Coal fire identification through satellite remote sensing considering the landscapes

Coal fire is a serious problem in China’s mining areas, which has become a major threat to regional environment and energy security. The timely and accurate grasp of the distribution range of coal fires and the clarification of their spatiotemporal evolution law are crucial for effective control and management in coal fire areas. In this paper, a coal fire identification method based on satellite remote sensing considering the difference of land cover is proposed. Firstly, based on Landsat satellite images, the atmospheric correction was carried out using the radiative conduction equation method to invert the absolute surface temperature. Secondly, the supervised classification method of support vector machine is used to obtain the land cover type and determine the abnormal region of coal fire. Finally, the distribution characteristics of coal fire in time and space are deduced by regression analysis method, and the coal fire prediction model is constructed. Wuhai City, Inner Mongolia Autonomous Region is selected as the research area, and the Landsat time series satellite images from 2018—2023 are used as the data source to carry out the coal fire anomaly identification and prediction experiment. The results show that the coal fire identification results of the proposed method are consistent with the field verification and low-level monitoring results, and the predicted region and coal fire degree are consistent with the satellite inversion results, which confirms the reliability of the proposed method for the “sweep target” identification of coal fires in large areas, and provides reliable technical support for the timely discovery of coal fires in mining areas, the analysis of space-time evolution laws, and the restoration and treatment of coal fires.