Temperature and emissivity separation from ASTER data based on the urban land cover classification

Separation of land surface temperature (LST) and emissivity from the remote sensing data is crucial to the urban energy consumption estimation. The difficulty of the TES method is that a single multispectral thermal measurement with N bands presents N equations with N+1 unknown parameters (N spectral emissivities and one land surface temperature).Accordingly, the simultaneous estimation is an underdetermined case. Different assumptions were made to solve the equations. In this work, a new procedure was proposed to make an emissivity assumption based on the classification result. The Normalized emissivity method (NEM), Ratio method (RAT) and Minimum Maximum Difference (MMD) were combined to perform the land surface temperature and emissivity separation. The results indicated that the land surface temperature is between 297 k and 312 k.The retrieved emissivity in each band is less than 1 and the emissivity spectral shape is in good agreement with the laboratory measurements of the emissivity about the vegetation and other typical surface features. Therefore, the retrieval land surface temperature and emissivity are reasonable. At last, the error of the TES method was analyzed .The atmospheric correction procedure and the assumption of the emissivity and temperature is the main source of the error.