Ozone Profile and Tropospheric Ozone Retrieval from SCIAMACHY

Ozone Profile and Tropospheric Ozone Retrieval from SCIAMACHY Nadir Measurements: Preliminary Results X. Liu 1, C. E. Sioris 1, 2, K. Chance 1(xliu@cfa. harvard. edu) A 33 B-0902 1 Atomic and Molecular Physics Division, Harvard-Smithsonian Center for Astrophysics, Cambridge, MA, USA 2 Sioris Atmospheric Consulting, Brampton, ON, Canada 2005 Fall Abstract We adopt our GOME ozone profile and tropospheric column ozone algorithm to both operational and recent verification (with a 15 -20% correction in irradiance) SCIAMACHY data (290310 and 325 -339 nm). The retrieval performance is better for the original SCIAMACHY data, although a large systematic bias exists. The verification corrects much of the systematic bias, but it may introduce additional structures that degrade the retrievals. The preliminarily retrievals of 3 orbits of data (from the original SCIAMACH data) agree well with KNMI TOSOMI data to within 3. 8± 6. 8 DU (1. 1± 2. 1%, R=0. 993) and agree with SAGE-II ozone profiles (2 coincidences) to within ~15% down to ~20 km. More comparisons with ozonesonde, SAGE, and Dobson/Brewer data are necessary to validate and improve the retrievals. 1. SCIAMACHY q Launched on March 1 st 2002 on board ENVISAT q Observing backscattered, reflected, transmitted or emitted radiation from the atmosphere and Earth's surface, in 240 -2380 nm at moderate resolution (0. 2 -1. 5 nm) q Nadir-mode spatial resolution: varying, 30 km x 60/120/240 km; global coverage: 6 days q Ozone profiles and tropospheric ozone can be derived from backscattered spectra in the ultraviolet. However, retrievals Fig. 1 Correction to SCIAMACHY data require accurate absolute radiometric calibration in the recent verification. q Large radiometric calibration error (15 -20%) exists in current SCIAMACHY data (Fig. 1), mainly in the solar irradiance as seen from the recent test data. q Objective: test ozone profile retrievals from operational and verification SCIAMACHY data (3 orbits on Feb 11, March 30, and May 12, 2003, respectively) 2. Algorithm Description q We use our GOME ozone profile and tropospheric column ozone algorithm [Liu et al. , 2005, JGR] v Fitting window: 289 -307 nm, 325 -339 nm v 11 layers, use tropopause to divide the troposphere & stratosphere, 2 -3 layers in the troposphere v Wavelength calibration in radiance/irradiance/cross sections, undersampling correction, variable slit v Radiometric calibration: using 2 nd-order poly. correction in ch 1 and 2 nd-order surface albedo in ch 2 v Improve radiative transfer modeling: ECMWF/NCEP daily temperature, surface & tropopause pressure, GOMECAT clouds, GOME surface albedo database, GOCART tropospheric and SAGE stratospheric aerosols, on-line Ring effect correction, polarization correction v TOMS V 8 ozone profile climatology [Mc. Peters, et al. , AGU 2003 F] q Modification: fixed slit, 3 rd-order surface albedo, 290 -310 & 325 -339 nm, FRESCO clout-top pressure Fig. 5. Integrated total column ozone for the same orbit in Fig. 3 and comparison with TOSOMI. Fig. 6. Examples of averaging kernels (DFS=4. 3, 0. 9 in the trop) 4. Retrieval Characterization q Fig. 6: retrieved profile is well resolved at 5~40 km with small a priori influence, a priori influence is 0. 3 in the lowestlayer q Vertical resolutions: 8 -12 km at 20 -40 km, 816 km in the troposphere (Fig. 6) q 3. 5 -6 DFS in the atmosphere, 3 -5 DFS in the stratosphere. (Fig. 7) q Generally, 0. 5 -1. 5 DFS in the troposphere between ± 50ºN/S. q Uncertainties (random and smoothing): 0. 6%, Fig. 7. Degrees of freedom for signal in 1. 2%, 16% in total, stratospheric, and the atmosphere, stratosphere and tropospheric column ozone, 5 -10% at 20 -60 km, troposphere. 