1. Using SCIAMACHY to calibrate GEO imagers
D. Doelling
NASA LaRC
Benjamin Scarino, Dan Morstad, Arun Gopalan
SSAI
GSICS/GWRG Session Beijing, China, March 5-8, 2012

2. Background
In a perfect world, GSICS would use a well-calibrated visible hyper-spectral sensor to calibrate the GEO imagers
The CLARREO SW instrument will be absolutely calibrated to 0.3% at hyper-spectral wavelengths with a 100x100km footprint
Earliest CLARREO instrument will be 2022
The sensor that most resembles CLARREO visible hyper-spectral sensor would be SCIAMACHY
0.3µm to 2.1µm range to derive trace gas concentrations
30x240km footprint
Calibrated using a solar diffuser
Absolute calibration uncertainty between 2% to 6% depending on wavelength
Can the current operational SCIAMACHY be used as a calibration reference sensor?

3. Outline Use Aqua-MODIS as the visible reference for all comparisons
Aqua-MODIS more stable than Terra-MODIS
Not based on if Aqua or Terra-MODIS is brighter
Inter-calibrate SCIAMACHY and Aqua-MODIS 0.65µm channel using NSNOs
Determine SCIAMACHY stability compared against Aqua-MODIS
Determine relative calibration difference
Inter-calibrate GEO with SCIAMACHY pseudo (GEO SRF) radiances using ray-matching
Validate with ray-matched Aqua-MODIS and GEO calibration

4. SCIAMACHY Aqua-MODIS NSNO (top view)
ENVISAT
ENVISAT ground track
Aqua ground track
SCIAMACHY footprint centers
SCIAMACHY NSNO footprint
NSNO footprint
SCIAMACHY earth view
SCIAMACHY limb view
Aqua
• NSNO local time is 11:45AM, near local noon, assuring azimuthal match
• The Envisat and Aqua ground intersect is at 71°N and occurs 14 x daily
• Perform monthly regressions of SCIAMACHY and Aqua radiances between April and September

5. SCIAMACHY Aqua-MODIS 0.65µm, July 2010N
Force
Stderr 2.8%
SCIAMACHY pseudo Aqua-MODIS radiance
Aqua-MODIS radiance
• 15 minute coincident, <70° SZA
• ~ km sub-sampled MODIS pixels are averaged into a 30x240km SCIAMACHY footprint

6. SCIAMACHY Aqua-MODIS 0.65µm, 2004-2010
• Monthly standard deviation of force fits = 0.44%
• Relative temporal trend of -0.2%/decade
• SCIAMACHY is temporally calibrated as well as Aqua-MODIS
• both use solar diffusers for on orbit calibration
• both are within their uncertainty of 2% at 0.65µm

7. GEO SCIAMACHY precise ray-matching (top view)
GEO sub-satellite point
Envisat ground track
22.5° GEO VZA
240 km
30 km
7.5° VZA
22.5° VZA
40 km radii
ray-match
tolerance

8. GEO SCIAMACHY precise ray-matching (side view)
240 km
80 km diameter ray-match tolerance
GEO satellite
Envisat satellite
SCIAMACHY footprints
Earth surface
2° GEO view angle range
15° SCIAMACHY view angle range
22.5° VZA
Enivisat ground-track

9. GEO SCIAMACHY precise ray-matching
Precise ray-matching yields ~40 matches per year
ENVISAT has a 35 day repeat cycle
SCIAMACHY has a 50% Earth view duty cycle
Each SCIAMACHY footprint location is matched once a month
Expand tolerance radii
Precise ray-match => 40-km radii
Approximate ray-match => 160-km radii
Does off nadir ray-matching introduce angular view biases?
Test using GEO radiance calibrated against Aqua-MODIS
Can’t verify if sensor over its lifetime if it is degrading

10. Aqua-MODIS tp Meteosat-9 spectral band adjustment factor (SBAF)
Slope N
Stderr 0.28%
• Bands are similar
• Most of the slope is from solar constant difference
• Uncertainty due to SBAF is 0.28%
• Regress SCIAMACHY Aqua-MODIS and Meteosat-9 pseudo radiances over GEO/LEO equatorial ocean matching domain
• Takes into account both spectral response function and solar constant differences

11. Aqua-MODIS/Meteosat-9 raymatching
September 2009
Met-9/Aqua
Force Fit 0.547 N
Stderr %
Trend 1.4%/decade
Stderr 0.53%
• temporal uncertainty of ray-matched calibration is 0.53%

