CCSD1Z00000100000052 CCSD1R00000300000032 DELIMTER=EOF; TYPE=CCSD1F000001; CCSD1C00000400000013 ADI=NURSCL04; CCSD1R00000300000032 DELIMTER=EOF; TYPE=CCSD1D000002; UPPER ATMOSPHERE RESEARCH SATELLITE CRYOGENIC LIMB ARRAY ETALON SPECTROMETER STANDARD FORMAT DATA UNITS DATA SET CHANGE HISTORY DOCUMENT September 1997 A.E. Roche, J.B. Kumer, J.L. Mergenthaler, R.W. Nightingale and G.A. Ely Lockheed Martin Advanced Technology Center Lockheed Martin Missiles and Space Palo Alto, California A. INTRODUCTION --------------- The SFDU documentation for the CLAES data set is hierarchical (see below for other referenced SFDU's). This document describes the CLAES data set change history for each version of the catalogged data. The current data set (Level 2, Level 3AT, and Level 3AL) was produced by the CLAES Batch Data Processing Software (BDPS), version 6.06 and yielded data version 8 (V8). The V8 processing did not target improvement in all CLAES species. Species that were targeted for improvement, and that were significantly improved include N2O5, F11 and the aerosols. These have been transferred to the DAAC. The V8 species O3(B9), ClONO2, HNO3, N2O, NO2 and temperature have also been transferred to the DAAC. These are improved generically by comparison with V7 in that: 1) The V8 data have been processed for the period 21 October 1991 until 5 May 1993, when the CLAES cryogen was totally expended. The V7 data was processed on just a segment of that period, from 6 January 1992 to 5 May 1993. 2) An improved calibration approach that corrected for a trend in responsivity near missions end (e.g., see discussion by Roche et al., J. Geophys. Res., Vol. 101, No. D6, 9679 - 9710, 1996) was used in V8 but not in V7. 3) The embedded error bars in V8 agree better with independent error estimates (e.g., such as the track intersect method as described by Kumer et al., J. Geophys. Res., Vol. 101, No. D6, 9657 - 9678, 1996) The V8 species F12, CH4, and H2O have not been transferred to the DAAC. These inadvertantly regressed in V8. The source of the problems in V8 have been identified and these will be improved over V7, and therefore V8, in our next version, V9. However, at present the highest version on the DAAC for these species is V7. The latter is recommended for use pending their replacement with V9. The V8 species NO is changed considerably from V7, but has not been transferred to the DAAC, pending further validation. A list of supporting documents is given below. Standard Format Data Unit Document List: NURS1I00CL00 FileClass: Whole Data Set NURS1I00CL01 FileClass: Level 3AT Data NURS1I00CL02 FileClass: Level 3AL Data NURS1I00CL03 FileClass: Level 2 Data NURS1I00CL04 FileClass: Data Set Change History (*) [(*) Current Document] CHANGE HISTORY -------------- Keyword Description ------- ----------- DataVersion: Version 7. This is the version of the data set as stored in the UCSS catalog on the CDHF. This is the second version of CLAES data to be put into the public domain on the NASA Goddard Space Flight Center Distributed Active Archive Center (NASA-GDAAC). The species F12, CH4, and H2O are available in this version. Version 8. This is the third version of CLAES data to be put into the public domain on the NASA-GDAAC. The species O3, ClONO2, F11, HNO3, N2O, N2O5, NO2, aerosols and temperature are available in this version. SoftwareVersion: The CLAES BDPS Versions 5.01 and 6.06 produced the data versions 7 and 8, respectively. DataComments: Comments for each species in V7, summarized from the UARS Validation Workshop (19-22 September 1994) and the validation papers in the special UARS Validation issue of the J. Geophys. Res. (Vol. 101, No. D6, 1996) are listed in section C below. The changes in V8 are also listed in section C where appropriate. Table 1 CLAES spectral (blocker) filter regions, data versions, and subtypes of retrieved parameters available at highest version on the NASA-GDAAC. Spectral filter band data center and number version subtypes of retrieved parameters -------------------- ------- --------------------------------------------- 780 B9 V8 O3B9(ozone), CLONO2, AERO780 790 B8 V8 TEMP(from CO2 radiance), O3B8(ozone), AERO790 843 B7 V8 CFCl3, AERO843 880 B6 V8 HNO3, AERO880 925 B5 V8 AERO925 925 B5 V7 CF2Cl2 1257 B4 V8 N2O, N2O5_OTHER *, AERO1257 1257 B4 V7 CH4 1605 B3 V8 NO2, AERO1605 1605 B3 V7 H2O 1897 B2 V8 AERO1897 1897 B2 V7 NO * The subtype name N2O5_OTHER refers to the specie N2O5. There is an additional CLAES blocker filter region near 2843 cm-1 where data have been obtained for the future retrieval of HCl, aerosol and for properties of OH chemi-luminescence emissions. Data have been obtained in this region by use of special detectors as described by Roche, et al., JGR, 98, p10763, 1993. B. CLAES V8 ERROR ESTIMATION PROCEDURE The procedure for error estimation for the CLAES data version V7 is described in appendix A of the paper by Kumer et al. (J. Geophys. Res., Vol. 101, No. D6, 9621 - 9656, 1996). Basically, the CLAES multi-emitter multi-channel problem is broken down to the equivalent of a single emitter and single channel problem for each species i by using a linear least squares approach for fitting calculated radiances to the observed data. Next, a method similar to the Newtonian iterative algorithm (equation (99) on page 621 in the paper by Rodgers ( Rev. Geophys. and Space Phys., Vol. 14, 609-624, 1976) is used for retrieval and error estimation for each species. The error is driven by the radiance residuals, and for the multi-emitter multi-channel case this can some times be larger than would be predicted from random instrument noise alone as the result of systematic effects or undetected noise spikes, for example. The result was that in many cases the embedded V7 error estimates were too large on comparison with independently derived estimates of precision as reported in the special UARS Data Validation Issue, J. Geophys. Res., Vol. 101, No. D6, 1996. The embedded error estimates in V8 have been considerably improved. The goal was to provide embedded error roughly equivalent to the independently derived estimates of precision. In some cases the error is driven by estimated instrument noise rather than radiance residuals in order to achieve this goal. Also, a coding error in the embedded error bars for O3 (B9) as reported by Bailey et al. (J. Geophys. Res., Vol. 101, No. D6, 9621 - 9656, 1996) has been fixed. Finally, a coding error, as is described below, has been introduced that is specific to the aerosol subtype AERO790 in the V8. C. V8 RETRIEVED PARAMETER STATUS The status of the V8 retrieved parameters listed in Table 1 above is discussed in the following sections. Each discussion starts with a summary of the V7 findings, from the validation reports if published, followed with a summary of the comparison of V8 with V7. Only the V7 status is given for F12, CH4, H2O, and NO, for which there are no V8 paramters. --------------------------------------------------------------------- 1.0 TEMPERATURE (Temp) V7 temperature has been subjected to an extensive validation exercise that is documented by Gille et al. (J. Geophys. Res., Vol. 101, No. D6, 9583 - 9601, 1996). Users of this data are advised to carefully read this paper. Overall there is not a large difference between V8 & V7 data, therefore much of the detailed V7 validation information carries over to V8, except as noted. 1.1 Summary of the V7 Temperature Validation Temperature profiles extending from approximately 15-60 km are available for the period from January 4, 1992 to May 5, 1993. Comparison of many pairs of retrievals at the same location (near 32N or 32S) measured on sequential orbits (time separation 96 minutes) shows a precision ranging from approximately 0.8K at 68 hPa to about 3.5 K at 0.2 hPa, which agrees with simulations incorporating random noise and short-period spacecraft motions. U.K. Meteorological Office (UKMO) analyses show general agreement, with CLAES tending to be cooler by about 2K, except in the tropics and high-latitude winter conditions. This is supported by comparisons with individual and several lidars which indicate that agreement is within 2 K throughout the profile (except for a narrow layer around 3 mb). An error analysis also indicates that systematic errors should be roughly 2K, independent of altitude. The systematic differences at low latitudes appear to be due to tropical waves, which have vertical wavelengths too short to be seen by the TIROS Operational Vertical Sounder (TOVS) instruments. In the high latitudes CLAES is biased increasingly cooler relative to radiosondes as the temperature drops below about 190K. The bias is exacerbated by polar stratospheric clouds. 1.2 V8 Temperature Summary In V8 the number of processed days increased to 458. The processing began earlier in the mission, namely on 21 October 1991 and ran until 5 May 1993. Again there are data gaps in this period corresponding to days when the CLAES door was closed for yaw-round maneuvers, calibration, etc. For the part of the mission which overlaps, V8 temperatures are lower than V7 by between 0.5 and 1.5 K on the average in the region 20 and 50 km. Since V7 tended to also be cool, by 1-2K vs NMC and correlative data, the V8 data is be further biased. The V8 temperatures retain larger cold biases in the tropics (1-10 mb) and the southern polar winters. ---------------------------------------------------------------------- 2.0 AEROSOL (AERO1897, AERO1605, AERO1257, AERO925, AERO880, AERO843, AERO790, AERO780) Aerosol extinction coefficients have been retrieved from CLAES infrared thermal emission measurements in eight spectral regions (1897 cm-1, 1605 cm-1, 1257 cm-1, 925 cm-1 880 cm-1, 843 cm-1, 790 cm-1 and 780 cm-1), which are referred to as subtypes AERO1897, AERO1605, AERO1257, AERO925, AERO880, AERO843, AERO790, and AERO780, respectively. These coefficients are suitable for studying the evolution of the sulfate aerosol, cirrus and polar stratospheric clouds. The Version 7 aerosol precision and accuracy is discussed by Massie et al., J. Geophys. Res., vol. 101, p. 9757, 1996. This paper is highly recommended to users of the V8 data, since most of the comments apply to this data with the exception of changes noted below. 