SP-531 Envisat Validation Workshop

In order to assess the level-2 data retrieval accuracy of selected trace gases (O3, NO2, H2O) from SCIAMACHY and MIPAS, which are part of the atmospheric instrumentation on the recently launched ENVISAT satellite, operational and some scientific products are validated by comparison with other space borne instruments (GOME, HALOE, SAGE II, POAM III). First results of these satellite intercomparisons are presented in this paper. SCIAMACHY O3 vertical columns (version 3.53 and 4.0) are on average 5% (+/-5%) lower than GOME O3 columns. SCIAMACHY NO2 slant columns of the operational product (4.0) and IUP retrieval [1] show a consistent offset to GOME NO2 slant columns (GDP version 2.7) of –30% and -10%, respectively. While the NO2 total columns of the IUP product show a stable offset of –20%, version 4.0 of SCIAMACHY NO2 total columns show a strong variation with latitude (–60% at 70°S to 0% at 70°N), which is even worse for version 3.53. Since no operational products of SCIAMACHY limb measurements are available so far, only scientific limb products for O3 and NO2, retrieved by IUP [2], are compared to POAM III measurements. These preliminary results give confidence that reliable profile data can be retrieved from SCIAMACHY limb measurements. MIPAS O3 profiles (4.53) show good results between 20-60 km with +/10 % compared to HALOE and between –35 and 0% compared to SAGE II. MIPAS H2O profiles (4.53) agree also well between 20-55 km with differences between +5 to +15 % compared to HALOE. Below 20 km both, MIPAS H2O and O3 profiles show large deviation to HALOE and SAGE II measurements. 1 STATUS OF VALIDATION Table 1: Workplan for AO-ID 651: Satellite Instruments with launch date, measurements geometry and their data products for validating level 2 products of ENVISAT instruments GOMOS (G), MIPAS (M), and SCIAMACHY (S). Green: first validation has been performed, black: planned validation, blue: comparisons to new instruments planned during main validation phase. Asterisk (*) indicates that so far only SCIAMACHY profiles retrieved by IUP have been validated Instrument Data product Geometry Envisat Instrument SAGEII (10/84) O3 profiles* NO2 profiles H2O profiles occultation G,M,S* HALOE (9/91) O3 profiles* NO2 profiles H2O profiles CH4 profiles occultation G,M,S* only M,S POAM III (3/98) O3 profiles NO2 profiles occultation S* GOME (4/95) O3 columns NO2 columns nadir S TOMS (7/96) O3 columns nadir S SABER (12/01) O3 profiles H2O profiles limb G,M,S ACE-FTS (5/03) O3 profiles NO2 profiles H2O profiles CH4 profiles occultation G,M,S The SCIAMACHY (Scanning Imaging Absorption Spectrometer for Atmospheric Chartography, see e.g. [3]) instrument, the MIPAS (Michelson Interferometer for Passive Atmospheric Sounding, see e.g. [4]) and GOMOS (Global Ozone Monitoring by Occultation of Stars, see e.g. [5-7]) are space-based atmospheric instruments launched __________________________________________________________________________________________________________ Proc. of Envisat Validation Workshop, Frascati, Italy, 9 – 13 December 2002 (ESA SP-531, August 2003) onboard ENVISAT (Environmental Satellite) in March 2002. These three ENVISAT instruments provide information of a large family of trace constituents, which play an important role in atmospheric chemistry and in climate change issues. A prerequisite of scientific use of the trace gas data is to achieve a high accuracy in the derived trace gas products. The use of independent satellite measurements to validate trace gas products of these instruments has the great advantage that pole-to-pole coverage for all seasons is available and that validation activities are not limited to a certain period and location. The work plan (Table 1) of this AO-project (AO-ID 651) includes validation of SCIAMACHY, MIPAS and GOMOS O3, NO2, H2O and CH4 data by comparison with established and well validated satellite instruments GOME (the Global Ozone Monitoring