Validation of Edition 1 and V3 GERB products by comparison with CERES

1. Validation of Edition 1 and V3 GERB products by comparison with CERES S. Dewitte Royal Meteorological Institute of Belgium GIST, Exeter, 10/2006

2. Overview • Radiance comparisons ARG, SHI, BARG • Scene type dependence SW ARG • Radiance, Flux, ADM • LW ARG Flux • Regional distribution (cold cloud, hot desert) • Empirical correction GIST, Exeter, 10/2006

3. Used data • GERB, Edition 1/V3 , 3 space-time products: • ARG (E1): GERB PSF (basic 45 km), 17’ • SHI (V3): 9 km, 15’ (SEVIRI is used) • BARG (V3): 45 km, 15’ • CERES, ES-8 Edition 1-CV + in-flight gain changes + Revision 1 for SW gain • FM1,FM2,FM3: All cross-track ! • Time period: 25/3-30/4/2006 GIST, Exeter, 10/2006

4. GIST, Exeter, 10/2006

5. ARG and BARG agree within 0.2 % • SHI 0.7 to 0.9 % higher than ARG GIST, Exeter, 10/2006

6. GIST, Exeter, 10/2006

7. ARG and BARG agree within 0.5 % • SHI 0.8 to 1.6 % lower than ARG • Day night differences vary with CERES instrument, largest (1.7 to 2.5 %) for FM2 GIST, Exeter, 10/2006

8. Space-time conversions GERB-like 6’ GERB PSF 6’ HR 6’ 45km L,F C L,F ARG SHI BARG 15’ snapshot 15’ mean 17’ mean L,F: temporal averaging radiances and fluxes C: temporal averaging GERB/GERB-like GIST, Exeter, 10/2006

9. Interpretation • ARG & BARG agreement: validates spatial transformations • Small SHI differences: due to temporal differences • Snapshot <-> Mean: not ‘real’ GERB error • Interpolation of C (SW) GIST, Exeter, 10/2006

10. Scene type dependency SW ratios • Clear sky: t < 1 • Ocean, Dark/Bright Vegetation, Dark/Bright Desert • Thin clouds: 1 < t < 3.16 • Medium clouds: 3.16 < t < 10 • Thick clouds: 10 < t GIST, Exeter, 10/2006

11. ARG SW Rad. G/C GIST, Exeter, 10/2006

12. ‘Good’ values for thick clouds: • GERB 1.3 % higher than CERES FM1 • GERB 2.3 % higher than CERES FM2 & FM3 • Strong dependence ratio on cloudiness: spectral response problem GIST, Exeter, 10/2006

13. ARG SW Flux G/C GIST, Exeter, 10/2006

14. Bright scenes: similar to radiances -> dominated by spectral error • Cs ocean: increase due to ADM error • Cs dark vegetation: decrease due to ADM error GIST, Exeter, 10/2006

15. ARG SW ADM GIST, Exeter, 10/2006

16. Cs ocean ADM error: +2.9 to 4.8 % • Cs dark vegetation ADM error: -4.9 to -6.6 % • Other scenes: ADM OK within 2 % • Medium clouds: ADM OK within 0.4 %, best case ! GIST, Exeter, 10/2006

17. Day, FM1 GIST, Exeter, 10/2006

18. Night, FM1 GIST, Exeter, 10/2006

19. Conclusions GIST, Exeter, 10/2006

20. Day, FM1: Before/after correction GIST, Exeter, 10/2006

21. Night, FM1: Before/after correction GIST, Exeter, 10/2006

22. Conclusions • ARG & BARG agree, ARG & SHI agree within 1 %. • Strong dependence of GERB/CERES SW ratios on cloudiness, good agreement (1.3 -2.3 %) for thick clouds. • SW ADM errors for cs ocean & cs dark vegetation 3.5%. • Empirical correction for LW fluxes advised. GIST, Exeter, 10/2006

23. Level 2 Release paper • S. Dewitte, L. Gonzalez, N. Clerbaux, A. Ipe, C. Bertrand, “The Geostationary Earth radiation Budget edition 1 data processing algorithms”, submitted to Advances in Space Research. GIST, Exeter, 10/2006

24. Possible improvements • Better spectral response • New ground measurements • In-flight measurement: comparison with Sciamachy ? • Derive new surface SW ADM’s ocean + dark vegetation • Clear-sky: feasable to derive from geostationary • Derive NB+BB: no propagation errors GERB-like -> GERB • Derive empirical corrections to LW ADM’s using only GERB + Meteosat data. GIST, Exeter, 10/2006