Abstract: Aiming at the problem that the spatial and temporal resolution of land surface temperature (LST) have the contradiction with each other, a new downscaling model was put forward, based on the TsHARP(an algorithm for sharpening thermal imagery) downscaling method, this research makes improvements by selecting the better correlation of spectral index(normalized difference vegetation index, NDVI; normalized difference build-up index, NDBI; modified normalized difference water index, MNDWI; enhanced bare soil index, EBSI) with LST, i.e., replaces the original NDVI with new spectral index according to the different surface land-cover types, to assess the accuracy of each downscaling method based on qualitative and quantitative analysis with synchronous Landsat 8 TIRS LST data. The results show that both models could effectively enhance the spatial resolution while simultaneously preserving the characteristics and spatial distribution of the original 1 km MODIS LST image, and also eliminate the “mosaic” effect in the original 1 km image, both models were proved to be effective and applicable in our study area; global scale analysis shows that the new model (RMSE:1.635℃) is better than the TsHARP method (RMSE:2.736℃) in terms of the spatial variability and accuracy of the results; the different land-cover types of downscaling statistical analysis shows that the TsHARP method has poor downscaling results in the low vegetation coverage area, especially for the bare land and building-up area(|MBE|>3℃), the new model has obvious advantages in the description of the low vegetation coverage area. Seasonal analysis shows that the downscaling results of two models in summer and autumn are superior to those in spring and winter, the new model downscaling results are better than the TsHARP method in the four seasons, in which the spring and winter downscaling improvement is better than summer and autumn.

Key words: MODIS, downscaling, land surface temperature, TsHARP method, land-cover