1. Status and plans for AMSR-E calibration Keiji Imaoka JAXA Earth Observation Research Center (EORC) X-CAL Meeting University of Central Florida, Orlando June 30, 2010

2. AMSR-E Calibration Status Last updated in 2005 (Ver.2), different from RSS calibration for US L2A. Derivation of effective temperature of HTS (warm load) based on regression equation of PRTs and utilization of receiver physical temperature/gain relationship. Empirical correction of possible non-linearity of receivers. In doing above radiometric corrections, we have relied on cross calibration with SSM/I and TMI, and comparison with geophysical field from other data sources (e.g. Reynolds SST). Currently no scan bias correction, except for 30-points scan edge correction for 6.9GHz channels. Known problem with geolocation errors (affects incidence angle).

3. Multiple regression model of T eff using eight PRT readings. Coefficients of the regression model were determined by using SSM/I oceanic Tb (18GHz and higher channels) and computed Tb (6 and 10GHz channels) based on the Reynolds OI-SST analysis. SSM/I data were provided by the Global Hydrology Resource Center (GHRC) at the Global Hydrology and Climate Center, Huntsville, Alabama, USA. Reynolds OI-SST dataset were made available by NOAA. Utilize Relationship between receiver temperature and its gain variation. Applying this equation to HTS measurement and assuming T eff derived by regression model as T OBS, b RX can be computed by regression analysis. Using this value, gain variations can be compensated by the equation. Radiometric correction : Two steps PRT readingsHTS Effective Temp. T OBS : Scene Tb (K) T CSM : Deep space Tb (K) C ’ OBS : Digital counts of scene C ’ CSM : Digital counts of deep spece G 0 : Nominal gain b RX : Gain sensitivity to rec. temp. ( ℃ -1 )  T RX : Rec. temp. departure from mean value ( ℃ ). Step 1 : PRT methodStep 2 : RxT method

4. Radiometric correction : PRT+RxT 23GHz Vpol T b_ HTS & Treciever From PRT method

5. Radiometric correction : Results Comparison between computed Tb based on OI-SST and AMSR- E Tb by (a) simple two-point calibration and (b) presented method for 6.925-GHz vertical polarization on July 4, 2002. (c) Daily average of difference between computed and AMSR-E Tb as a function of position in orbit for simple two-point calibration (cross) and presented method (closed circle). AMSR-E Tb (K) Computed Tb from OI-SST (K) Position in orbit (*0.01 revolution) Difference (K) Computed Tb from OI-SST (K) (a) (b) (c)

6. TMI and SSM/I over rain forest (2004) (F14 V-H) VS (F14-TMI)(F15 V-H) VS (F15-TMI) 2003.01-2003.09 Land Match-up Less than 10 min. 2003.01-2003.09 Land Match-up Less than 10 min. SSM/I Local Time 20:21 (Asc) SSM/I Local Time 21:31 (Asc) 10G 19G 21G 37G 85G

7. TMI and AMSR/AMSR-E (Ver.1, 2004) (AMSR V-H) VS (AMSR-TMI)(AMSRE V-H) VS (AMSRE-TMI) 2003.04-2003.10 Land Match-up Less than 10 min. 2002.06-2003.12 Land Match-up Less than 10 min. 10G 18G 23G 36G 89G

8. Tb difference (L1B-L2A): 36.5GHz (2004) AscendingDescending July 1-2, 2003 L1B: V2

9. Tb difference (L1B-L2A): 23.8GHz (2004) AscendingDescending July 1-2, 2003 L1B: V2

10. Lunar emission in CSM counts CSM counts are influenced by lunar emission. Detected and removed by computing departure angle between CSM viewing direction and moon direction. Contamination from the Earth Contamination from the Moon AMSR-E CSM CAL Count (July 20, 2002, Path No. 195, Ascending)

11. Possible spillover into CSM counts CSM counts are influenced by Earth’s emission. Good correlation was found between variations of contamination and Earth’s Tb of about 100 scans before. Current processing system subtract this contamination by assuming CSM spillover. Spillover factor of approximately 0.4 % was statistically found and used for correction. Before After Calibrated map of 6GHz-H is made of 233 descending paths, in July 1-16, 2002.

12. 11 Comparison (2009, before TMI corr.) WindSat AMSR-E 10G18G23G36G89GLTANEIA Windsat10.6518.723.837.0-18:0055.3, 53.0 SSM/I-F13-19.3522.23537.085.518:3353.1 SSM/I-F14-19.3522.23537.085.519:0853.1 TMI10.6519.3521.337.085.5-52.88 AMSR-E10.6518.723.836.589.013:3055.0, 54.5

13. 12 Comparison (2009, before TMI corr.) 10GHz AMSR-E WindSat Frequency [GHz] Correction to TMI [K] TMI AD Avg. TMI-ASC TMI-DSC

14. Vrangelja Attu Rishiri Taiwan SaipanHawaii Mare Balleny Easter Galapagos Jamaica Longs Coasts Falkland South Georgia Gough Price Edward Heard Mauritius Socotra Svalbard Jan Mayen Cape of Good Hope ： The location of GCPs, 70GCPs are selected equally all over the world. Reference Areas (GCP)

15. Geo-location of AMSR-E 89GHz A-horn (Current version, based on one-year products in 2003) Geo-location of AMSR-E 89GHz A-scan (Using “Optimal parameters, based on one-year products in 2003) Geometric parameterCurrent valueOptimal value Roll angle offset [deg.]0.00+0.09 Pitch angle offset [deg.]0.00+0.04 Yaw angle offset [deg.]0.00+0.13 Off Nadir angle [deg.]47.5047.42 Detemined “Optimal Parameters”

16. AMSR-E Calibration Plans Calibration analysis and improvements Bias/Systematic errors: Improve cross calibration (X-CAL results are the good reference), non-linearity corrections, and understanding of the related issues (e.g. HTS and spillover). Orbital/Time-varying errors: Improve HTS correction methods. Relative errors: Derive scan bias errors for wider swath (234 points for low-freq). Geolocation errors: Correction almost completed. Everything needs to be consistent to new incidence angles. Others: Revisiting RFI issue to prepare for masking in AMSR2 data. Version upgrade Minimum (possibly unofficial) update (e.g. geolocation and bug fixing) is now under consideration in this JFY. Full update after completed cross calibration with AMSR2, which will be launched into A-Train constellation in JFY 2011.