONSULTING THE MLS TEAM. THE PARAMETER FILES SHOULD BE EXAMINED AND ONLY PROFILES WITH MMAF_STAT='G', 't', or 'T' AND QUALITY_CLO=4 SHOULD BE USED FOR SCIENTIFIC STUDIES. Profiles with MMAF_STAT set to 'T' (which occurs quite infrequently) may have slight biases at 46 and 100 hPa. MOREOVER, ONLY DATA HAVING POSITIVE ESTIMATED UNCERTAINTIES (QUALITY FIELD IN LEVEL 3 DATA FILES) SHOULD BE USED. PARAMETER #6: SO2 This document supplements the MLS SO2 paper by Read et al. [GRL 20, 1299-1302, 1993], which should be studied by any user of these data. CHANGES FROM VERSION 3 TO VERSION 4: Version 4 incorporates a recently-measured SO2 pressure broadening function which is 3 % larger at 250 K than the Version 3 value. Radiances for tangent pressures of 100 hPa are included in Version 4 by spline interpolation of the radiances to minimize vertical sampling gaps sometimes created in Version 3 (but a criterion to neglect opti- cally thick radiances is still in effect in Version 4). A post-launch estimate of the pointing offset between the 63- and 205-GHz radiometers is now used [this was obtained from MLS antenna scans across the Moon]; the pointing offset change from Version 3 corresponds to a tangent height change of about 430 m. Retrieved values of SO2 can also be somewhat affected by changes which have occurred in the ClO and O3_205 retrievals, as well as by changes in the spectral baseline. ESTIMATED PRECISION AND ACCURACY: UARS Estimated V4 - V3 Standard Single Profile Estimated Average Index Pressure Precision (1) Accuracy (2) Differences (3) 14 4.6 hPa 1.7 ppbv 3 ppbv and 15 % 0.0 ppbv 12 10 hPa 2.0 ppbv 3 ppbv and 15 % 0.0 ppbv 10 22 hPa 2.2 ppbv 3 ppbv and 15 % -0.3 ppbv 8 46 hPa 4.3 ppbv 3 ppbv and 15 % -2.0 ppbv 6 100 hPa 8.4 ppbv unknown -1.0 ppbv Notes: (1) Precisions (1-sigma) are typical values obtained from the retrieval algorithm and are consistent with the observed rms variability in situations where SO2 is below the instrument noise level, except at 100 and 46 hPa, where the observed rms variability is about 3.5 ppbv. The quality field in the Level 3A files gives the estimated precision of individual profiles. Precision can be improved by averaging together individual profiles. (2) Accuracies are based on a known SO2 bias of up to 3 ppbv during non- volcanically-enhanced aerosol conditions and on a 15 % scaling uncer- tainty estimate based on the ClO scaling error estimate (measurements made at nearby frequencies). Also, comparison of derived quantities such as bulk atmospheric loading and decay rates agree within 10 to 20 %, consistent with the estimated MLS accuracy for SO2. These accuracies do not include the random noise which, for a single profile, is the estimated precision. (3) Average differences (V4 minus V3 data) are derived from average profiles for selected days/regions under non-enhanced conditions. Under enhanced conditions from Mt. Pinatubo (tropics for Sep. 21, 1991), the average differences are as listed in the above Table, except at 46 hPa and 10 hPa, where the differences are -4.0 and +1.0 ppbv, respectively. SYSTEMATIC EFFECTS: Steady-state SO2 concentrations are expected to be < 0.1 ppbv globally. Zonal averages of MLS SO2 (which improves the precision significantly) years after the Pinatubo eruption give unrealistic values as high as 3.0 ppbv at some heights. This is caused by a very weak but not understood spectral feature in the measured radiances which doesn't have the proper SO2 emission shape, but can be partially fit by SO2, causing a bias in the retrieval. There are also some weak contaminant atmospheric emitters whose concentrations are unknown which also couple into the SO2 measurements. Therefore, MLS can usefully measure SO2 only when its concentrations exceed 3 ppbv, which can occur following a significant volcanic event; when using MLS SO2, one will need to subtract a non-decaying "SO2 baseline" from any MLS measurement. It is also possible and unknown (mainly because SO2 was not routinely produced as Level 3 files in the preceding versions) whether there is a seasonal variation in the SO2 baseline. SO2 is measured by the ClO radiometer and is retrieved simultaneously with ClO. Consequently, in the absence of correlative measurements, the ClO scaling uncertainty of 15% is assumed for SO2. CAVEATS: SO2 PROFILES HAVE NOT BEEN VALIDATED WITH COINCIDENT PROFILE COMPARISONS FROM OTHER EXPERIMENTS because these data