Coal Fire Zone Identification by Using an Enhanced Ensemble Learning Model and Multi-Modal Remote Sensing Data

The spontaneous combustion of coal seams presents a significant ecological and public health hazard, making it a global concern. In this study, we employed multi-modal remote sensing data to extract multi-dimensional anomalous characteristics of fire zones. We proposed an enhanced three-tier ensemble learning model (ETELM) for the identification of coal fire zones. This model identifies the Rujigou fire area in Ningxia, China through the fusion of homogeneous and heterogeneous learners, coupled with multiple abnormal elements. The results show that our model exhibits a substantial congruence with the actual fire area, and the precision of meta learners has markedly surpassed that of the base learners. These findings provide valuable insights for the monitoring and identification of fire zones.