The study of the land surface temperature retrieval with the proposed neural network model

Land surface temperature (LST) is a key parameter of surface physical process. It´s a effective way to measure the LST by remote sensing data. But, The radiance leaving the earth-atmosphere system which can be sensed by a satellite borne radiometer is the sum of radiation emission from the earth surface and each atmospheric level that are transmitted to the top of the atmosphere. It can be separated from the radiance at the top of the atmospheric level measured by radiometer. However, it is very difficult to measure the atmospheric radiance , especially the synchronous measurement with the satellite. In this paper, based on the knowledge of atmospheric radiative transfer, the moderate spectral resolution atmospheric transmittance algorithm and computer model is selected to model the atmosphere. The retrieval of atmospheric elements , and the surface parameters , will also be presented. At the same time, a two-layer BP neural net model is constructed with three input nodes, and three output nodes. To successful retrieve LST of the field data of Qinghai Lake in the Qinghai province.