do not exist [however, comparison of derived quantities such as bulk atmospheric loading and decay rates agree within 10 to 20 %]. There are large gaps in the MLS measurement track between orbits. This could severely complicate analyses of small highly localized eruptions such as the 21 April 1993 Lascar eruption which was easily detected by MLS (actually a much stronger signal than the residual Pinatubo plume measured by MLS after instrument activation); the Lascar eruption was very localized compared to the MLS sampling capability, thus making it impossible to determine the SO2 decay rate or even plume dispersion dynamics. The retrieval for sulphur dioxide uses a constant a priori estimate of 0.1 ppbv. The 100 hPa SO2 value will be very strongly biased toward 0.1 ppbv (the a priori estimate), and so will the values at pressures smaller than 4.6 hPa. Also, the 100 hPa measurements tend to be especially susceptible to minor spectral artifacts. SO2 VALUES OUTSIDE THE RANGE FROM 46 TO 4.6 hPa SHOULD NOT BE USED FOR SCIENTIFIC STUDIES. THE PARAMETER FILES SHOULD BE EXAMINED AND ONLY PROFILES WITH MMAF_STAT='G', 't', or 'T' AND QUALITY_CLO=4 SHOULD BE USED FOR SCIENTIFIC STUDIES. Profiles with MMAF_STAT set to 'T' (which occurs quite infrequently) may have slight biases at 46 and 100 hPa. MOREOVER, ONLY DATA HAVING POSITIVE ESTIMATED UNCERTAINTIES (QUALITY FIELD IN LEVEL 3 DATA FILES) SHOULD BE USED. PARAMETER #7: HNO3 ESTIMATED PRECISION AND ACCURACY: UARS Estimated Standard Single Profile Estimated Index Pressure Precision (1) Accuracy (2) 10 22 hPa 4.5 ppbv TBD 8 46 hPa 3.0 ppbv TBD 6 100 hPa 2.0 ppbv TBD (1) Precisions (1-sigma) are based on observed variability in a narrow latitude band centered around the equator, where meteorological variability is expected to be small relative to the estimated retrieval error. Because natural atmospheric variation is not completely negligible, the true precisions may be slightly better than these estimates. These precisions are generally consistent with uncertain- ties estimated by propagating the radiometric noise through the retrieval algorithm. The quality field in the Level 3A files gives the estimated precision of individual profiles. Precision can be improved by averaging together individual profiles. (2) The estimated accuracy has not yet been quantified because a sufficient number of comparisons with correlative data sets has not been made. However, a preliminary synopsis of possible systematic errors is given below. SYSTEMATIC EFFECTS: Initial comparisons (over a limited data set) with colocated UARS CLAES HNO3 observations show that the MLS HNO3 agrees well at 100 hPa but is usually 0-2 ppbv lower at 46 hPa and 0-4 ppbv higher at 22 hPa (parti- cularly in the polar regions). However, even where biases between the two data sets exist, there is good correspondence in the morphology of the CLAES and the MLS HNO3 fields. Preliminary comparisons with other correlative data sets also indicate that MLS overestimates the peak in the HNO3 profile. Under conditions of enhanced SO2 (after a major volcanic eruption), there is a strong negative bias (about 5 ppbv) in HNO3 in the equato- rial region at 22 hPa and a similar but smaller negative bias at 46 hPa. CAVEATS: The retrieval for nitric acid uses an a priori estimate based on a month-dependent latitude-dependent climatology developed by the UARS science team. The profiles contained in the Level 3A files extend from 464 hPa to 0.46 hPa, but the range of useful MLS information is primarily from 100 to 22 hPa (at other pressures, values are mostly climatological). ONLY HNO3 VALUES AT 100, 46, AND 22 hPa SHOULD BE USED FOR SCIENTIFIC STUDIES (keeping in mind the error bars and the fact that validation efforts are not as far along as for other MLS products). THE PARAMETER FILES SHOULD BE EXAMINED AND ONLY PROFILES WITH MMAF_STAT='G', 't', or 'T' AND QUALITY_O3_205=4 SHOULD BE USED FOR SCIENTIFIC STUDIES. Profiles with MMAF_STAT set to 'T' (which occurs quite infrequently) may have slight biases at 46 and 100 hPa. MOREOVER, ONLY DATA HAVING POSITIVE ESTIMATED UNCERTAINTIES (QUALITY FIELD IN LEVEL 3 DATA FILES) SHOULD BE USED. NOTE: The quality control measures outlined above are not always sufficient to filter out occasional large spikes in the retrieved HNO3 values; such remaining spikes are not yet understood and must be removed on an individual basis (by inspection).