1. Vicarious calibration by liquid cloud target
Japan Meteorological Agency
Meteorological Satellite Center
Arata OKUYAMA
Prepared for 23 March 2011 GSICS meeting in Daejeon, Korea

2. Approach and error budget
Application to other satellites and its validation
Calibration monitoring Web page

3. Calibration Methodology
Vicarious calibration
Rebuild visible calibration table,
Based on comparison of observation with simulation.
Radiative transfer calculation
To make simulation accurately, RT calculation is done on temporally and spatially stable area
= Reference target
The reference target should be over wide range of brightness to obtain reliable regression line.
Cloud-free ocean : Dark target
Cloud-free land : Medium brightness target
Uniform liquid cloud top : Bright site
(Deep convective cloud top : Brightest site)
Observed
Calculated
Liquid cloud

4. for liquid cloud target
Necessary data
for liquid cloud target
Tools
RT model (RSTAR)
Cloud property retrieval tool (CAPCOM)
Input factors to RT model
Sun and satellite geometry
Atmospheric profile
WV, Pressure, Temperature (JRA-25 or JCDAS)
Ozone (TOMS)
Surface condition
Sea surface wind speed (JRA-25 or JCDAS)
Scattering particle
Cloud (retrieved from MODIS L1B)
Error amounts for each factor are estimated, and summed up. 4

5. Error budget for liquid cloud target
Random error
Systematic error
Spectral response function (MODIS and MTSAT)
Stratospheric aerosol
Factor
RTM input parameter
Typical uncertainty
Radiance uncertainty [%]
Solar and sensor geometry
Sat zenith angle
0.2 deg
0.2
Solar zenith angle
~ 0 deg
~ 0
Relative azimuth angle
Atmospheric profile
Ozone
2 %
0.1
Water vapor
Assumed 30 %
Surface
Sea surface wind speed
Cloud parameters
Optical thickness 8%
3.8
Effective radius 4%
Total
RSS
MODIS L1B

6. Systematic error ? Calibration coefficients based on two MODISs
MTSAT-2 Calibration coefficients by MODISs
Calibration coefficients based on Aqua/MODIS and Terra/MODIS are not same.
It means that cloud optical parameter retrieved from Terra or Aqua MODIS L1B contains systematic error.
Dave pointed out Terra/MODIS has degradation trend relative to Aqua/MODIS.
(on Dec GSICS Web meeting)
slope
intercept
2010
2011

7. Validation by aerosol product Aqua vs Terra
Comparison of satellite retrieved aerosol with ground observation
Calibrate satellite data using Terra/MODIS or Aqua/MODIS
Retrieve aerosol optical thickness (AOD) from calibrated data
Compare AOD with ground observation data
Calibrated data based on Aqua/MODIS seems to be better than the one based on Terra/MODIS.
It means calibration coefficients based on Terra/MODIS have some systematic error. But it is difficult to estimate the error amount.
Based on Terra
Based on Aqua
Ground obs.
Sat
RMS
2010 Jul.-Aug.-Sep.
Terra : 0.960
Aqua : 0.851

8. Approach and error budget
Application to other satellites and its validation
Calibration monitoring Web page

9. Application to GOES and METEOSAT
GOES and METEOSAT data in 2002 is calibrated by this approach.
The approach seems to work fine.
September 2002
GOES-8
GOES-10
METEOSAT-5
Radiance(W/m2/sr/um)
METEOSAT-7 DN 9
Collaborate research of Chiba University and JMA/MSC

10. Retrieved downward SW flux at ground surface
2002/09/
Averaged Downward SW flux at ground surface
Downward SW flux at the surface
METEOSAT-7
METEOSAT-5
GMS-5
GOES-10
GOES-8
0 deg.
63 deg.
140 deg.
225 deg.
285 deg.
EXAM SYSTEM [Takenaka et al., 2011, JGR-atmosphere] 10

11. Validation of the flux by BSRN and SKYNET
METEOSAT-7(322.5E-31.5E)
METEOSAT-5(31.5E-101.5E)
GMS5(101.5E-182.5E)
GOES-8(225E-345E)
GOES-10(165E-285E)

12. Approach and error budget
Application to other satellites and its validation
Calibration monitoring Web page

13. MTSAT-2 Calibration Monitoring page
Renewed in July 2010
Click logo
MTSAT-2 IR opened in July 2010
Jump to GSICS Portal (WMO)
MTSAT-2 VIS opened in Sep. 2010

14. MTSAT-2 IR Calibration Monitoring page
Regression Coefficients of
MTSAT Digital Counts
as a function of Hyper Sounder Radiances
Intercept
Slope

15. MTSAT-2 IR Calibration Monitoring page
TB Difference at Standard Radiance
Introduced
GSICS Correction
(Re-Analysis Corr.)
30-day Moving Ave. Introduced

16. MTSAT-2 Visible Calibration Monitoring page
Monthly statistics and time sequences
Scatter plot
Regression Coefficients

17. MTSAT-2 Visible Calibration Monitoring page
JMA GSICS Top Page
Guide
Scatter plot
Regression Coefficients (Time Series) 17

18. Summary Error budget for the liquid cloud target is estimated.
Terra/MODIS based calibration coefficients seems to have bias. It is important to estimate systematic error amount.
Vicarious calibration approach is applied to GOES and METEOSAT. The result is verified through the solar flux product as an example of validation.
Calibration monitoring Web site is available on

19. Thank you !

21. Comments
If we utilize different types of targets, how to determine weight ?
If simulations on each target have different systematic error, how combine them ?
How to validate calibration coefficients ?
Are update of the coefficients necessary ?

23. Consistency check of the approach
The radiance simulation approach is evaluated by using MODIS data.
MODIS carries onboard calibrators for visible bands and it is well-calibrated. Its observations are reliable.
Computation and observation show good consistency
RMSE is around 1%
MODIS
Radiance comparison
Liquid Cloud
MODIS L1B
Aerosol
Cloud
Sim. Radiance
retrieve
RT calc
compare
Atmos. prof.
Simulated radiance
Cloud-free Land
Cloud-free Sea 23
MODIS L1B (Band-1) 23