2.1 V7 Aerosol Summary The average precision for aerosol retrieved from the spectral channels from 1605-780 cm-1 varies from about 7 to 35 % for pressures from 21 to 100 hPa. The measurement accuracy for volcanic aerosol is estimated at between 20 and 35% for the 780 and 790 cm-1 channels for moderate and heavy aerosol loading. The range of pressures where the error bars indicate high quality data is between 20 and 68 mb, although this varies with aerosol loading. In this pressure range, the sulfate aerosol extinction profiles for the 790 cm-1 channel show general agreement with the ISAMS 12.1 micrometer aerosol extinction coefficient retrievals and SAGE II retrievals made at visible wavelengths that have been scaled to the infrared for this comparison. Major caveats for data use are: AERO1257 data are set to near zero for altitudes where the continuum extinction coefficients become less than 6.3E-05 km-1 to accomodate the retrieval of N2O5; AERO1605 data contains a contribution from O2 pressure-induced-absorption that can be corrected by the user; the daytime AERO1897 data contain a significant contribution from solar scattering; there is a 25-35% difference between AERO790 data and AERO780 data when theory would suggest a maximum difference of 3%. Finally, the extinction coefficients for fairly thick PSCs (extinction coefficients greater than about 1.2E-03 km-1), which occur mainly during the coldest Antarctic winter period, are set to a low extinction climatology. Work is on going to correct these problems in the next version. Sulfate aerosol spectra, when viewed as relative values of extinction derived from all channels, shows general agreement with expectations based on measured aerosol size distributions and sulfuric acid optical properties. However, there are differences in the relative values of the aerosol extinction coefficients amongst channels of up to 50%. [Massie et al., 1994] 2.2 V8 vs. V7 Aerosol Comparisons V8 aerosol retrievals show much improved detection of PSCs in the polar winters and of thick tropical cirrus. The low extinction climatology is no longer in use. If the cloud is thick, a high value is shown for the retrieved aerosol, i.e., something greater than 1.2e-03 /km, and a large error bar (greater than 100%) is assigned. There is a caveat to this general approach in the 790 cm-1 channel where the error bar is erroneously, always less than about 10%. The general trend in the 8 different cataloged aerosol products, including the two most commonly used to-date for science studies, AERO780 and AERO790, are 5 to 10% reductions in the absolute values of the extinction coefficients. AERO925 and AERO1605 show larger changes that are discussed below. AERO1897 This product has undergone little validation, but in past versions has shown diurnal variations. While it is still "noisy" and is not a high confidence aerosol product, AERO1897 data show improvement in that areas of thick tropical cirrus and PSC's are now more apparent. Otherwise, V7 and V8 are very similar. Due to the cloud detection improvement V8 is recommended. AERO1605 AERO1605 data show improvement in that areas of thick tropical cirrus and PSC's are now apparent. The O2 pia continuum has been subtracted. There is a pronounced diurnal signature above about the 20 hPa level which is apparently interference from NO2 also retrieved in this channel. Best quality is found between about 20 and 68 hPa with k_ext less than about 0.004/km. AERO1257 AERO1257 data show improvement in that areas of thick tropical cirrus and PSC's are now apparent. The magnitude of the aerosol extinction at the sulfate peak (20-28 km) is reduced 5-15% from V7. The V8 profiles are cut-off above about 10 hPa since more of the continuum in this region was included with N2O5, which also retrieved in this channel in V8. Best quality is found between about 20 and 68 hPa with k_ext greater than 6.3E-04/km and less than about 0.004/km. AERO925 AERO925 data show improvement in that areas of thick tropical cirrus and PSC's are now apparent. The magnitude of the aerosol extinction at the sulfate peak (20-28 km) has increased up to 50% from V7. At this point it appears that AERO925 V8 extinction is in better relative agreement with AERO780 and AERO790. Due to the cloud detection improvement V8 is recommended. Best quality is found between about 20 and 68 hPa with k_ext less than about 0.002/km. AERO880 AERO880 data show improvement in that areas of thick tropical cirrus and PSC's are now apparent. The magnitude of the aerosol extinction at the sulfate peak (20-28 km) has decreased up to about 10% from V7. Due to the cloud detection improvement V8 is recommended. Best quality is found between about 20 and 68 hPa with k_ext less than about 0.002/km. AERO843 AERO843 shows improvement in that areas of thick tropical cirrus and PSC's are now apparent. The magnitude of the aerosol extinction at the sulfate peak (20-28 km) has decreased up to about 5% from V7. Due to the cloud detection improvement V8 is recommended. Best quality is found between about 20 and 68 hPa with k_ext less than about 0.002/km. AERO790 AERO790 shows improvement in that areas of thick tropical cirrus and PSC's are now apparent. The magnitude of the aerosol extinction at the sulfate peak (20-28 km) has decreased up to about 5-10% from V7. Due to the cloud detection improvement V8 is recommended. Best quality is found between about 20 and 68 hPa with k_ext less than about 0.002/km; error bars too small in thick clouds. AERO780 AERO780 shows improvement in that areas of thick tropical cirrus and PSC's are now apparent. The magnitude of the aerosol extinction at the sulfate peak (20-28 km) has decreased up to about 5-10% from V7. Due to the cloud detection improvement V8 is recommended. Best quality is found between about 20 and 68 hPa with k_ext less than about 0.002/km. ------------------------------------------------------------------------- 3.0 OZONE (O3B9) Ozone is retrieved from CLAES radiances in spectral channels 8 (O3B8) and 9 (O3B9), with the latter (O3B9) being the preferred product. Version 7 O3B9 data products have been subjected to an extensive validation effort as documented by Bailey et al., (J. Geophys. Res., vol. 101, p. 9737, 1996), and Cunnold et al., J.Geophys. Res., 101, 10335, 1996). The present documentation starts with a summary of the V7 validation findings taken mostly from the Baily et al. 1996 paper and follows with a summary of the comparison of V8 with V7. A brief summary of the characteristics of O3B8 is given for completeness, but O3B9 remains the recommended product. 3.1 Summary of V7 Ozone Validation The CLAES V7 ozone data set provides global coverage from 80S to 80N for 388 days. Approximately 1200 profiles are obtained each 24 hour period for each constituent over a nominal altitude range of 120 to 0.1 mb (15 to 65 km). To arrive at estimates of systematic error we have (1) compared CLAES profiles for both tracers with a wide variety of correlative data from ground-based Lidar and microwave instruments, balloon instruments , and space-borne sensors including SAGE, SBUV, and ATMOS; (2) carried out empirical estimates of experiment systematic error based on knowledge of instrument characteristics. In most cases close coincidence in latitude (+/-2 deg), longitude (+/- 10 deg) ,and time (+/- 6 hours) was available with correlative data over the CLAES altitude range. The majority of correlative ground and balloon data were confined to mid-northern latitudes, typically near 34N. To arrive at estimates of random error we: (1) looked at the variability in the CLAES profile data for multiple crossings of the 32 degree latitude turnaround points; ( 2) made empirical estimates based on knowledge of instrument characteristics; ( 3) made predictions based on an independent retrieval algorithm (4) compared with the production algorithm-embedded errors. The measurements have also been examined for spatial and temporal distributions and variations known to exist in historical data. Overall, the CLAES data have been shown to agree well with expectations. Regarding data precision, that estimated from the measured profiles has been shown to agree with empirical error budget analyses and predictions based on independent retrieval of simulated radiances. The overall tendency is for the estimated precisions to exceed the observed variances in the 80N and 32S latitude turnaround-point data. The estimated random errors give reasonably close agreement with the data above about 10 mb: the agreement is best above about 5 mb, where the system noise dominates the estimates. The estimated precision is most pessimistic in comparisons with the 80N observations below 10 mb where there is approximately a factor of three difference. Observation and estimate are in better agreement below 10 mb for the 32S case and actually come into good agreement below about 40 mb. The V7 algorithm-generated profile error estimates (i.e., cataloged errors) are higher than observations or empirical estimates by a factor of 3 to 4. (See Cunnold et al. 1996, Fig 4c). An inadvertent systematic component which made the embedded error a factor of 2.4 too high was identified and rectified in V8. Comparison of individual coincident profiles confirm that the overall profile shapes are represented well by CLAES, and that the shapes vary as expected with season and latitude. The ability of CLAES to observe and retrieve small scale and atypical vertical structure to within its resolution limits is also confirmed by these comparisons. Regarding absolute accuracy, in the altitude range of the peak in the ozone mixing ratio profile (~2 - 10 mb), the CLAES measurements are in almost all cases 10-20% higher than those reported by other measurement systems. At the upper altitudes (~0.2 - 1.0 mb), CLAES is generally lower than the correlative measurements with the differences becoming larger with increasing altitude. Around 1.0 mb, the differences are typically of the order of 10% below other measurements. At 0.2 mb, the differences can be as large as 50% lower. For altitudes between 10 and 20 mb, CLAES is in many cases lower than the correlative measurements by about 20%. At the very lowest altitudes, the differences are very dependent on location, time, and the correlative measurement technique being compared. There is some indication that suggests that uncertainties due to aerosol may affect