Through the comparison between temperatures estimated from remotely sensed data and those actually measured, we discuss causes of discrepancy between them. We apply regression analysis to the data and pay particular attention to regression coefficient which as a shole represents causes for the error. The coefficient obtained by taking ground truth data as independent variables and estimated temperatures as dependent variables tends to be less than 1. Atmospheric effect on the coefficient is studied, being based on a simple model. Vertical temperature profile, another possible cause for the tendency, is also discussed on the basis of laboratory experiments.
Date of this Version