Abstract: In order to accurately extraction the Inner Mongolia Daliuta mine coalfield fire, based on night aerial thermal infrared hyperspectral remote sensing data(TASI), the method of the emissivity normalization method and ground synchronous temperature measurement data were adopted to establish space-to-ground regression equation for retrieving land surface temperature. The coal fire were extracted by the reasonable threshold.On this basis,the non-hot anomaly were separated from the area of coal heat anomaly, such as the cooling water of power plant,heat buildings,chimney,thorough the characteristic difference of the emissivity spectrum of the objects.The interfering factors were eliminated on the accurate extraction of the coal fires burning effect of anomaly area. The results show that the 1-16 bands of TASI are suitable for infrared temperature inversion for the coalfield fire.The inversion precision of the surface temperature meet the study on quantitative detection.The accurated extraction of coalfield fire thermal anomaly range were able to provide some decision support for underground coal fire extinguishing.