Experiment, [8]), TOMS (Total Ozone Mapping Spectrometer, [9]), HALOE (Halogen Occultation Experiment, [10]), and SAGE II. (Stratospheric Aerosol and Gas Experiment II Instrument, [11]). During the main validation phase of ENVISAT instruments the same atmospheric ENVISAT data products will be intercompared to new satellite instruments SABER (Sounding of the Atmosphere using Broadband Emission Radiometry, [12]) and ACE-FTS (Atmospheric Chemistry Experiment Fourier-transform spectrometer, [13]). At the time at this workshop only a subset of all validation tasks could be completed, because ENVISAT data delivery was very limited. Since all GOMOS data and MIPAS NO2 and CH4 data were only available after late October and SAGE II and HALOE data only until the end of October, no coincidences could be determined. Therefore, only operational products of SCIAMACHY O3 and NO2 columns (also scientific NO2 columns) and MIPAS O3 and H2O profiles have been validated until now and are presented in this paper. For SCIAMACHY no operational limb profiles were distributed so far, so only scientific products were validated. These are compared with HALOE, SAGE II and POAM III (The Polar Ozone and Aerosol Measurement III, [14]) measurements and some results are presented here. 2 SCIAMACHY VALIDATION RESULTS 2.1 SCIAMACHY and GOME DOAS O 3 and NO 2 column products Vertical column densities of O3 and NO2 can be derived from SCIAMACHY and GOME UV-VIS nadir measurements. The Differential Optical Absorption Spectroscopy (DOAS, [15]) is used to derive trace gas columns from GOME and SCIAMACHY spectra. For SCIAMACHY and GOME, the DOAS algorithm determines O3 and NO2 slant columns in the 325-335 nm (UV) and the 425-450 nm (VIS) spectral window, respectively. In addition, SCIAMACHY retrieves ozone slant columns in the 425-450 nm (VIS) spectral window. Both retrieval versions (3.53 and 4.0) of SCIAMACHY level 2 data correspond to GOME GDP 2.4. These versions use the US standard atmosphere for NO2 which usually leads to an underestimation of vertical number densities under polluted conditions. Compared to version 3.53, version 4.0 of SCIAMACHY has an improved level-1 product with an polarisation correction, an updated SCIAMACHY sun spectrum, a new dark current and spectral calibration applied. For version 3.53, and probably for version 4.0, as well GOME absorption cross sections were used in the spectral fitting (pers. comm. by J. Frerick, ESA). GOME level-2 data used for these comparisons are of version GDP 2.7. This GOME version improves NO2 columns retrieval in the tropics by inclusion of O4 and H2O in the fitting. GOME GDP 3.0 [16] has already been implemented for older data (19951999) and will be soon implemented for recent dates. This version improves total O3 (tO3) retrieval by applying TOMS V7.0 climatology for O3 and an air mass factor classified by vertical column amount and latitude. 2.2 Comparison of SCIAMACHY O3 total columns (UV) with GOME Since GOME/ERS-2 and SCIAMACHY/ENVISAT are flying in the same orbit only 30 minutes apart, numerous collocated measurements can be detected (up to 28000 a day). In order to quickly compare collocations of a day up to a months period, and in addition to that, to overcome the difference in ground pixel size of 30 km x 60 km of SCIAMACHY and 40 km x 320 km for GOME, the following method was applied: All O3 total column data of a certain time period (day, week, month) are spatially binned into regular 2.5° and 2.5° grids. Fig. 1 shows binned tO3 of 2002/10/24 measured by SCIAMACHY (3.53) and GOME (GDP 2.7). For most orbits SCIAMACHY makes alternating nadir and limb measurements. Only part of all SCIAMACHY orbits of one day were available. When all measurements are available, SCIAMACHY achieves global coverage after six days, while GOME measuring in nadir geometry achieves global coverage after three days. If both instruments have measurements in the same grid, then the mean of the data of one instrument