A Simple Method for Land Surface Temperature Retrieval from AMSR-E

In this paper, we present a simple but efficient algorithm to derive land surface temperature (LST). The major point of our algorithm is to build linear regress equations between brightness temperature of AMSR-E and LST product from MODIS. Each equation corresponds to a group of land surface cells, within which the emissivities are similar. We found that for a given land surface type and a given latitude range, land surface emissivities are comparatively steady within a period (one month for example). In this way, we can separate all of the AMSR-E brightness temperature and MODIS LST product data into groups by land surface type, latitude range and date range. In each group, the regress equation can be build. We used 2004 data to build the regress equations and validated those used 2004 and 2006 data. We found that: 1, in snow covered area the accuracy of retrieved LST is poor. 2, in desert, accuracy of retrieval LST is also poor. 3, in other most land cover, retrieved LST by AMSR-E fit well with the MODIS LST product.