A mutiple-band algorithm for retrieving land surface temperature and emissivity from EOS/MODIS data

Abstract

We intend to propose a multiple-band algorithm which can simultaneously retrieve land surface temperature and emissivity from MODIS data. We build four radiance transfer equations for MODIS band 29, 31, 32, 33, which involve six unknown parameters (average atmosphere temperature, land surface temperature and four bands emissivity). We also analysis the emissivity characteristics of common objects about 80 kinds under 1KM scale of MODIS data provided by ASTER spectral database between thermal band 29, 31,32, 33 and find that there is approximate linear relationship between them. In order to describe them more accurately, the land surface at least are classified into two groups. The first is none- water and snow covered surface, the second is water and snow covered surface. For the two types, the average emissivities error of band 31and 32 are all under 0.00504. The max emissivity error is under 0.0097 for band 31 and 32. So we can obtain six equations and six unknown parameters. On the other hand, we can also just utilize MODIS band29, 31, 32 and another equation that the average atmosphere temperature is the function of land surface temperature to retrieve land surface temperature and emissivity. We obtain the transmittance through building the relationship between water content and transmittance by simulation of MODTRAN in different conditions because the transmittance of TIR is mainly determined by the influence of water content in atmosphere and we can retrieve water content of atmosphere from MODIS data. © 2005 IEEE.