the results for CLAES and, in some cases, the correlative data as well. This is particularly true for observations prior to July 1992. The statistics of the correlative data comparisons show that, globally, CLAES tends to have a positive bias at 100 mb of the order of 20 - 30% , a slight negative bias at 46 mb, almost no bias at 10 mb, and overall negative biases of -10% at 1 mb and -25% at 0.46 mb, respectively. Apart from the 100 mb surface, the observed biases are in reasonable agreement with empirically predicted ranges, suggesting that the systematic uncertainties estimated for the various calibration, instrument, and a prior components are also reasonably representative of the actual behavior of the experiment. The larger-than-predicted bias apparent in the correlative data comparisons at the 100 mb surface, may well reflect residual interference in the CLAES retrievals from the Pinatubo tropical sulfate layer. Cunnold et al., 1996 also indicate that vs SAGE II CLAES are ~15% high at 46 mb in the tropics, and ~20% low near 0.32 mb. Time series comparisons with ground based lidar and microwave instruments at fixed locations show good representation of the seasonal variability at various altitudes. Day-to-day variations as seen by CLAES are somewhat larger than observed by the ground-based systems. The differences are largest at higher altitudes where the CLAES signal to noise ratio is lowest. However, some of the difference is also due to lack of exact spatial coincidence. Comparisons with microwave observations at high altitudes (0.147 mb) under day and night conditions indicate that CLAES senses the diurnal ozone variation but indicates a smaller amplitude with less precision. At these altitudes, the signal to noise ratio for CLAES becomes a problem and the retrieval may be over constrained. When viewed on larger scales, the CLAES data display the characteristics that have been noted in previous ozone climatologies. The monthly zonal mean cross sections exhibit vertical and horizontal gradients that are consistent with past observations. The seasonal variations as observed by CLAES are also consistent with historical data when compared at several different altitudes and latitudes. 3.2 V8 Ozone Comparisons Overall, based on correlative data comparisons and global statistics, V8 O3B9 appears either very similar to or slightly better than that of V7, the judgment depending somewhat on altitude and latitude range. The factor of 2.4 numerical error in the algorithm-generated profile errors has been eliminated. V8 O3B9 vmrs at extratropical latitudes, increase between 5 and 10% at and above about 10 mb, show little change between about 15 and 20 mb, and show a general decrease of 5-10% below 20 mb. In the tropics larger decreases, up to 25%, are seen below 20 mb. These changes tend to bring the CLAES data into better agreement with global correlative data, especially at low altitudes in the tropics where V7 data were high by 10-20% vs SAGE, and still higher vs MLS. They also improve agreement with correlative data above 2 mb where V7 was generally low, especially at altitudes between 1 and 0.4 mb where V7 differences between 10 and 25% have been reduced to under 10%. Statistically the mean difference near the peak of the ozone profile appears to have improved somewhat for V8. For example versus about 220 ATMOS profiles between 50S and 28N, V8 shows a maximum 10% difference between 5 and 10 mb compared with 20% for V7. Many individual profile comparisons, however, show that V7 was already somewhat high near the peak, and for these cases V8 will increase the discrepancy. Near 46 mb at extratropical latitudes V7 was low by about 5% and this difference will increase to 10-15% with V8. The zonal mean morphology and seasonal behaviour show little change. 3.3 Comments on O3B8. O3B8 mixing ratios in both software versions, V7 and V8, are systematically high by 10-15% wrt O3B9 and correlative data above about 10 mb in the winter and spring of 1992 and low by the same amount starting in August of 1992, contiuning into May of 1993. Both versions are systematically low below about 20 mb by as much as a factor of 2. This low bias is somewhat worse in V8 vs V7. There is some indication that the altitude of the peak mixing ratio for O3B8 is lower than that of O3B9 by the order of 1 km, and the O3B8 altitude profiles show more anomalous vertical structure. In addition, the O3B8 fields appear noiser than those of O3B9 throughout the data set. Apart from the biases in absolute quantities, the O3B8 zonal mean fields do exhibit similar seasonal and interranual morphology to that seen in the O3B9 fields and in correlative data. There is an on-going effort to improve the O3B8 data. However, as noted earlier, O3B9 has been shown through extensive validation and analysis to be a scientifically superior and self consistent product, and it's exclusive use is recommended. ------------------------------------------------------------------------- 4.0 WATER VAPOR (H2O) 4.1 V7 H2O Summary Only The utility of the CLAES V7 H2O data is limited in several ways. First there are non-LTE daytime enhancements of the order 1 to 2 ppmv. Therefore, day time data have this known deficiency. Secondly 'run-away profiles' of H2O are retrieved in polar winter conditions of a hot low altitude mesopause, and cold high altitude tropopause. Finally, the night time CLAES values for the more benign atmospheric cases are typically the smallest of the UARS measurements, suggesting some caution in their use. Best agreement between night time CLAES and other UARS sensors occurs below the 10 mb level where differences vs HALOE are of the order + and - 10 %. Agreements in this altitude range with mid-latitude correlative data are of the order + and - 15 %. Values in the tropics at these altitudes are less than any other UARS sensor with minima some times less than 3 ppmv in the January 1992 and April 1992 validation periods. At higher altitudes, between 10 and 1 mb, CLAES measurements are 10 to 20 % less than correlative data, and 10 to 30 % less than MLS values with limited regions some times less still (high latitudes in January and tropics in April). At 1mb and above the CLAES values agree better with correlative data, and the other UARS sensors. For the data that might be of use, namely, nighttime data tropical to mid-latitudes, and summer high latitudes, these comparisons suggest 20 to 30 % systematic error, and precision of about 0.5 to 1.0 ppmv. The precision derived from mid-latitude turn around data is typically of the order of 0.2 to 0.7 ppmv from about 40 mb to 2 mb, and tending towards 1 to 2 ppmv above and below this altitude range. This is somewhat less than indicated by the reported error bars, probably due to the systematic component of these as discussed in paragraph B above. In the nighttime the CLAES data can be used on the range from the tropopause on up to altitudes higher than 1 mb, subject to the limitations discussed above. Within the above limitations the CLAES data show reasonable qualitative agreement with climatology and with the other instruments with respect to zonal mean latitude pressure, and longitude-pressure structure. And, especially in comparison with MLS, time track profiles show good consistency over many orbits. Given this, and the observation that we see good day-by-day repeatability in the zonal mean and profile data, corroborates our estimates of precision given above. A general statement based on the UARS Validation Workshop meetings held to date is that the CLAES version V7 H2O retrievals have deficiencies that might make them difficult to use for some scientific investigations. The main issues to be addressed in future retrieval versions include, the high values in the polar winter regions, the bias (low) wrt correlative data, day-night differences, apparent spiking in some of the high latitude data, and some low level dependence on initialization. ------------------------------------------------------------------------ 5.0 NITROUS OXIDE (N2O) AND METHANE (CH4) Nitrous Oxide (N2O) and Methane (CH4) version 7 data products have been subjected to an extensive validation effort as documented by Roche et al., (J. Geophys. Res., 101, 9679, 1996). Overall, as noted below, there is little substantive difference between N2O V8 and V7 quantities. On the other hand V8 CH4 is systematically larger than V7 CH4, its use is not recommended, and it has therefore not been moved to the GDAAC. 5.1 Summary of V7 CH4 and N2O Validation The CLAES V7 N2O AND CH4 data sets provide global coverage from 80S to 80N for 388 days in the period from January 9, 1992 to May 5, 1993. Approximately 1200 profiles are obtained each 24 hour period for each constituent over a nominal altitude range of 120 to 0.1 mb (15 to 65 km). To arrive at estimates of systematic error we have (1) compared CLAES profiles for both tracers with a wide variety of correlative data from ground-based, rocket-, aircraft-, balloon-, and space-borne sensors, and (2) carried out empirical estimates of experiment systematic error based on knowledge of instrument characteristics. In most cases close coincidence in latitude ( +/-2 deg), longitude (+/- 10 deg) ,and time (+/- 6 hours) was available with correlative data over the CLAES altitude range. The majority of correlative ground, rocket, and airborne data were confined to mid-northern latitudes, typically near 34N, in the spring and fall, the exception being the mm-wave N2O data at 76oN in February-March 92. The ATMOS experiment provided much more extensive latitudinal coverage, i.e., 54S to 31N, although confined to approximately 7-day periods in the spring of 1992 and 1993. To arrive at estimates of random error we: (1) looked at the variability in the CLAES profile data for multiple crossings of the 32 degree latitude turnaround points; (2) made empirical estimates based on knowledge of instrument characteristics; (3) compared with the algorithm-embedded errors. N2O profile errors: CLAES N2O is within + 14.7% and - 13% of the correlative data, over the range 68 to 1.5 mb. On average, compared with correlative data, the CLAES values are about 7% high below 6.8 mb and about 10% low above 6.8 mb. Empirical estimates for N2O systematic error lie between 15 and 22% for the altitude range 68 to 1 mb, in