New Scheme for Estimating Land Surface Temperature from AMSR-E Over the Continental United States

摘要

Land surface temperature (LST) is a key variable in the processes of energy and water balance between Earth's surface and atmosphere. To date, considerable researches have been focused on this issue, especially the thermal infrared (TIR) methods. Whereas TIR measurements are only limited to clear-sky condition, no observation is possible under cloudy conditions. While, passive microwave (PMW) as an alternative to the TIR measurements can penetrate the clouds. In this paper, the optical LST was derived from AMSR-E brightness temperatures by building a strong linear relationship over the continental United States. Unlike the previous studies, the algorithm was conducted by further using the corresponding ground measured LSTs which consist of both clear and cloudy conditions. The linear relationships were built on three sub-regions which is defined by NDVIs. The results show that the root-mean-square error (RMSE) of derived LST ranges from 2.50K to 3.23K for ascending overpass and 1.54K to 2.71K for descending track. This accuracy is proven to be better than existing work.