Abstract: Conventional methods for identifying coal-rich areas mostly rely on a single remote sensing parameter or empirical index combinations, suffering from inadequate information, subjective parameter selection, and limited generalization ability. Focusing on more than ten mining areas in Shanxi Province, including Zhengzhuang and Panzhuang, this study developed a design method for an integrated index based on multi-source heterogeneous remote sensing data. First, by labeling the longitude and latitude coordinates of mines and collection wells on Google Maps, six parameters were selected: surface subsidence, soil moisture, surface temperature, methane concentration, mineral alteration, and vegetation erosion. Then, a design of experiments (DOE) framework was introduced to systematically explore parameter combinations, and a hybrid-weight prediction model was developed using the entropy weight method and principal component analysis. Finally, the contributions of various parameter combinations were assessed based on matching accuracy of 200 m. The results indicate that the optimal core subset integrating the six parameters improved the identification accuracy to 92.7%~95.0%. The established index was verified in the Panzhuang, Heshun, and Xishan mining areas, achieving accuracies of greater than 91%. This result demonstrates that this index exhibits interpretability and strong generalization ability. The design method developed in this study provides a novel technical pathway for rapid exploration and environmental monitoring of coal mines.