15 -30% at 0 -20 km. Fig. 3 Retrieved profiles (DU/layer) for one orbit (nadir Fig. 4 Tropospheric column ozone for pixels) of original data. The red line shows the tropopause. the same orbit shown in Fig 3. m Fig. 11. Same as Fig. 10 except for the corrected SCIAMACHY data. 5. Comparison with SAGE-II (15 -60 km) q Two coincidences between SCIAMACHY & and SAGE-II data are found. q Average all SCIAMACHY retrievals within ± 1º lat. and ± 5º lon. of the SAGE-II observations. q Integrate/interpolate SAGE-II/SCIAMACHY to a common altitude grid (similar to SCIMACHY) q The retrievals from the original data agree with SAGE-II data to within 15% at 20 -60 km, better than those from the corrected data. Summary and Future Outlook 3. Examples of Retrievals q More successful retrievals with the original data than the corrected data (7598 vs. 5030) q Smaller fitting residuals in the original data (on average, 1. 6% in 290 -310 nm and 0. 15% in 325 -339 nm) (Fig. 2). Ch 2 fitting residuals are slightly better than those in GOME. q Fitted on-line correction at 310 nm: 18%± 8% for original data, 7%± 10% for corrected data. Large error exists. q Figs 3 -5: O 3 profiles, tropospheric column O 3, and integrated total column O 3 for an orbit on Feb 11, 2003. The tropospheric Fig. 2 Examples of fitting (original data). ozone distribution agrees reasonably with GOME retrievals. Fig. 10. Two examples of comparisons between SCIAMACHY and SAGE-II ozone profiles (at SCIAMACHY retrieval grids) for the original data. Fig. 8. Comparison between integrated and TOSOMI total column ozone (left) and their differences (right) as a function of latitude for two orbits of the original data. 5. Comparison with TOSOMI q TOSOMI: KNMI total ozone retrieval algorithm (DOAS fitting + air mass factor) for SCIAMACHY data [Valks and van Oss, 2003, TOGOMI ATBD] q TOSOMI vs. Ground-based observations (87 locations): -1. 5± 4. 9% on average [Brinksma, 2004] q Due to inconsistencies in viewing geometry, geolocation, cloud fraction, we use coincidences with differences <1º in SZA, <2º in VZA, and < 0. 2 in cloud fraction between ours and TOSOMI q Generally good agreement between our Fig. 9. Same as Fig. 8 right but for the retrievals and TOSOMI. Larger differences and corrected data. more scattered at higher latitudes (Figs 5, 8 -9) q Despite large difference in normalized radiance between the original and corrected data, the comparisons with TOSOMI are similar. q Overall, the retrievals from the original data agree slightly better with TOSOMI total column ozone: 3. 8 ± 6. 8 DU (1. 1 ± 2. 1%) vs. 0. 5 ± 8. 4 DU (0. 1 ± 2. 4%) q We adopt our GOME ozone profile and tropospheric column ozone algorithm to both original and corrected SCIMACHY data. q Large systematic bias exists in SCIAMACHY data. The 2 nd-order polynomial correction in channel 1 and 3 rd-order polynomial in surface albedo partly correct the radiometric calibration errors. q Overall, the retrievals from the original data are better in terms of fitting residuals, number of successful retrievals, and comparison with SAGE-II and TOSOMI data. q The integrated total column ozone agrees with TOSOMI to 3. 8 ± 6. 8 DU (1. 1 ± 2. 1%), but shows larger differences and scatters at high-latitudes. q The retrieved ozone profiles agree with SAGE-II data to within 15% at 20 -60 km. q More validation against SAGE, ozonesonde, and Dobson/Brewer observations are necessary to further test and improve the retrieval algorithms. References Ø Brinksma, E. , Validation of SCIAMACHY column ozone with ground-based data, KNMI, 2004. Ø Liu, X. , K. Chance, C. E. Sioris, R. J. D. Spurr, T. P. Kurosu, R. V. Martin, M. J. Newchurch, Ozone Profile and Tropospheric Ozone Retrieval from Global Ozone Monitoring Experiment (GOME): Algorithm Description and Validation, J. Geophys. Res. , 110(D 20), D 20307, 10. 1029/2005 JD 006240, 2005. ØMc. Peters, R. D. , J. A. Logan, and G. J. Labow (2003), Ozone climatological profiles for version 8 TOMS and SBUV retrievals, Eos. Trans. AGU, 84(46), Fall Meet. Suppl. , Abstract A 21 D-0998. Ø Valks, P. , and R. van Oss, TOGOMI Algorithm Theoretical Basis Document, KNMI, November 2003. Acknowledgements This study is supported by NASA and by the Smithsonian Institution. We acknowledge ESA for providing SCIAMACHY data, KNMI TEMIS for providing FRESCO cloud data and TOSOMI total ozone data, and NASA LRAB for providing SAGE-II data. We thank Sander Slijkhuis for clarification in viewing geometry in SCIAMACHY data.