reasonable agreement with the correlative data comparisons at all altitudes. As was the case for CH4, there were very few correlative comparisons below 20 mb in the tropics. The observed N2O variability ranges from about 20 ppbv to 5 ppbv (7 to 15%) from 68 to 2 mb, in good agreement with empirical and algorithm-generated estimates. Overall, useful information is contained in the N2O individual profiles from 68 to 1 mb, but our best confidence range is from 46 to 2 mb, extending to 68 mb at extratropical latitudes. Over this range we assign a total error (systematic +random) of < 17% and precision of 20 to 5 ppbv or about 7% over the mid stratosphere. CH4 profile errors: The results of these validation exercises indicate that CLAES CH4 is on average within 15% of the correlative data, with a positive bias, over the altitude range 68 to 0.68 mb. The empirically estimated systematic errors for CH4 are of the order of 14% from 46 to 0.46 mb, in good agreement with correlative comparisons but increase to 47% at 68 mb, while correlative comparisons continue to indicate better than 15% agreement at this level. This would suggest that the empirical estimates are overly pessimistic for CH4 at the lower levels. We also noted that there were very little correlative CH4 data below about 20 mb in the tropics. Data precision lies between 0.04 and 0.08 ppmv (5 to 15%) from 30 mb to 0.46 mb, in reasonable agreement with both the empirically estimated and the processing algorithm-generated errors. At lower altitudes, the observed variability drops to unrealistically low values as the data begin to be influenced by climatology. The algorithm estimated errors of 0.15 ppmv (9%) at 68 mb, and 0.08 ppmv (5%) at 46 mb are considered more representative of the CH4 variability for low-altitude and midlatitude situations. Overall, although there is useful information in the CH4 individual profiles from 70 to 0.1 mb, our best confidence lies between 46 mb and 0.46 mb, extending to 68 mb at mid and higher latitudes. Over the high confidence range we assign a total error (systematic + random) of < 17%, and precision between 0.08 ppmv and 0.05 ppmv or about 7% over the mid stratosphere. N2O/CH4 global field analysis: The CLAES daily zonal mean fields exhibit overall agreement with the major morphological and seasonal features seen in previous global field data (mainly SAMS) and predicted by recent models, including the steep descent of the mixing ratio isopleths towards the winter poles, the double-peaked latitudinal structure at high altitudes near the vernal equinox, and the southern midlatitude "surf-zone" in the austral winter. Several morphological features were pointed out for situations for which there have been essentially no previous measurements. These include: the differential behavior of the tracer isopleths near and inside the Antarctic polar winter vortex where the CH4 isopleth slopes are seen to increase with respect to those for N2O poleward of about 50S; "local" maxima in the tracer fields in the tropics (20S-20N), visible in CH4 prior to April, 1992 between 12 and 6 mb, and in N2O prior to August, 1992 between 40 and 20 mb. The origin of the isopleth differential behavior is not currently obvious, but may be associated with latitudinal gradients in the velocities of descent of the tracers close to and inside the vortex. The local maxima, whose features are confined to the tropics and diminish with time are likely to be associated with the Mt. Pinatubo sulfate aerosol layer, especially the N2O feature whose altitude is in the vicinity of the layer maximum. The CH4/N2O scatter diagrams generally show compact linear relationships for N2O VMR>50 ppbv for extratropical cases as previously observed (mainly in aircraft and balloon data), but show curvature for tropical latitudes for both high and low sulfate aerosol loading. Previous measurements contain little tropical data, but 2-D models indicate some curvature in the relationship in the tropics. In conclusion, we have shown that the CLAES V7 N2O AND CH4 altitude profiles are in agreement with correlative data within 15%, that the profiles for both gases have precisions in the low to mid stratosphere of the order of 7%, and that the global mean fields are correlated and exhibit the morphological and seasonal features seen in previous data. Although there are clearly areas for retrieval-algorithm improvement, these results indicate that the version 7 data are of good overall reliability and can be used with good confidence for quantitative and qualitative studies of stratospheric and lower-mesospheric structure and dynamics. 5.2 N2O Comparison V8 vs V7 The V8 zonal mean isopleth morphology and spatial structure have changed very little from those of V7 for all 4 seasons and most latitudes, including the southern polar winter region. N2O V8 profiles show little change wrt V7 below about 40 mb. Between 40-5 mb (~23-37 km) the V8 N2O profiles show enhanced VMRs with differences of ~0-10% in the tropics (~20S-20N) and ~10-20% in the mid latitude region. The V8 mean differences with the ATMOS correlative data are also larger by ~10-20% than those of V7 in this altitude range. Since the overall V7 mean percent differences with the correlative data over this range showed CLAES to be larger by 2-15%, the V8 N2O data have even larger differences. This increase between 40 and 5 mb is most likely due to a cooling of retrieved temperatures in this altitude range in the V8 data and will be corrected in V9. Above 5 mb V8 vmr show increases between 5 and 15% with the largest increases at midlatitudes. This improves agreement somewhat with correlative data at high altitudes (4-1.5 mb) where V7 was low. ------------------------------------------------------------------------ 6.0 CHLORINE NITRATE (ClONO2) Version 7 ClONO2 has been subject to an extensive validation exercise documented by Mergenthaler et al. (J. Geophys. Res., Vol. 101, No. D6, 9603 - 9620, 1996). Overall there is not a large difference between V8 and V7. Therefore, much of the detailed V7 validation information carries over to V8, except as will be noted in the V8 to V7 comparison discussion. 6.1 Summary of V7 ClONO2 validation Volume mixing ratio profiles extending from approximatey 15-35 km are available for the period from January 9, 1992 to May 5, 1993. The CLAES sampling and the morpology of ClONO2 measured by CLAES are presented graphically by Roche et al. (J. Atmo. Sci, Vol. 51, no. 20, 2877-2902, 1994). CLAES measurements provide the first near-global dataset of this stratospheric species. Data quality has been evaluated through: (1) an analysis of estimated uncertainties and biases in the remote sensing process, (2) comparison with 2D chemical model calculations, (3) comparison with correlative data, and (4) an examination of various known limitations. The precision of CLAES CLONO2 volume mixing ratio (VMR) retrievals are within 15% in the pressure range 10 < P < 50 mb. The upper limt on estimated systematic error is 28% in the range of 10 < P < 100 mb based on studies of error sources in midlatitude retrievals. The major source of systematic uncertainty is ClONO2 spectral parameters. The global distribution of ClONO2 computed with the Lawrence Livermore National Laboratory two-dimensional stratospheric chemistry model and the CLAES measurements agree qualitatively. However, above the profile peak the calculated concentration frequently exceeds the measurement. CLAES and ATMOS measurements show relatively good mid-latitude agreement suggesting that the major source of discrepancy is in the model. A possible explanation in terms of a missing reaction ClO+OH ->HCl+O2 is suggested. Also, the ClONO2 diurnal cycle constructed from more than 30 days of CLAES data agrees well with the model. The CLAES ClONO2 data differ from correlative data acquired on flights of the shuttle-based ATMOS, and balloon-borne instruments by less than 25% on the average in the 10 < P < 50 mb range. At altitudes above 10 mb the CLAES measurement is biased low with respect to correlative measurements. This discrepancy at high altitudes is consistent with the analysis showing a large increase in the systematic errors above 10 mb. Heavy tropical volcanic aerosol from the Mt. Pinatubo eruption in June, 1991 apparently interfered with ClONO2 retrievals in the period before July, 1992, causing anomalous peaks in the 20 < P < 30 mb region accompanied by very small concentrations below the peak (P > 30 mb). A similar effect associated with thick polar stratospheric clouds has been identified. Overall, this validation study indicates that the majority of these data are of good quality and should be very useful in quantitative and qualitative chemical studies of the stratosphere. 6.2 ClONO2 Comparison V8 with V7 Compared with V7, the V8 ClONO2 shows a relatively uniform decrease between 5 and 10% from about 46 to 10 mb at extratropical latitudes and up to 25% decrease near the equator (15S - 15N). The equatorial change is reduced to about 5% by March 1993. Outside the equatorial region, the zonal mean morphology and seasonal variation shows little change from V7. There is some indication of a slight (5%) increase in the global values after March, 1993, which is likely due in part to a re-calibration which removed a downward responsivity trend between August 1992 and April 1993. The reduced values cause V8 to be in somewhat better agreement with the ATMOS data, for which we have 30 coincident profiles between 54S and 28N in 1992 and 1993. The CLAES-ATMOS global mean difference is about 5% for V8 vs 10% for V7 in the region of peak ClONO2 values (25 to 10 mb). Compared to the much sparser correlative data from balloon-borne instruments (5 mid-latitude, 2 high-latitude coincident profiles, all northern hemisphere), since V7 data was already on the average 15% low at the peak, V8 data moves lower still. There are two significant improvements in the V8 data. The first is that there are far fewer spikes jumping to very high VMR values than with V7. The second improverment is the reduction of the sharp peak in VMR that appears above the reservior of Pinatubo aerosol in the pre-July 1992 data and has been described in the validation paper. In summary, the version 8 data present improvements over the version 7 data with respect to most