1. COMS Visible Channel Calibration Dohyeong KIM Ho-Seung LEE Won-Seok LEE Tae-Hyeong OH Sunmi NA National Meteorological Satellite Center Korea Meteorological Administration Republic of Korea GSICS Annual Meeting(Williamsburg, Virginia / 2013. 3. 4-8) 1

2. Contents Data Method Results Summary Issue 2

3. Data target moonoceandesertcloud period Feb. 2011~Dec. 2012 Aug. 2011~Dec. 2012 obs.Local Area observation mode Full Disk observation of Earth with moon Pacific Ocean Indian Ocean Simpson Desert in Australia 0.5° surface 3

4. Moon irradiance computed from the ROLO where D m-v : the moon-viewer distance in km i.e. Moon-satellite D m-s : the moon-sun distance in AU Method (moon) *ROLO: Robotic Lunar Observatory *USGS: US Geological Survey COMS Moon observation L1A data COMS ephemerides data Moon mask extraction ROLO* model of USGS* Moon irradiance as measured by the COMS/MI where N P : the total number of pixels in the lunar disk image R P : an individual radiance measurement(i.e. pixel) on the Moon Ω: the solid angle of a pixel (sr), corresponding to one visible detector, i.e. 28 µrad 16/28: the over-sampling rate of MI along E-W axis Ref : COMS Radiometric Calibration Document 4

5. Method (Desert) COMS data Target selection MODIS BRDF Parameter(2004~2006) 11 targets in Simpson desert of Australia Simulated TOA Radiance NameLatitudeLongitude A-26.275136.625 B-25.775136.825 C-25.775137.975 D-25.775137.125 E-25.625136.825 F-25.625137.975 G-25.475136.825 H-25.475137.975 I-25.475137.125 J-25.625137.875 K-25.325137.575 11 targets RTM Input Spectral BRDF - MODIS BRDF Parameter - ASTER Spectral albedo NCEP TPW OMI O 3 Developed by Dr. Chun (SNU) 5

6. Method (Water Cloud) 0.5°х 0.5° Grid Format COMS 0.675-μm Reflectance MODIS Cloud Products 0.5°х 0.5°Grid Format Thresholds of Cloud Targets Over ocean regions Overcast clouds 5 ≤ Cloud optical thickness (COT) Matching Time & Location ∆Time ≤ 5 minutes RTM Inputs Surface – Ocean BRDF Atmosphere – Tropical profile Cloud – MODIS COT, particle size (r e ), Cloud altitude Mie scattering model for CTT ≥ 273 K Baum scattering model for CTT ≤ 227 K SBDART 3D Radiative Effects Developed by Dr. Ham (SNU) 6

7. Method (Deep Convective Cloud) COMS 0.675-μm Reflectance COMS 10.8-μm Brightness Temperature RTM Inputs Surface – Ocean BRDF Atmosphere – Tropical profile Cloud properties - COT = 200, particle size (r e ) = 20 μm Baum scattering model SBDART 3D Radiative Effects Thresholds of DCCs (1) Cold cloud condition to select clouds overshooting tropopause layer - TB 10.8 ≤ 190K where TB 10.8 : Brightness Temp. at 10.8-µm band (2) Homogeneity checks to exclude cloud boundary pixels - STD(TB 10.8 ) ≤ 1K - STD(R 0.6 )/Mean(R 0.6 ) ≤ 0.03 where R 0.6 : Visible reflectance Developed by Dr. Ham (SNU) 7

8. Results (Scatter Plot) (1/2) ISSUE 1.1 Do you know why? 8

9. Results (Scatter Plot) (2/2) (W·m -2 ·sr -1 ·μm -1 ) 9

10. The mean values of radiance ratio with desert/cloud targets are all about -8% from August 2011 to December 2012. The mean value of irradiance ratio with moon target is about -5% from February 2011 to December 2012. Results (Time Series) (1/3) 10

11. The mean value of reflectance ratio with desert target is about +2 % (positive). In case of cloud target, the mean values are in the range of -9 to -11% (negative). Results (Time Series) (2/3) ISSUE 1.2 Any comments on positive value? 11

12. The reflectance ratio with WC target is consistent with the radiance ratio with it. But the reflectance ratio with desert/DCC targets are inconsistent with the radiance ratio. Results (Time Series) (3/3) 12

