On the characteristics of the coefficients in the sea surface temperature estimation functions due to the split-window method

Abstract

The split-window method is known as an effective sea surface temperature estimation algorithm by using NOAA/AVHRR data. Two fundamental structures are presently used for the purpose, i.e. the double varible function and the split-window function, but their coefficients derived from the regression analysis are empirically known to be almost equal. The similarity of the coefficients was investigated mathematically. It is shown that the similarity comes from the data statistics specific to the split-window method, i.e., the comparable ranges of the sea-truth temperature and the satellite brightness temperatures, and the high correlation coefficients between them. The variations of the coefficients were observed by the simulation using the match-up data set collected in Mutsu bay.