is compared to the mean of the data of the other instrument as follows: (tO3 of SCIAMACHYtO3 of GOME)/ tO3 of GOME (1) Fig. 2 shows the results of the comparison of binned SCIAMACHY (3.53) and GOME (2.7) total ozone for one day (2002/10/24) and a six-day period (2002/10/24 to 2002/10/29). Fig. 3 shows for the same comparisons the results as a function of latitude. Excluding the Antarctic region (>70°S), SCIAMACHY is 5% (+/-5%) lower than GOME. GOME tO3 data still show a very good accuracy (around 3% [17]) despite aging of the instrument. Fig. 1. Total ozone in Dobson Units (DU) of 2002/10/24 measured by SCIAMACHY version 3.53 (left) and GOME GDP 2.7 (right) after binning into 2.5° x 2.5° grid boxes Fig. 2. Comparison of binned SCIAMACHY (3.53) with binned GOME (GDP 2.7) total O3 for 2002/10/24 (left) and 2002/10/24 to 2002/10/29 (right) Fig. 3. Comparison of binned SCIAMACHY (3.53) with binned GOME (GDP 2.7) total O3 for 2002/10/24 (left) and 2002/10/24 to 2002/10/29 (right) as a function of latitude The comparison of SCIAMACHY 4.0 tO3 with GOME GDP 2.7 tO3 (2002/08/23) shows the same bias (Fig. 4). The only improvement in version 4.0 is the removal of bad pixels where measurements have been taken after twilight. A possible reason for SCIAMACHY tO3 showing an offset to GOME tO3 might be that in both SCIAMACHY versions 3.53 and 4.0 still GOME absorption cross sections are used in the algorithm. Fig. 4. Comparison of binned SCIAMACHY (4.0) with GOME (2.7) total O3 for 2002/08/23 as a function of latitude 2.3 Comparison of SCIAMACHY O3 total columns (VIS) with GOME Fig. 5. Comparison of binned SCIAMACHY (3.53, left) and (4.0, right) VIS total O3 with GOME (GDP 2.7) total O3 UV as a function of latitude Fig. 5 shows results from the comparison of the VIS SCIAMACHY tO3 product version 3.53 and 4.0 with UV GOME (GDP 2.7) tO3 using the same method as described earlier (in 2.2). Results of the comparison show that there is still a large deviation with SCIAMACHY (3.53) VIS tO3 showing up to 200% difference to GOME. SCIAMACHY (4.0) VIS tO3 comparisons to GOME show much better results, but still a very strong scatter around +/-80%. As in the case for GOME, the UV tO3 product is considered more reliable than the VIS ozone data. 2.4 Comparison of SCIAMACHY NO2 columns (UV) with GOME SCIAMACHY NO2 vertical column densities (VCD) of the operational product (3.53 and 4.0) and of the scientific product by IUP were compared to GOME (GDP 2.7) NO2 VCD using the same method as described in chapter 2.2 for the comparison of the tO3 data product. The scientific NO2 product from the SCIAMACHY data was derived using the DOAS analysis developed for the GOME instrument [1]. Starting from level-1 data with only the spectral calibration and dark signal correction applied, NO2 slant columns are retrieved in a non-linear fit using the wavelength window 425 450 nm. In addition to NO2, absorptions of O3, O4, H2O and the Ring effect are taken into account, using SCIAMACHY FM reference spectra where possible and a Ring spectrum calculated with the radiative transport model SCIATRAN [18]. Prior to the fit, the wavelength axis of the background spectrum is aligned with a Fraunhofer atlas. In contrast to the standard GOME analysis, no solar spectrum was used as a background, as SCIAMACHY solar reference measurements provided with the operational data still suffer from unresolved problems. Instead, an arbitrarily selected measurement over the Pacific is used, which leads to an unknown offset in the resulting slant columns that has been accounted for by using the slant column measured by GOME at this location. Once the slant columns are retrieved, they are converted to vertical columns using an air mass factor calculated by SCIATRAN assuming a US standard atmosphere. While this a priori assumption is acceptable in most cases, it will introduce errors at high latitudes in particular in winter. In general, the retrieval is very similar to that of the