13. Summary The observed values with ocean/desert/cloud targets are correlated with the simulated values(R=0.99) from August 2011 to December 2012. The mean values of radiance ratio with desert/cloud target are all about -8% from August 2011 to December 2012. The mean value of irradiance ratio with moon target is about -5% from February 2011 to December 2012. The results of time series with desert/cloud/moon targets seem that there is no great change for confirming existence of performance degradation during calibration period. 13

14. Issue 1.Desert/Cloud/Ocean targets 1.1 [Scatter plot] The reason why the unnormally data with large bias? 1.2 [Time series] The reason why the reflectance ratio with desert target is positive value? 2.Moon target 2.1 In case that the moon is not full moon(decrescent, increscent), how to process the moon mask extraction in your own? 2.2 How to check the latest version of ROLO model? (see next slide!!) ? 14

15. ROLO model used in KMA(1/2) Ref : COMS Radiometric Calibration Document 15

16. ROLO model used in KMA(2/2) Ref : COMS Radiometric Calibration Document ISSUE 2.2 Is this the latest version of ROLO model? Who can help to check it? 16

17. Thank you

18. Back-up slides

19. S-/L-/Ka-Band Comm Meteo Imager Ocean Color Imager Solar array Bus ■ Communication, Ocean, and Meteorological Satellite (COMS) ■ 8-year program (2003 – 2010) ■ Launch date : June 26, 2010 (French Guiana Kourou) ■ Location : 128.2E ■ Mass : 2,500kg ■ Design life time : 7 years ■ Contractor : EADS Astrium (ITT for MI) COMS channel Wave length( ㎛ ) Visible 0.67 Shortwave IR(IR4) 3.7 Water Vapor(IR3) 6.7 IR1 10.8 IR2 12.0

20. Full Disk : every 3 hr NH : every 15 min Korea : ~ 8 min FD LA ENH Observation Schedule

21. Atmospheric Model Ozone(OMI) Total Precipitable Water Ocean Model Wind Speed Wind Direction Pigment Concentration (Sea WiFS monthly climatology) Radiative Transfer Model(6S) Ocean Target Desert Target Radiative Transfer Model(6S) Atmospheric Model Ozone(OMI) Total precipitable Water MODIS BRDF COMS geometrical information MODIS geometrical information COMS Observation data COMS geometrical information Aerosol Model AOT(aerosol optical thickness) at 550 nm(MODIS)

22. Water Cloud Deep Convective Cloud COT(MODIS) Cloud Optical Thickness 0 ≤ COT ≤ 100 CTT(MODIS) Clout Top Temperature 150 ≤ CTT ≤ 350 CTP(MODIS) Cloud Top Pressure 1 ≤ CTP ≤ 1100 SBDART MODIS geometrical information COMS Observation data COMS geometrical information COT(MODIS) Cloud Optical Thickness 100 ≤ COT ≤ 400 SBDART MODIS geometrical information COMS Observation data COMS geometrical information

23. Results (ocean/desert/cloud) (2/2) Ongoing project

24. Results (ocean/desert/cloud) (1/2) Ongoing project

25. To measure the total degradation of the instrument visible channel Comparison between the Moon signal and ROLO model(moon irradiance computation model) of USGS – as a Function of Phase angle(Sun-Moon-Earth) Imager Response Where is the Moon irradiance as measured by the Imager is the Moon irradiance computed from the ROLO model, under the same conditions(phase angle, positions of sun, moon, satellite, etc.) 11 Method (moon)

26. Comparison - COMS/MI Moon observation vs. ROLO* model of USGS* COMS/MI Response Ratio *ROLO: Robotic Lunar Observatory *USGS: US Geological Survey Where is the Moon irradiance as measured by the COMS/MI is the Moon irradiance computed from the ROLO model Trend (in percentage)

27. Trend shows that there is about 3.4% COMS visible channel degradation from Feb 2011 to May 2012. (1-k)×100(%) TIME(month) Change of Ratio k Trend For obtaining more precise result, We need more data through the long term observation. (1-k)×100(%) 12

28. Results (moon) month KMA has been testing COMS visible channel calibration using moon for long-term monitoring. k (%) ADD ISSUE Seasonal variation? Starting warm season Starting cold season