A Refinement and Cross-Validation of the Special Sensor Microwave Imager (SSM/I) Calibration-Validation (CV) Brightness Temperature Algorithm
AIR FORCE INST OF TECH WRIGHT-PATTERSONAFB OH SCHOOL OF ENGINEERING
Pagination or Media Count:
The Air Force Weather Agency AFWA currently uses an algorithm to calculate surface temperatures from microwave observations taken by the Special Sensor Microwave Imager SSMI aboard the orbiting platforms of the Defense Meteorological Satellite Program DMSP. This algorithm, called the Calibration-Validation CV algorithm, used multiple linear regression to calculate coefficients relating microwave brightness temperatures and land surface temperatures. Because the coefficients in this algorithm do not take into account the identity of the individual satellite, the question arose whether this assumption was valid. This thesis used multiple linear regression, stepwise linear regression, and qualitative regression on 3700 data sets from October of 1996 and September of 1997, including microwave brightness temperatures from three satellites. This data was analyzed to determine if satellite identity had a significant impact on CV regression coefficients. Analysis indicated that satellite identity does not have a significant impact on regression coefficients for five of the eight CV land types investigated. Analysis of two CV land types indicated data set identity had a significant impact, while there was insufficient data to determine the impact for one CV land type. In addition to the qualitative regression, stepwise linear regression was performed on five land type categories using combined data for all satellites. Regressed RMSEs ranged from 2.825 K to 3.743 K, while R squared values ranged from .7295 to .8613. Preliminary analysis indicated refinement of CV brightness temperature coefficients might yield better accuracy for the algorithm.