operational processor. Differences in the NO2 columns from both analyses are therefore probably related to the use of Pacific pixel as a background spectrum and in the not yet optimised software settings of the operational processor. Fig. 6. NO2 vertical column densities (VCD) of 2002/08/23 measured by SCIAMACHY 4.0 (left), GOME GDP 2.7 (middle), and SCIAMACHY IUP retrieval (right) binned into a 2.5° x 2.5° grid Fig. 7. Comparison of binned NO2 vertical column densities (VCD) from SCIAMACHY calibration orbits 509 and 510 (2002/08/23) with GOME. Left: SCIAMACHY (4.0) and right: SCIAMACHY (IUP retrieval) with GOME (GDP 2.7) for as a function of latitude Fig. 6 shows binned NO2 VCD from 2002/08/23 based on SCIAMACHY (4.0), IUP retrieval and GOME (GDP 2.7) data. Figs. 7 and 8 show results of comparing these SCIAMACHY (4.0 and IUP retrieval) NO2 VCD and their slant columns (SCD) to GOME (GDP 2.7). SCIAMACHY (4.0) VCD show a strong variation with latitude compared to GOME data ranging from –60% at 70°S to 0% at 70°N, while SCIAMACHY IUP retrieval show a stable offset of – 20% with a strong scatter (south of 60°S down to –40%). SCIAMACHY NO2 SCD from both algorithms (4.0 and IUP retrieval) show a negative offset compared to GOME; for the operational product (4.0) it ranges from –45% to –30% between 70°S and 50°N and for the IUP retrieval the offset is –10% +/–10%. From Fig. 7 and 8 one may conclude that both spectral fitting and air mass factor contribute to the differences between SCIAMACHY operational product and GOME. In addition, comparisons of binned NO2 VCD of 2002/11/08 from SCIAMACHY (3.53 and IUP retrieval) and GOME had been performed (Fig. 9). These comparisons were made to look for differences in NO2 VCD in regard to level-2 data versions and level-1 products used for the IUP retrieval. Level-1 data used for 2002/11/08 in the IUP retrieval was an older version compared to the one used at 2002/08/23. Deviations of SCIAMACHY (3.53) NO2 VCD compared to GOME (GDP 2.7) NO2 VCD are even higher than for SCIAMACHY (4.0), ranging from –50% at 70°S to +140% at 70°N. For SCIAMACHY (IUP retrieval) compared to GOME (GDP 2.7) deviations range from 0% at high latitudes to +50% in the tropics and differ quite a bit from the results of 2002/08/23. Fig. 8. Comparison of binned NO2 slant columns (SCD) from SCIAMACHY calibration orbits 509 and 510 (2002/08/23) with GOME as a function of latitude. Left: SCIAMACHY (4.0) and right: SCIAMACHY (IUP retrieval) with GOME (GDP 2.7) F Fig. 9. Comparison of binned NO2 vertical column densities (VCD) at 2002/11/08 from SCIAMACHY with GOME as a function of latitude. Left: SCIAMACHY (3.53) and right: SCIAMACHY (IUP retrieval) with GOME (GDP 2.7) for as a function of latitude 2.5 Comparison of SCIAMACHY O3 profiles with POAM III Since no operational product of SCIAMACHY trace gas product profiles have been available so far, scientific IUP products of SCIAMACHY O3 and NO2 profiles have been compared to collocated HALOE, SAGE II and POAM III measurements. First preliminary results of ozone profiles from SCIAMACHY limb and POAM III occultation measurements which are taken at the same day and close to each other (about 400 km apart) are shown in Fig. 10 [2]. Ozone profiles of SCIAMACHY are derived from two retrieval schemes: the method of A. Rozanov uses a differential fitting employing Chappuis bands, the method of C. v. Savigny uses a 3 wavelength retrieval in the same bands. POAM III has a slightly better vertical resolution of 1-2 km in the stratosphere and has been validated against ozone sondes and other satellite measurements [14]. All three profiles agree fairly well above 20 km despite the different viewing geometry and vertical resolution of SCIAMACHY and POAM III. The integrated subcolumns between 10 and 40 km are 378 and 384 DU for SCIAMACHY and POAM III, respectively. The total O3 for the geolocation of the SCIAMACHY measurement from GOME and TOMS are 459 and 456 DU, respectively.