Evaluation of the algorithms for land surface temperature retrieval from MTSAT data

Multi-functional Transport Satellite (MTSAT) is a Japanese geostationary satellite launched in 2005 and provides hourly data with 5 bands including two thermal infrared bands in the 10.5-12.5μm region. Hourly LST data enables us to improve our understanding of surface energy balance. In this research, we have evaluated split-window equations to retrieve hourly Land Surface Temperature (LST) from thermal infrared data of MTSAT. Radiative transfer code was used to simulate brightness temperatures measured at sensor on MTSAT, and the coefficients of split-window equations were derived from the simulation results with a statistical regression method. As a result, though surface emissivity or atmospheric water vapor that are input parameters to estimate LST contain errors, the accuracy of the split-window equation which requires both parameters have been higher than that of the equation which require only surface emissivity. These split-window equations are able to estimate LST with the RMSE less than or equal to 0.7K and 1.09K, respectively, in the case of viewing zenith angle lower than 40 degree and no atmospheric water vapor and surface emissivity errors.