of the correlative data and the removal of artifacts. Therefore, we recommend the use of the version 8 data over that of version 7. ----------------------------------------------------------------------- 7.0 NITROGEN DIOXIDE (NO2) As noted below, NO2 shows good day by day consistency, exhibits predicted diurnal behaviour, and is closely correllated in spatial/temporal structure with the ISAMS instrument on UARS, also measuring NO2 in emission for the same air mass as CLAES. The CLAES V7 NO2 vms are however systematically low vs LIMS and ISAMS V10. Comparison for Jan 9 1992, and discussion by Reburn et al (J. Geophys. Res. Vol. 101, No. D6, p9873-9895, 1996) indicate that at the region of the maximum in NO2 the CLAES V7 NO2 is 30 % less than both LIMS and ISAMS V10 in daytime conditions, and is 14 and 28 % less than LIMS and ISAMS V10, respectively, in nighttime conditions. In computing these biases the 10 % reduction in ISAMS V10 recommended by Reburn, and the reductions in LIMS as the result of the recent re-analysis of LIMS by Remsberg et al (J. Geophys. Res. Vol. 99, pages 22965-22973, 1994) have been included. The CLAES V8 NO2 is increased vs V7 by up to 20% near the altitude peak, but this still leaves V8 with an overall low bias. Although the CLAES team has not yet published a specific validation paper for NO2, several authors have used the data for scientific analyses and have described approaches to dealing with the systematic bias (.e.g. Dessler et al., JGR, 101, 12515, 1996, and GRL, 23, 339, 1996). Significant improvement in mitigating the systematic error in NO2 is a major focus of the next software build (V9). 7.1 V7 NO2 Summary NO2 is being retrieved at all latitudes between 80S and 80N and is reported on the nominal pressure range from about 100 mb to 0.1 mb. In most cases we have fair confidence on the range from about 20 to ~ 0.3 mb for night time data, and typically up to about 2 mb for day time data. Our confidence in this range is based mainly on comparison with daily zonal mean crossections in comparison with LIMS and climatology. There are also some limited profile comparisons as discussed in section 9.2.2.1 below. Direct comparison with solar occultation instruments such as SAGE are limited in value, but are consistent. The comparison with other UARS instruments, including (a) ISAMS, confirms reasonable diurnal, seasonal and regional structure, and (b) HALOE, the same, although the comparison here has the same problem as with SAGE. In general the CLAES NO2 data (i) show good resemblance to climatological zonal mean structure from ~ 0.3 to > 100 mb, (ii) are approximately 20 to 30 % less than climatology that is largely based on the LIMS data, (iii) have good day to day consistency in zonal mean maps, (iv) show physically realistic diurnal dependence, (v) show no apparent aerosol degradation, and (vi) are insensitive to apriori. Zonal Means: General zonal mean structure is in reasonable agreement with LIMS and/or climatology, however, the CLAES values are smaller than LIMS by ~ 20 to 30%. There is good day to day consistency in the CLAES zonal mean maps. Day and night maps show the expected diurnal variation. There is no apparent aerosol degradation. There are features in the CLAES south polar zonal means that are not directly comparable with LIMS data. Profiles: For this validation exercise there were available two directly comparable data sets from the FIRS-2 and BLISS. CLAES data were within a few % of the BLISS data, and were considerably smaller than the FIRS-2 data. For that case scaled LIMS data more closely resembled CLAES than FIRS-2. Time track comparisons: The CLAES and ISAMS data are highly correlated on time track comparisons. The ISAMS data are ~ a factor x 1.8 greater than CLAES, while CLAES is of the order 20 to 30 % less than corresponding LIMS comparisons. The time tracks show the expected diurnal and latitudinal variations. Error discussion: Based on the comparisons cited above we believe conservatively that CLAES NO2 systematic error is of the order 30% for all cases except perhaps polar winter conditions involving large vertical temperature gradients. In atmospherically benign regions, including just about all conditions except polar winter, we believe the algorithmic error estimates represent a pessimistic case, typically of the order 0.5 to 1.0 ppbv near 46 mb, ~ 10% of the reported value near the peak, and indicating significant results to levels well above 1 mb in most night time cases. In the polar winter cases involving large vertical temperature gradients the error can become very large, especially at altitudes below the stratopause. The problem is associated with large temperature gradients between a relatively high altitude tropopause and low altitude stratopause. For example, zonal mean temperature changes by > 70 K, i.e., from < 196 K to > 266 K, on going from 32 to 47 km (i.e., ~ 10 to 1.0 mb) at 76S on 8/23/92. For this example errors of the magnitude of the returned data are reported in the CLAES data below 32 km. Degradation in the accuracy begins at ~ 44 km. 7.2 NO2 V8 vs V7 Comparison The NO2 V8 retrieval algorithm was not targeted for specific improvement, although it was included in the improvements in radiometric calibration, forward radiance calculation, and radiance fitting procedures generic to all product retrievals in the V8 software. As a result of these changes, and also because the V8 temperatures cooled by 1-1.5 K, the NO2 values increased by about 10-20% vs V7, which improves agreement with correlative data (mostly LIMS and ISAMS). V8 has not changed the situation re the large errors in polar night winters. ------------------------------------------------------------------------ 8.0 NITRIC ACID (HNO3) Nitric Acid version 7 has been subject to an extensive validation exercise that is documented by Kumer et al. (J. Geophys. Res., Vol. 101, No. D6, 9621 - 9656, 1996). It is very important for users of the data to carefully read this paper. There is not a large difference between V8 and V7, therefore, much of the detailed V7 validation information carries over to V8, except as will be noted in the V8 to V7 comparison discussion. 8.1 Summary of the V7 HNO3 Validation The most interesting science aspects of the CLAES HNO3 data are two-fold. First, they provide a global view of the autumn - winter - spring evolution of the de-nitrification event at south high latitude, an essential element of the dramatic ozone reduction in that region in the spring. Second, the data were obtained during a period when the atmosphere is recovering from effects of the heavy Pinatubo aerosol loading, and they lend critical insight on the relative magnitude of the transient aerosol effects on comparison with other, more permanent mechanisms for year to year variability, the QBO for example. In the V7 validation report we have examined the quality of these data. These have been compared with correlative vmr profiles, LIMS data, column data time series and some published mapped data. Accuracy and precision are inferred from the comparisons and these are consistent with estimates that are based on instrument and retrieval characterization. One year deseasonalized trends with considerable hemispheric asymmetry are strikingly evident in the data. These are quantitatively supported by correlative data in all cases in which these are available for suitable time periods in a given location. These are documented and credible trend mechanisms are established. Profile comparisons: These were made with concurrent correlative measurements obtained by ATMOS in late March - early April 1992 and in early April 1993, 11 profiles obtained by various instrumentation on mid-latitude balloon (approximately 35N), and interspersed in time near the equinoxes, spring and summer, and 6 northern (approximately 68N) winter profiles from balloons. In 1992 there were 17 sunset ATMOS profiles on the range from 31.8S to 55.5S, 21 sunrise profiles from 16.7S to 18.2N, and in 1993 only the 14 sunset profiles were concurrent and these were from 27.4S to 49.5S. Registration for the balloon profiles and the ATMOS 92 sunset case is within 0.4 km. The registration offset is larger for the ATMOS 92 sunrise and ATMOS 93 sunset cases in which many fewer profiles were processed to low enough altitude to be used for registration purposes. Good peak vmr agreement is achieved in all cases. The CLAES values tend to be less than those for the correlative data for cases where the CLAES peak vmr is < or = approximately 8 ppbv. In all these cases, namely ATMOS 92 sunrise, ATMOS 93 sunset and mid-latitude balloons, agreement within variability is just missed. Conversely, for CLAES peak vmr > or = approximately 10 ppbv, the tendency is for the CLAES values to be slightly larger than correlative, but to agree within variability. At both 70 and 10 mb the CLAES vmr agrees within variability with the balloon measurements. For the most part agreement within variability is achieved with ATMOS, the variabilities are larger than in the case of the balloon comparisons. From 10 to 3 mb the large majority of the data points from midlatitude balloons agree within the combined instrumental error estimate, the majority of comparisons with ATMOS were within variability, and there were no high latitude balloon data in this altitude range. There were 5 midlatitude balloon profiles down to 100 mb and on the average CLAES was less by 24 + or - 35 %. None of the midlatitude profiles went lower than 119 mb. At high latitude it was possible to compare 3 profile pairs at 100 mb and there CLAES was less on the average by 33 + or - 10 %. Zonal mean vmr comparison with LIMS: These comparisons show good agreement in overall structure and in subtle features such as polar winter enhancements above 10 mb and polewards from 50 degrees latitude. These comparisons for January and April showed LIMS values in the tropics and south in much closer agreement with the 1993 case. Closer agreement between the LIMS and the 1993 CLAES, rather than the CLAES 1992, is predicted by Gray and Ruth, (Geophys. Res. Lett. Vol. 19, No. 7, 673-676, 1992) on the basis of QBO phasing. As in comparison with some of the correlative profiles above, the CLAES top and bottom side vertical gradients are steeper than LIMS. Vmr accuracy and precision: Our highest confidence for the CLAES HNO3 V7 vmr is on the range 70 to 3 mb, with comparison with correlative data as summarized in table 2 in the validation paper. Although the few comparisons at 100 mb are also very encouraging, caution should be exercised for use outside this range. The precision on the range 70 to 3 mb is of the order 0.3 to 1.0 ppbv. This precision was derived from data repeatability. The embedded error estimates and the estimates based on instrument and retrieval characterization agree within a factor 2 or better. Zonal mean column comparisons: Comparisons over the time period of the CLAES V7 processed data with long time series column data reported by by David et al. (Geophys. Res. Lett., Vol. 21, No. 11, 1003-1006, 1994) and by Koike et al. (Geophys. Res. Lett., Vol. 21, No. 7, 597-600, 1994) at 20N and 45S respectively, were in agreement within data variability. Observations of deseasonalized trends, and mechanisms for these: The CLAES V7 data period of processed data included 9 January through 15 April of both 1992 and 1993. Data obtained in a group of days, or on a day, within the 1992 period were compared with data obtained in the corresponding group of days, or day, within the 1993 period to examine one year deseasonalized trends in the data. There is an obvious trend for the southern hemispheric and equatorial HNO3 to decrease in going from 1992 to the corresponding time in 1993. The decreasing trend is not globally universal, it is reversed for north mid and high latitudes. The decreasing trends in the south are larger than increases in the north so that the global average is a decrease. We believe the mechanism for the global averaged decrease is the result of the diminishing influence of heterogeneous conversion of N2O5 to HNO3 as the Pinatubo aerosol settles out during this time period, and the HNO3 recovers towards pre-Pinatubo conditions (designated the 'recovery from Pinatubo aerosol cloud, rfPac mechanism'). We believe the increase in the north is the result of hemispherically asymmetric QBO like effects that are strong in the northern hemisphere and weak in the southern hemisphere, and are phased to produce an increase in HNO3 over the this time period of just the right magnitude to more than offset decrease due to the rfPac. We believe this interpretation is supported by characteristics of the MLS ozone column data in this time period as were reported by Froidevaux et al. (J. Atmo. Sci., Vol. 51, No. 20, 2846-2866, 1994). Various authors of theoretical studies have predicted QBO HNO3 effects of the required magnitude. 8.2 Comparison HNO3 V8 with V7 Compared with V7, in the 46-15 mb peak region V8 HNO3 shows small increases (< 5%) for vmr < 6 ppbv, and between 5 and 10% increase for vmr > 9 ppbv. Values much in excess of 9 ppbv occur at the profile peak mostly in winter high latitudes, so that compared with V7, V8 will bias the peak winter values high, poleward of about 50 degrees, by 5-10% compared with lower latitudes. Other than this the zonal mean morphology is essentially unchanged. There is also no significant change in the seasonal characteristics of the data. V8 is in better agreement with most of the correlative balloon data, which is composed mainly of midlatitude values in the 6-8 ppbv range where V8 data begins to show small increases (V7 data were biased low wrt to these balloon data). Compared with ATMOS data, which is weighted towards smaller values of vmr, there is almost no difference between versions 7 and 8, both showing about + and - 10% difference in the peak region. Above 10 ppbv, however, CLAES V7 generally exceeded the correlative data (mainly high latitude winter 1992 ballon profiles) by 10 to 20%. The V8 data therefore for these situations will move farther away from the correlative data. The V8 zonal mean column, which now includes the extra days from earlier in the mission, maintains good agreement, approximately the same as V7, with the 20N David et al. data, and with the 45S Koike et al. data for all but the south look of 11 June to 10 July. For that south look (most notably on the days 13, 14, 30 June and 1 July) the V8 are considerably larger than the Koike et al. data and do not agree as well as did the V7. For these problem days we have executed runs with updated retrieval software and have obtained good agreement. This indicates the problem is with V8 in these special cases of very large HNO3 in the high- to mid- southern latitudes. For now we recommend V7 for these cases. Our next HNO3 data version, V9, will incorporate the software improvements and will be the recommended version for all cases including the high- to mid-southern latitudes during the problem south look period. ------------------------------------------------------------------------ 9.0 DICHLORODIFLUOROMETHANE (CF2Cl2 or CFC-12) Although there are some improved morphological features in the V8 CLAES CF2Cl2 data, such as the less evident upwelling artifact noted in the polar winter vortices, the V8 volume mixing ratios have increased substantially with respect to those of version 7. The V8 CF2Cl2 data are systematically 15 to 40% too high. This, in turn, introduces substantial biases with respect to the correlative data. We recommend that V8 CF2Cl2 data not be used for quantitative studies, and instead, that V7 CLAES CF2Cl2 should continue to be used. V8 CF2Cl2 has not been transferred to the GDAAC. 9.1 CF2Cl2 V7 Summary Only Below is a summary of V7 CF2Cl2 validation findings taken from the paper entitled "Global CF2Cl2 Measurements by UARS Cryogenic Limb Array Etalon Spectrometer: Validation by Correlative Data and a Model", by R. W. Nightingale et al., J. Geophys. Res., 101, pp. 9711-9736, April 30, 1996. The CLAES V7 CF2Cl2 data set provides global coverage from 80S to 80N for 388 days in the period from January 9, 1992 to May 5, 1993. To arrive at estimates of the systematic error we have (1) compared CLAES CF2Cl2 profiles with a wide variety of available correlative data from balloon and space-borne sensors, and (2) carried out empirical estimates of experimental systematic error based on knowledge of instrument characteristics. To demonstrate that the CLAES CF2Cl2 follows the tracer gas seasonal morphology, we have compared CLAES CF2Cl2 maps with those from model calculations and with those from CLAES N2O. In most cases close coincidence in latitude (+/-2 degrees), longitude (+/-10 degrees), and time (+/-6 hours) was available with correlative data over the CLAES altitude range. The correlative balloon data were confined to the northern latitudes, both at 35N and 68N, mostly in the spring and fall. The ATMOS experiment provided much more extensive coverage at south latitudes from 56S to 17S, although confined to approximately 7-day periods in the spring of 1992 and 1993. To arrive at estimates of random error we (1) looked at the variability in the CLAES profile data for multiple crossings of the +/-32 degrees latitude turnaround points, (2) made empirical estimates based on knowledge of instrument characteristics, and (3) compared with the algorithm-embedded errors. CF2Cl2 Profile Errors: The results of these validation exercises indicate that the mean fractional differences of the CLAES CF2Cl2 mixing ratios are within +6% and -19% of the available mid-latitude and tropical correlative data over the range of 16 to 34.7 km (100 to 6.8 mb). Based primarily on the percentage random error estimates being substantially larger above 32 km and on the increased small scale structure seen in the zonal-mean plots below 23 km in the tropics and lower in the mid-latitude region, the altitude range of most confidence in the CLAES CF2Cl2 data has been selected to be 18.7 to 32.0 km (~68 to 10 mb). In this range the mean fractional differences are between +6% and -13%. On average, compared to the mid-latitude, spring through fall correlative data, the CLAES values do not show a bias. Empirical estimates of the CF2Cl2 systematic error for the same latitude and seasonal conditions cover the range of 12 to 14% over 17 to 32 km, in reasonable agreement with the mean fractional differences between the CLAES and correlative data. The systematic error estimates apply for the mid-latitude spring, summer and fall seasons and all seasons in the tropics. The instrument portion of the systematic error is only weakly dependent on latitude through the retrieved temperature. Thus, the systematic error can be extended to apply to those spatial regions mentioned above where there were no correlative data. Based on the mean comparisons with the correlative data and on the empirical estimates of the systematic error, we have assigned a systematic error of 14% to the profiles over the range of 18.7 through 32 km and 22% for 33-35 km and 16-18 km. This takes into effect the 440 pptv artifact that can sometimes cause up to a -14% error in the tropics for altitudes below about 23 km. The systematic errors can be larger in winter, especially in the cold polar regions beyond 60 degrees, primarily due to the increased temperature errors in this region. Comparisons of the CLAES profiles with those correlative data at 68N in January and March of 1992 showed larger percent fractional differences than for the mid-latitude comparisons. The CLAES CF2Cl2 observed data repeatability, or precision, obtained at orbit turnaround, lies between 32 and 11 pptv (6 to 13%) over the range of 17 to 32 km, in good agreement with the empirical estimates of the random errors (6 to 9%). For this range an average precision of 9% has been assigned. Algorithm-generated estimates of the CF2Cl2 error lie between 20 and 30% for the altitude range 17 to 32 km and can be larger at higher altitudes. These error estimates exceed the other error estimates and the correlative data comparisons at all altitudes, suggesting that the algorithm-generated estimates may include some additional systematic component. For those data users who need to utilize the algorithmic errors, we recommend that the errors be scaled by a factor of 1/2. Overall, although there is useful information in the CLAES CF2Cl2 individual profiles from 16 to 48 km, our best confidence altitude range was chosen to be 18.7 to 32 km (~68 to 10 mb). In this range the total error, or accuracy, (RSS of systematic and random errors) is ~17%, applicable