1. CERES-GERB meeting, Boulder, 3/2004 Validation of GERB unfiltered radiances S. Dewitte Royal Meteorological Institute of Belgium

2. CERES-GERB meeting, Boulder, 3/2004 Overview u GERB-CERES comparison –Reflected solar radiance –Emitted thermal radiance - Night u GERB-SEVIRI comparison –Day-night consistency

3. CERES-GERB meeting, Boulder, 3/2004 GERB-CERES: Used data u 19/12/2003-29/2/2004 –Only night time used for thermal comparisons u GERB: L2 ARG –Unfiltering with SEVIRI –Version 2 : geolocation from matching u CERES FM2 –GERB mode, RAPS, Principal plane scan –Edition 1: pre-flight gains u Coangular: Difference viewing direction < 20 deg.

4. CERES-GERB meeting, Boulder, 3/2004 Reflected solar radiance comparison u Clear sky/overcast separated by cloud amount C=(L-Lcs)/(Lov-Lcs) –CS: C 50% u Clear sky separated by ADM surface type –Ocean, dark/bright vegetation, dark/bright desert

5. CERES-GERB meeting, Boulder, 3/2004

6. 95% confidence intervals u GERB /CERES reflected solar radiance= Ocean: 0.931+/-0.009=[0.922-0.941] Dark veg.:0.951+/-0.006=[0.944-0.957] Bright veg.:0.969+/-0.011=[0.958-0.980] Dark desert:1.001+/-0.010=[0.991-1.011] Bright desert:0.980+/-0.006=[0.974-0.986] Clouds:0.959+/-0.009=[0.950-0.968]

7. CERES-GERB meeting, Boulder, 3/2004 Spectral Response

8. CERES-GERB meeting, Boulder, 3/2004

9. u CERES FM2 LW 0.9 % too high

10. CERES-GERB meeting, Boulder, 3/2004 95% confidence interval u GERB/(ed. 1 CERES) emitted thermal radiance= 0.988+/-0.007=[0.982-0.995] u After in-flight correction CERES= 0.997+/-0.007=[0.991-1.004]

11. CERES-GERB meeting, Boulder, 3/2004 GERB LW day night consistency u (GERB LW)/(SEVIRI BB LW) known for every ARG pixel u Calculate mean for every image u Plot mean ratio as function of hour of day u Time interval: 12/2003-2/2004

12. CERES-GERB meeting, Boulder, 3/2004

13. Day-night consistency u Discontinuity of 0.5 % between 5:45 and 6:00 UTC –Origin: geolocation from matching ? u Off-scale values around midnight –Origin: straylight ? u Mean daily variation of order 0.5 % -> OK

14. CERES-GERB meeting, Boulder, 3/2004 Conclusions u GERB/CERES reflected solar radiance differences are within target accuracy only for clear sky desert. –Cause: Spectral response ? u GERB/CERES night time emitted thermal radiance differences are within target accuracy. u GERB/SEVIRI emitted thermal radiance differences show day/night consistency within 0.5 % –Exception: stray light ? u Next steps: –Check reflected solar radiance differences with updated spectral response. –Validate BARG, SHI unfiltered radiances.