Search

This study calibrated a refined split-window algorithm for land surface temperature (LST) retrieval based on Fengyun-2D (FY-2D) meteorological satellite. First, FY-2D land surface emissivity (LSE) was predicted from Moderate-resolution Imaging Spectroradiometer (MODIS) LSE based on sensors spectral similarities. The retrieved FY-2D LST data were validated in an arid region where the traditional split-window algorithm generally performed unsatisfactorily. Validation results show R2 (coefficient of determination) and RMSE (root mean square error) values range 0.53–0.67 and 2.86–6.21 K, respectively, against ground observed LST. Better LST retrievals were observed over vegetated regions with an RMSE value of ~2.8 K. Spatially, the FY-2D LST was highly correlated (R2 = 0.83) with and showed marginal differences (±2 K) from MODIS LST for ~40% of the whole area.