for the spring, summer and fall mid latitudes and for all seasons in the tropics. As discussed above an average precision of 9% was assigned for this range. CF2Cl2 Global Field Analysis: Examples of 1-day polar area maps at 21 mb were presented for both CLAES CF2Cl2 and N2O projected onto a southern hemispherical globe at different seasons showing good large- and small-scale correlation between the tracer fields. Good correlation was also seen in the zonal-mean global field data, shown for north- and south-looking cases for all seasons. Additional comparisons of CF2Cl2 in the form of 16-month time series of zonal-mean latitude distributions at two pressure levels were similar to those from the LLNL-2D model. Overall agreement was seen with the major morphological and seasonal features. These included the steep descent of the mixing ratio isopleths towards the winter poles, the double-peaked latitudinal structure at high altitudes near the vernal equinox, and the mid-latitude "surf-zone" during the winter season. An apparent upwelling feature not present in the model was seen in the zonal-mean-temporal maps of CF2Cl2 in the coldest southern polar winter region. This feature has been determined to be a V7 software artifact and will be corrected in a later version. CF2Cl2/N2O scatter diagrams for the southern spring of 1992 and 1993 showed compact linear relationships for N2O volume mixing ratios > 100 ppbv for the extratropical data cases. More scatter in the data at lower altitudes in early 1992 than in 1993 may be associated with the CF2Cl2 440 pptv artifact, a possibility still under investigation. However, for the tropical latitudes the correlation relationships show some curvature for both high- and low-altitude data. Although there are clearly areas for improvement of the retrieval algorithm, the results of this study indicate that the V7 CLAES CF2Cl2 data are of good overall reliability in the spring, summer and fall mid-latitude and tropic regions. ----------------------------------------------------------------------- 10.0 TRICHLOROFLUOROMETHANE (CFCl3, or CFC-11) The first version of trichlorofluoromethane to be generated was that in version 7. CFCl3 is retrieved from CLAES blocker region 7 centered at 843 cm^(-1). As stated below, "considerable caution" was recommended in the use of the V7 CFCl3 data. Some caution is still advised in V8, which is best used after July 1992 and away from the high-latitude winter hemispheres. A summary of V7 is included to serve as a baseline for understanding the much improved V8 CFCl3 product. 10.1 V7 CFCl3 Summary CLAES is the only instrument on UARS measuring CFCl3. Unlike CF2Cl2, which has a distinct spectral structure, CFCl3 has a continuum-like emission spectrum at the CLAES resolution with little spectral contrast to distinguish it from aerosol and other atmospheric (or instrument) continuum backgrounds. Therefore, it is very susceptible to interference effects from such backgrounds, and particularly from the Pinatubo tropical aerosol cloud below 20 mb and intense PSCs in the polar winter vortices. There were no correlative data available for comparison at the time of the Langley 1994 workshop. Comparisons were made with the LLNL-2D model for January and April 1992. These comparisons showed general morphological agreement with the model, but there were strong indications of interference from the Pinatubo cloud in the tropics and from PSCs near the winter poles, as expected. There was also a suggestion that some fraction of the data was unduly influenced by the climatology used as the first guess in the retrieval. We expect to significantly upgrade the retrieval algorithm in future versions to mitigate these effects. The tropical aerosol cloud had dissipated significantly by late July of 1992, and the zonal mean data fields appear substantially less noisy beyond this date in tropical and subtropical regions. Any use of the data therefore in the current retrieval would be best confined to dates later than July 1992. Additionally, even beyond this date, polar winter regions should be avoided (i.e., from about 50 deg latitude to the winter pole for July-September 1992 in the south and December 1992-March 1993 in the north). We recommend considerable caution in using V7 CFCl3 even during these "optimal" periods, as we consider the retrieval to be quite preliminary and very likely to be subject to significant change in future versions. 10.2 Comparison of V8 CFCl3 with that of V7 The V8 CFCl3 product received specific attention in the L1-3 retrieval algorithms in addition to the generic changes in the V8 algorithms mentioned elsewhere. The product has improved in a number of areas over that of V7. CFC-11 zonal mean plots show substantial improvement over those of V7, in that isopleth shapes and structure are now more similar to those of V7 CFC-12 and V8 N2O tracers than to those for the V7 CFC-11 (we have little or no correlative data for comparison). Besides these similarities with the other CLAES tracer gases, the improvements include the significant reduction in the tendency of the isopleths to show anomalous upwelling in northern winter high latitudes. The V8 CFCl3 data still have distinct problems that are the subject of on-going work for the next version, including tropical aerosol interference before July 1992 and PSC interference in the 1992 south polar winter. So CFC-11 is best used after July 1992 and away from the high latitude winter hemisphere regions. Also in January and February 1992 the isopleth fall off is too rapid for the higher volume mixing ratios, compared to those for F12 and N2O. In addition the southern mid-latitude and polar regions before February 13, 1992 show missing CFC-11 vmr from about 18 to 24 km in altitude stretching south from about 25 degrees S. Even with these problems the V8 CFC-11 compares better with the CFC-12 and N2O than does the V7 CFC-11. ------------------------------------------------------------------------- 11.0 DINITROGEN PENTOXIDE (N2O5_OTHER) Dinitrogen pentoxide version 7 has been subjected to an extensive validation exercise that is documented by Kumer et al. (J. Geophys. Res., Vol. 101, No. D6, 9657 - 9678, 1996) in the UARS special validation issue. It is very important for users of the data to carefully read this paper. 11.1 Summary of the V7 N2O5 Validation Findings The CLAES and ISAMS on the UARS have made the first near-global N2O5 measurements. Only CLAES data on the `standard UARS pressure surfaces,' 3.16, 2.15 and 1.47 mb, were addressed by V7 validation study. Only the V7 data on these surfaces are recommended for serious science studies. A comparison of the diurnal data variation with the model shows generally good qualitative agreement, except during late morning and afternoon when the CLAES data show offsets from zero that are to first order independent of local solar time. The magnitudes of the offsets are in the range 0 to 1.5 ppbv, dependent on pressure surface, latitude and time of year; and the time scale for sensible change in an offset (at a particular latitude and pressure surface) is of the order of 36 days, the typical period of a UARS hemispheric view. On subtraction of the offsets from the CLAES data, much better agreement with the model is noted. The offsets are thought to be due to instrument continuum emission that is not yet well understood, but candidate mechanisms include continuum radiance from aerosol, weak continuumm radiance from species that are not modeled in the CLAES V7 retrieval process, uncharacterized scattering within the instrument, far wing spectral out of band, and/or a slightly broader etalon transmission than is used in the forward model calculation. The CLAES data are compared with 12 profiles of sunrise ATMOS data obtained on latitudes from approximately 0 to 28N during the ATLAS 1 mission on the March 24 - April 7, 1992. ATMOS data are reported as high as 3.16 mb for 11 of those 12 profiles, and the average ATMOS value there is 2.40 ppbv with a one sigma variability of 0.29 ppbv. For these 11 comparisons the offsets-subtracted sunrise zonal mean CLAES average value and variability are 3.18 +/- 0.53 ppbv, while for no offset subtraction they are 4.15 +/- 0.55 ppbv. On the average the CLAES data with offsets subtracted are greater than the ATMOS 1992 sunrise data by 33% at 3.16 mb. The ATMOS data extend up to 2.15 mb for one profile of the 12 and agree with the CLAES data with offsets subtracted there within < 0.1 ppbv (i.e., 5%). There are sunrise zonal mean ISAMS data available for some of these profiles and at 1.47 mb the CLAES data with offsets subtracted and ISAMS data agree within better than 0.13 ppbv on the average. At the lower altitudes of 3.16 and 2.15 mb the average values of the CLAES data with offsets subtracted are greater than ISAMS by 1.05 +/- 0.49 and 0.39 +/- 0.19 ppbv, respectively, i.e., the CLAES data with offsets subtracted are greater than ISAMS by 37 and 23%, respectively, on these surfaces. During the ATLAS 2 mission from April 8 through 16, 1993 ATMOS sunrise data were obtained in 4 profiles near 64N latitude. The CLAES views from the sun shaded side of the UARS did not allow for CLAES to obtain data in the north high latitudes at that time. However, the offsets-subtracted sunrise zonal mean CLAES observations on the north look prior to the time of the ATMOS observations, extrapolated to the time of those observations, were consistent with the ATMOS observations. The balloon borne JPL MARK IV instrument obtained sunrise data near 35N on 15 September up to approximately 2.15 mb and the offsets-subtracted sunrise zonal mean CLAES data on the average compare within 0.3 and 0.15 ppbv at 3.16 and 2.15 mb, respectively, i.e., within approximately 30 and 15%, respectively. During an N2O5 enhancement event, which was nearly coincident with the early-mid January 1992 stratospheric warming event, the CLAES and ISAMS show detailed global scale features that are highly correlated. Values of about 5.5 ppbv observed in the most intense enhancement region are the largest ever reported. Examination of the early mid January 1992 data indicates users should sometimes (but not always) be cautious of CLAES V7 data at 3.16 mb during large enhancement events. The problem is never encountered at the highest recommended level, 1.47 mb. But, examination of data throughout the mission suggests that users should also sometimes (but not always) be cautious in the use of near-sunrise data at 3.16 mb in the latitude range 32S to 32N, especially during CLAES north-viewing periods. Excluding the offsets, CLAES systematic errors estimated on the basis of instrument and retrieval process characterization range from 14% at 3.16 mb to 21% at 1.47 mb. However, equatorial correlative comparisons suggest the CLAES data with offsets subtracted are too large by the order 30 to 40% at 3.15 mb, but improve at the higher altitudes towards the estimated value. Also, comparisons with ISAMS and the MARK IV suggest improved agreement at 3.16 mb at mid and high latitudes. The CLAES precisions have been estimated by three methods that agree reasonably well. At 3.16 and 1.47 mb, respectively, the characterization method estimates precisions of 9 and 26%, the track intersects method estimates 19 and 24%, and the embedded error estimates are 8 and 36%. 