Land surface temperature(LST) is a key parameter in earth environment, the thermal infrared band that can detect LST plays an important role in spectroscopy. Aiming to the latest optical and thermal bands of HJ-1B satellite, the LST retrieval over Ningxia plain was implemented using a mono-window algorithm without atmospheric water vapor content input, based on the COST model for atmospheric correction. Considering the difficulty of obtaining simultaneous ground measured data, the MODIS LST product was adopted as a standard to test the approach. The comparison and validation indicate that this method has good reliability with accuracy of less than 1 K. In addition, the sensitivity analysis is performed for land surface emissivity, and the result shows that this variable was not sensitive to LST, because the LST error is less than 0. 5 K when it varies at medium level. This study proves that the satellite data has higher availability for detecting LST.