1、China Radiometric Calibration Site(CRCS) and Global Sites Calibration for FY3 MERSI RSBs,Ling SunNSMC, CMA2012-3-7,AcknowledgementsNa Xu, Jing-Jing Liu, Xiu-Qing Hu,Dunhuang CRCS Vicarious Calibration Method, Result, MODIS-based Evaluation Using SNOGlobal Sites Calibration TrackingMethod, Result, MO
2、DIS-based Evaluation Using SNOSummary,Outline,Formula of the FY-3 MERSI Reflectance Calibration:,Background,MERSI RSB Spectral Specifications,FY-3A (launched on May 27, 2008) ,local equator-crossing time of 10:30 A.M. (descending southward) FY-3B (launched on Nov 5, 2010), local equator-crossing tim
3、e of 1:30 P.M. (ascending northward),(0100%),FY-3A MERSI,FY-3B MERSI,There are anomaly electronic gain jumps in SWIR bands 6 and 7 possibly induced by electrostatic discharge.,The antifouling cover of space radiant cooler was not successfully opened. Thus, bands 6 and 7 are in abnormal working condi
4、tions without radiant cooler since Jan. 21, 2011.,Dunhuang CRCS Vicarious Calibration,FY-3A MERSI imageCRCS Dunhuang,Reflectance-based method Radiance-based method,BRDF in 2008,Reflectance-based VC,RTM: V6S and MODTRANAerosol model: continental or desert modelSurface reflectance bidirectional model:
5、 BRDF model AMBRALS Parameters corrected with real-time surface reflectance and convolved using SRF,Coefficient of Variation of FY-3A MERSI,Calibration uncertainties (CV= Std/Mean) are within 3% for most bands.,There are anomaly electronic gain jumps in SWIR bands 6 and 7 possibly induced by electro
6、static discharge.,PDif% between simulated Aqua MODIS TOA radiance and sensor measurements from 2008 to 2011,Bands at 645, 858, 469, 555, 1240, 1640, 2130, 412, 443, 490 and 905 nm. Over estimation for bands 4(0.555 m), 7(2.13 m) and 17(0.905 m).The average differences are within 3% for window bands
7、with wavelengths less than 1 m.,Radiance-based VC,MODIS-based CRCS Calibration Evaluation,SNO data screening criteria:,Band match-ups:,Nadir overpass time difference: within 10 minutes. Mapped in 10241024 using Lambert projection centered at nadir. Angle requirement: SenZ35, SolZ90, |cos(SenZMODIS)
8、/ cos(SenZMERSI) 1| 0.01.Space uniformity requirement : CV2% in each 44 box.,470nm,550nm,650nm,865nm,412nm,443nm,490nm,520nm,565nm,650nm,685nm,765nm,865nm,905nm,940nm,1640nm,2130nm,Difference between Re-calibrated FY-3A MERSI ARef(0100%) and MODIS using SNO,Dunhuang (40.138N, 94.32E),Libya1 (24.42N,
9、 13.35E),Libya4 (28.55N, 23.39E),Arab2 (20.13N, 50.96E),Lanai (20.49N, -157.11E),Sites with stable surface properties : Gobi and desert targets: Dunhuang, Libya1, Libya4 and Arabia2, ocean site: Lanai (MOBY).,Global Sites Calibration Tracking,TOA radiation calculation similar as CRCS VC, but without
10、 in-situ measurements AOD550nm： MODIS monthly aerosol product Aqua deep blue for land , Terra for ocean Ozone amount: climatological monthly mean from TOMS Water vapor amount: climatological monthly mean from NCEP Surface directional reflectance: MODIS BRDF products for land sites MOBY measurements
11、for ocean site,Data within 10 days from 5 sites are used to get the calibration slope.,Calibration Coefficient Trending,(0.470, 50),(0.550, 50),(0.650, 50),(0.865, 50),Blue dot for CRCS VC,(0.412, 20),(0.443, 20),(0.490, 20),(0.520, 20),(0.565, 20),(0.650, 20),(0.685, 20),(0.765, 20),(1.03, 20),(0.8
12、65, 20),(0.905, 20),(0.940, 20),(0.980, 20),The mean percent difference between estimated coefficients and CRCS VC are below 3.5% (except for bands 14, 15 1719).,PDif% between Calibration Updating Model Estimationand CRCS VC,Normalized Response Change Trending,(0.470, 50),(0.550, 50),(0.650, 50),(0.
13、865, 50),(0.412, 20),(0.443, 20),(0.490, 20),(0.520, 20),(0.565, 20),(0.650, 20),(0.685, 20),(0.765, 20),(1.03, 20),(0.865, 20),(0.905, 20),(0.940, 20),(0.980, 20),Till Dec. 31, 2011,FY-3A MERSI Response Change, 900 nm: also noticeably response decrease; 600 to 900 nm: relatively stable with the ann
14、ual degradation rates within 1% ;B3(650nm): noticeably response increase over 1% in 4 years.,FY3B MERSI Calibration Coefficient Trending,470nm,550nm,650nm,865nm,412nm,443nm,490nm,520nm,565nm,650nm,685nm,765nm,865nm,905nm,980nm,940nm,103nm,Percentage differences 3.5% for most bands with the exception
15、 of band 8 (412 nm), 6 (1.6 m) and 7 (2.1 m).,PDif% between Calibration Updating Model Estimation and CRCS VC,Aug.,2011,Response Change Trending,FY-3B MERSI Response Change,Till Dec. 31, 2011,Similar pattern as FY-3A MERSI. Bands 3 (650nm) and 15 (765nm) with noticeably response increase over 1% per
16、 year.,MODIS-based Post-launch Calibration Updating Evaluation,470nm,550nm,650nm,865nm,412nm,443nm,490nm,520nm,565nm,650nm,685nm,765nm,865nm,905nm,940nm,1640nm,2130nm,Difference between Re-calibrated FY-3A MERSI ARef(0100%) and MODIS using SNO,No need to do the correction with CRCS VC.,FY-3B,470nm,5
17、50nm,650nm,865nm,412nm,443nm,490nm,520nm,565nm,650nm,685nm,765nm,865nm,905nm,940nm,1640nm,2130nm,CRCS VC Corrected,No need to do the correction with CRCS VC except for band 8.,2011-2,FY-3A MERSI,Pre-launch Calibration Coefficients Usability,The pre-launch calibration coefficients are obviously less
18、than those of post-launch initials for most bands except for bands 3, and the differences between them exceed 9% except for bands 11 and 20.,The pre-launch calibration coefficients are obviously less than those of post-launch initials for most bands except for bands 14, 17 and 18, and the difference
19、s between them exceed 5% except for bands 10, 15, 17 and 19. The pre-launch calibration coefficients could not be used after launch. The pre-launch calibration accuracy needs to be improved.,FY-3B MERSI,CRCS vicarious calibration: The uncertainty is within 3% for most bands. The method is validated
20、using MODIS, and the TOA reflectance PDif is almost 3% for window bands with wavelengths less than 1m.Post-launch calibration updating using global sites: Daily-based updating model using a linear function of DSL. SNO analysis shows good agreement between re-calibrated MERSI TOA apparent reflectance
21、 and MODIS measurements.The on-orbit response changes are wavelength dependent. Bands with wavelengths 900 nm also show noticeably response decrease; most bands from 600 to 900 nm are relatively stable. Pre-launch calibration should be intensively investigated. More validation should be taken, such as DCC.,Summary,Thank you,B8,B10,B11,B6,B7,B14,B15,B17,B16,B19,B18,B20,
