1. Inter-calibration of the SEVIRI solar bands against MODIS Aqua, using Deep Convective Clouds as transfer targets
Sébastien Wagner, Tim Hewison
In collaboration with D. Doelling and D. Morstad (NASA)
Contact:

2. Outline Context (GSICS)
Current calibration system (for the reflective bands)
Inter-calibration with MODIS using DCCs
Main steps
Some results
Conclusions and future work

3. Context... GSICS What is GSICS?
Initiative of CGMS and WMO
Effort to produce consistent, well-calibrated data from the international constellation of Earth Observing satellites
What are the basic strategies of GSICS?
Improve on-orbit calibration by developing an integrated inter-calibration system (GEO-LEO, LEO-LEO)
Best practices for prelaunch characterisation (with CEOS WGCV)
Benefits:
Improve consistency between instruments
Reduce bias in Level 1 and 2 products
Provide traceability of measurements
Retrospectively re-calibrate archive data
Better specify future instruments
EUMETSAT
CNES
JMA
NOAA
CMA
KMA
ISRO
NASA
WMO
USGS
NIST
JAXA
ROSHYDROMET
IMD
ESA

4. Context... GSICS principles
Systematic generation of inter-calibration products
For Level 1 data from passive satellite sensors (weather, climate and other environmental applications)
To compare, monitor and correct the calibration of monitored instruments to community references
By generating calibration corrections (NRT / Re-analysis applications) with specified uncertainties
Through well-documented, peer-reviewed procedures
Based on various techniques to ensure consistent and robust results
Delivery to users
Free and open access
Adopting community standards
To promote
Greater understanding of instruments’ absolute calibration
More accurate and more globally consistent retrievals
Inter-operability for more accurate environmental, climate and weather forecasting products
TRACEABILITY / UNBROKEN CHAINS OF COMPARISONS

5. GSICS in context GSICS Users GOS Level 1b/c Data (Radiances) SCOPE CM
Corrections
Statement of Needs
SCOPE CM
NWP centres
Developers of Lev2/3 products
...

6. What about the SEVIRI solar band calibration (GOS box)?
4 solar channels ( VIS06 / VIS08 / NIR16 / HRVIS )
Specifications for the solar channel calibration?
SEVIRI = 10% accuracy + 5% long-term stability
Achieved by vicarious calibration  SEVIRI Solar Channel Calibration syst. (Govaerts et al., TGRS2004)
Reference = RTM simulations of Top-Of-Atmosphere radiances (evaluated against well-calibrated polar-orbiting instruments)
Comparison with TOA measured signal above desert and ocean targets
Currently:
Meteosat 8: launch = 28/08/2002  operation = 19/01/2004
Meteosat 9: launch = 21/12/2005  operation = 18/07/2006
Meteosat 10: launch = 05/07/2012
VIS06
NIR16
HRVIS
VIS08

7. Using Deep Convective Clouds as transfer target for inter-calibration of SEVIRI with Aqua/MODIS
DCCs = invariant targets
Available all around the Earth (not the case for desert for example)  seen by all GEO satellites
Well characterized and easily detectable (infra-red threshold)
Higher part of the dynamic range
Seen by well calibrated instruments on board polar satellites  calibration transfer targets
Outcome:
Improved calibration and long term monitoring of the instruments
Consistent calibration across the GEO satellite constellation
Within the GSICS framework:
ATBD as provided by D. Doelling et al., 2011 (
Use of MODIS Aqua as a reference
Within EUMETSAT, two-fold approach:
Implementation of an inter-calibration algorithm based on the use of DCCs for VIS06
Development of a vicarious calibration algorithm to be added to the current SSCC system (absolute calibration + drift monitoring)

8. Main steps of the method as in Doelling et al, 2011
DCC identification
Time: for MET-9 (0.0 Lat / 0.0 Lon)  11:00 < t < 13:00
Geometry:
Lat / Lon between 0.0 and +/ degrees (with respect to the SSP)
SZA and VZA < 40 degrees
DCC identification  threshold using the MODIS 11m band and SEVIRI 10.8 m (BT<205K)
Spatial homogeneity (over boxes of pixels in the “11 m” BT + in the “0.6 m” radiances)
Conversion from counts to overhead sun + spectral transformation to account for Spectral Response Function differences between MODIS and SEVIRI
Use of an Angular Distribution Model (Hu model, Hu et al. 2004)
Use of correction factors as given by the GSICS ATBD (Doelling et al, 2011)
Construction of the Probability Density Functions on a monthly basis
Derivation of the gain from the calibration equation:

9. Example of DCCs tracking with MODIS and SEVIRI as implemented
Aqua/MODIS 11m
Met-9/SEVIRI 10.8m
Missing DCCS due to test on VZA
Test on the view zenith angle
MODIS Aqua (01/04/2010 – 13:00)
SEVIRI Met-9 (01/04/2010 – 12:57)

10. Preliminary results for a few cases...
MODIS Aqua DCC radiance – 0.6 m
SEVIRI Met-9 DCC counts – 0.6 m
What is wrong???
Shift in the PDFs + double peaking for 07/2008
Shift in the PDFs
Statistical uncertainty u / mean < 0.2%
With
 Sampling pb, but not only...
Derived gain for Julys
2008
2009
2011
2010
2012
1.2 %
1.7 %
-2.8%
-5.7 %

11. Conclusions
Implementation of the GSICS ATBD still on-going (uncertainty analysis still missing)
Limited to the VIS06 band
Still some issues to be solved
Questions:
What about the other visible channels (VIS08 and HRVIS)?
What about Rapid Scan Data? (limited spatial coverage)
Once problems are solved  Development of a vicarious calibration algorithm to be added to the current SSCC system (absolute calibration + drift monitoring)
For the NIR16, we would be sensitive to ice crystal. So it is more difficult to do.

12. Thank you