11.2 Comparison of N2O5 V8 with V7 The V8 is significantly improved over V7 in that it now can be used on more pressure levels, and the problem with artifacts at the 3.16 mb surface that sometimes occurred during enhancement events has been fixed. The V8 can now be used on the UARS standard pressure levels 10.00, 6.81, 4.64, 3.15, 2.16, and 1.47 mb. Comparison of the offsets-subtracted V8 with correlative data, and the associated V8 error characteristics, are similar to those found in V7. The problems with artifacts in V7 that were previously found at 3.16 mb have been corrected. However, the heavy aerosol early on in the mission does present problems at 10 mb in the cases listed below: YAW PERIOD PRESSURE LEVEL LATITUDE RANGE ----------------- ------------- -------------- 9/29/91 - 11/1/91 10 MB 24S to 24N 11/7/91 - 12/2/91 10 MB 24S to 24N 12/8/91 - 1/11/92 10 MB 24S to 8N 1/18/92 - 2/11/92 10 MB 24S to 8S 3/27/92 - 4/29/92 10 MB 24S to 8S In the majority of comparisons the offset-subtracted V8 agree better with the correlative data than do the offset-subtracted V7. For most uses it is recommended that the diurnally independent offsets should be subtracted from the data. The V8 offsets are not necessarily the same as V7. A data file and subroutine are provided and, as described below, can be used to generate the V8 offset and its error for arbitrary UARS_DAY, LATITUDE and PRESSURE. To determine the pressure (p), interpolation is used for 1.47 < p < 10 mb. Outside that range the appropriate limit values @ 10 mb and 1.47 mb are returned for p > 10 mb and p < 1.47 mb, respectively. Latitude is handled similarly. The subroutine, the data file it uses, and documentation are provided in the CLAES anonymous ftp area DATA1:[ANONYMOUS.N2O5_V8_OFFSETS] on the internet node claes.space.lockheed.com . The subroutine also provides flags to indicate if any of the parameters input to it, ie., UARS_DAY, LATITUDE and PRESSURE, are outside the limits of when and where the data were taken. The CLAES V8 offset subtraction subroutine CLSV8_N2O5_OFFSET_SUBTRACTION is written in two versions so that it may be incorporated into a FORTRAN or an IDL program with respective suffixes of .FOR ans .PRO. Using inputs of UARS_DAY, LATITUDE and PRESSURE, the subroutine will output the estimated CLAES N2O5 V8 diurnally independent offset and the associated estimated error (in units ppbv). The subroutine reads the CLAES mission V8 offset data from the file V8_OFFSETS.DAT. Interactive driver programs demonstrate how to include the subroutine into FORTRAN and IDL programs. ------------------------------------------------------------------------ 12.0 NITRIC OXIDE (NO) Various changes appear in the V8 retrieved NO vs V7, in part due to temperature and generic changes. Pending analysis of these changes,the V8 NO data have not been transferred to the DAAC. 12.1 V7 NO Summary only In general the CLAES NO data show (i) a resemblance to climatological zonal mean structure although they become unrealisticly large for altitudes above ~ 1 mb, (ii) reasonable day to day consistency in zonal mean maps, (iii) no apparent aerosol degradation and (iv) are not particularly sensitive (of the order 10% on a zonal mean on doubling the apriori) to apriori. There is no retrieval for night time conditions. To the extent it is possible, the comparisons with HALOE are reasonable. Other comparisons, and potential correlative data sources are discussed below in the main section on NO. Error discussion: Examination of CLAES production software error estimates indicates best confidence in the region from ~ 5 to 1 mb in general. This is supported by comparison of CLAES zonal means with climatology. Below 5 mb the climatology scaled by the data in the 1 to 5 mb region is reported. Above 1 mb the retrieval is unrealisticly too large, probably due to inadequate compensation for high altitude non-LTE NO emission. Based on the comparisons with our night time NO2, and potential systematic error in it as discussed in section 2.2.11 above, we estimate systematic error in our NO of ~ 30% too large at 2.2 mb. The assumption that NOy is mostly NO2 by night, and mostly NO by day, at 2.2 mb is used in this 1st order analysis. Repeatability in the daily zonal mean sense at 2.2 mb is also ~ 30%. Error bars on individual profiles can be considerably larger. Caveats: The NO spectral region is at relatively short wavelength, and the radiance data are therefore relatively noisy. There is considerable work to be done in optimizing the CLAES NO retrieval which is at present in a relatively immature state. Work that remains includes - Improved DC zero level component subtraction. This is an electronics problem that can be solved by the processing software in future versions. - Subtraction of high altitude non-LTE NO radiance still very crude. - non-LTE effects in the stratosphere may be important, may be responsible for apparent systematic error to the high side. - Develop capability to average zonal mean radiances & retrieve from these. ----------------------------------------------------------------------- 13.0 HYDROCHLORIC ACID (HCl) In the current CLAES production software, HCl is not retrieved. However, our radiance data indicate that we should be able to retrieve HCl once some special processing software is developed and in place. ------------------------------------------------------------------------ CLAES V0008 Data Products Summary: Parameter Retrieved Best confidence Errors in Best Comments Range(mb) Range (mb) Confidence Range [Sys] [Random] ----------------------------------------------------------------------- Temp 150-0.1 100-0.2 3-4 K 0.8-3.5K [1] O3(B9) 120-0.1 68-0.7 10-15% 0.1-0.3 ppmv [2] N2O 100-0.15 46-2 15-20% 20-5 ppbv [3] NO2 100-0.1 20-0.3(night) 30-40% 1-0.3 ppbv [4] 20-2 (day) HNO3 100-0.1 70-3 10-20% 1-0.3 ppbv [5] ClONO2 100-1 68-6 25% 0.1-0.2 ppbv [6] N2O5 46-1 10-1 10-40% 0.1 ppbv [7] CFCl3 100-10 46-20 mb see comments [8] AEROSOL 150-10 68-20 35% 10-35% [9] ----------------------------------------------------------------------- CLAES V0007 Data Products Summary CH4 100-0.1 46-0.2 15% 80-40 ppbv [10] CF2Cl2 100-2 68-10 14-22% 32-11 pptv [11] H2O 100-0.1 Alt/Lat/Dep 20% 1-0.5 ppmv [12] NO 100-1 5-1 30% 30% [13] Comments (V8 products) ----------------------- [1] Temperature: When compared to NMC and UKMO analysis CLAES temperatures are generally cold by 2-5 K from 1-100 mb, and are warm above the 1 mb level. They are very cold in the tropics near 10 mb and in the southern polar winter above 10 mb and near thick PSC's. [2] O3(B9): Tends to be biased high at low altitudes in tropics, possible due to aerosol interference [3] N2O Local maxima in the tropics, 40-20 mb, prior to September 1992. [4] NO2 Biased low by 30-40% vs correlative data. Generally errors increase rapidly below 20 mb. Potentially large errors in polar night winter conditions with large vertical temperature gradients. [5] HNO3 There may be a problem in the southern hemispheric mid to high latitude data in autumn and early winter for cases where the vmr excedes ~ 15 ppbv. In these cases the vmr may systematiclly be too large by more than the 20% value that is given in the summary table. [6] ClONO2 Errors increase substantially above 6 mb. Tropical (20S-20N) data below 26 km possibly showing interference from Pinatubo aerosol cloud prior to May 1992. Some PSC interference in Antarctic winter. [7] N2O5 Very good confidence from 10 to 1 mb for the offsets subtracted case. SW is available at CLAES anonymous ftp site to perform the subtraction. The data is also useful at higher and lower altitudes for some regions and times, but user should be VERY careful in assessing what the offset might be, as this has not been done by the CLAES team. The systematic error applies for the offset subtracted case. It varies with region, altitude, and time into the mission. [8] CFCl3 Likely significant interference from Pinatubo aerosol and polar winter PSCs. Most qualitatively-useful data for summer-fall northern hemisphere-looking periods between July 17 and October 26, 1992, and summer-fall southern hemisphere-looking periods between November 2, 1992 and April 15, 1993. [9] Aerosol The 1257 cm-1 V0008 retrieved aerosol absorption coefficients are set near zero for altitudes where the continuum extinction coefficients become < 6.3E-04 km-1; the daytime 1897 cm-1 has significant contribution from solar scattering; there is a 25% to 35% difference between the 790 and 780 cm-1 regions but theory suggests a 3% maximum difference. Comments (V7 products) ---------------------- [10] CH4 Local Maxima in the tropics, 12-6 mb, prior to May, 1992 [11] CF2Cl2 Unrealistic isopleth "upwelling" in coldest south polar winter conditions. Some indication of residual Pinatubo interference in tropics below 20 mb during first part of 1992. Vmrs at 440 pptv can be up to 14% low, usually low altitudes in tropics [12] H2O Non-LTE effects in daytime values; recommend using night time values. Polar winter values significantly too high; tropical values below 10 mb too low. Best confidence: mid-latitudes below 10 mb; mid to high latitudes in Summer hemispheres; all latitudes above 1 mb. [13] NO Daily zonal mean-daytime error estimates only. Non- physical values at altitudes above 1 mb. Scaled climatology at altitudes below 5 mb