Atmospheric correction for land surface temperature using NOAA-11 AVHRR channels 4 and 5

Abstract

In this work, a theoretical model that permits relating the land surface temperature with the temperatures measured by thermal infrared sensors has been developed. The model has been derived by linearization of Planck's function and atmospheric trasmittance. In this way a split-window equation is obtained, which depends on atmospheric water vapor, viewing angle, and channel surface emissivities. Simulations of satellite measurements of land surface temperatures are made using the atmospheric transmittance-radiance model LOWTRAN-7 for NOAA-11 AVHRR Channels 4 and 5. From these simulations the accuracies of linearizations have been checked. The dependence of the split-window coefficients on the aforementioned magnitudes has been analyzed. Thus, it has been demonstrated that the angular dependence can be avoided while the total water vapor dependence must be retained. Therefore, we have obtained a split-window equation for each atmosphere in which the emissivity dependence is linearized in terms of Channel 4 emissivity ϵ 4 and channel emissivity difference, ϵ 4–ϵ 5. The analysis of the emissivity influence shows that the emissivity measurement should be made with an accuracy of 0.005 in order to get an error below to 0.4 K on land surface temperature. Finally, two applications of the split-window method for obtaining sea surface temperature and crop temperatures have been included.