12. SCIAMACHY Meteosat-9 0.65µm ray-match
precise
approximate
Force fit 0.996 N
Stderr %
Force fit 0.991 N
Stderr %
Meteosat-9 radiances based on ray-matching
SCIAMACHY pseudo Meteosat-9 radiances
All matches between
• The force fits are within 0.5%, well within the uncertainty of 4%
• 13 times more points with approximate method

13. SCIAMACHY clusters
SCIAMACHY has 56 Earth view clusters, each with its own integration time, footprint resolution, and spectral range
Cluster for trace gas retrievals have highest spectral resolution
Meteosat-9 spectral response function (SRF) is contained nearly in cluster #24 spectral range
Cluster #24 has a footprint size of 30x60km, ¼ of SCIAMACHY
Spectral range of 0.613μm to 0.726μm  at  20nm resolution
Compute lost energy adjustment factor (LEAF)
Regress SCIAMACHY Met-9 SRF convolved pseudo radiances with Met-9 SRF limited by clustert #24 spectral range
Use all the approximate case footprints
The single LEAF value was 0.96

14. Meteosat-9 0.65µm spectral response function
The out of band energy ~4% based on LEAF

15. SCIAMACHY Meteosat-9 0.65µm ray-match
precise
approximate
Force fit 0.996 N
Stderr %
Force fit 0.991 N
Stderr %
Meteosat-9 radiances based on ray-matching
SCIAMACHY pseudo Meteosat-9 radiances
All matches between
Approximate cluster
Force fit 0.991 N
Stderr %
With LEAF
• All force fits are within 0.5%, well within the uncertainty of 4%
• LEAF is a single value and probably adds noise to the regression

16. GEO and Aqua-MODIS spectral response functions
• Now test the ray-matching with several GEOs and bands

17. Comparison of precise and approximate SCIAMACHY/Met-9 force fit 2007-2010
• All force fits are within 0.5%, well within the uncertainty of ~4%
• All approximate ray-matching is not adding any angular biases

18. Perform calibration validation
• If Aqua-MODIS/SCIAMACHY and SCIAMACHY/GEO, where GEO was calibrated against Aqua-MODIS, ray-matches were all performed perfectly then the following ratio would equal 1.0
• If the ratio is within the uncertainty than the methodology is validated
• The total uncertainty is the square root of the sum of the squares of the uncertainties of the individual terms
• All terms are temporal regression standard errors,
• For Met µm/SCIA it is 0.35%
Met-9/SCIA 0.65µm
Approximate footprint
annual means
Mean force fit 0.991
Trend 0.19%/year
Stderr 0.35%

19. Ray-matching calibration validation
Left column total uncertainty without SBAF, right column with SBAF
• Most ratios within 1% and all within the total uncertainty
• Validating that all three ray-matches are consistent
• Need to revisit the Met µm Aqua/Met-8 calibration, however this band has a large SBAF uncertainty since it is in the near IR absorptions bands

20. Operational SCIAMACHY Meteosat-9 0.65µm calibration
Perform 3-monthly (monthly) approximate SCIAMACHY-radiance/Met µm count footprint (cluster) regression force fits to compute the gain
Adjust the SCIAMACHY calibration to match Aqua-MODIS
Compare the calibration gains with other methods
All other methods use Aqua-MODIS as calibration reference
Aqua-MODIS/Met-9 ray-match (shown earlier)
Terra-MODIS adjusted first to Aqua-MODIS Met-9 ray-match using NSNOs over polar region
Libyan Desert
DCC

21. Meteosat-9 0.65µm gain comparison
• SCIAMACHY/Meteosat µm calibration is within 1% of all other calibration coefficients
• Has one of the lowest monthly uncertainty at ~0.55%
• SCIAMACHY based GEO calibration as good as other methods and has the advantage of not need any spectral corrections
• The calibration can be used to verify spectral corrections of other methods

22. Conclusions
SCIAMACHY can either validate an existing GEO calibration or if enough sampling is available operationally calibrate GEO visible sensors
SCIAMACHY/Meteosat µm calibration is within 1% of all other calibration coefficients
Has one of the lowest monthly uncertainty at ~0.55%
This calibration technique can be applied to future CLARREO instrument if visible pointing is not possible
SCIAMACHY is temporally calibrated as well as Aqua-MODIS
Monthly uncertainty of 0.44%
A relative trend of 0.2%/decade