Land surface temperature is higher in the zones of underground coal fire than in their surroundings areas. It is possible to monitor the coal fire and the heat anomalies using the remote sensing technology of thermal infrared. By taking the coal fire of Wuda in Inner Mongolia Autonomous Region as an example，the monitoring methods of underground coal fire were explored based on different seasons night ASTER thermal infrared images. By employing the TES-ADE algorithm and threshold segmentation method，land surface temperature was retrieved and coal fire areas were extracted. Finally the seasonal variation of surface temperature，the surface temperature differences between night and day，and the change and distribution of coal fire area were analyzed. The results showed that (1) it is effective to retrieve land surface temperature and to avoid the interference of sand, bare land and building based on the remote sensing of thermal infrared at night; (2) it is best time to extract the coal fire areas based on the winter night images of thermal infrared; (3) it is effective to monitor the change and distribution of coal fire areas based on the winter night images of thermal infrared.