Abstract: Xishan in Beijing, a historically significant coal mining area, exhibits the extensive distribution of coal gangue resulting from prolonged mining activities. The coal gangue, with a loose structure, experienced significant erosion and deposition during the "23·7" extremely heavy rainfall in 2023, triggering numerous debris flows that threatened the lives and property of residents downstream. To assess the erosion-deposition probability of coal gangue within debris flow gullies, this study investigated the Xishan coal mining area. With catchments as assessment units, 11 assessment factors were selected: the area, gradient, and slope of a catchment; topographic relief; topographic wetness index (TWI); normalized difference vegetation index (NDVI); treatment engineering; and the area, location, slope, and interferometric synthetic aperture radar (InSAR)-derived deformation of coal gangue. Accordingly, the information value - logistic regression (Ⅳ－LR) and information value - random forest (Ⅳ－RF) coupled models were constructed. The assessment results of the models were validated based on receiver operating characteristic (ROC) curves. The Ⅳ－LR and Ⅳ－RF models yielded areas under the curve (AUCs) of 0.84 and 0.92, respectively, demonstrating that the Ⅳ－RF model is more suitable for the assessment of the erosion-deposition probability of coal gangue within debris flow gullies in the study area. Moreover, the Ⅳ－RF model achieved more reliable results with increasing coal gangue scale. Overall, the results of this study provide foundational data for the prevention and medium- to long-term prediction of debris flow risks in the coal mining area and its adjacent areas. This study offers technical support for the identification and mitigation of debris flow risks in Beijing, thereby assisting in guaranteeing the geological safety of the capital city.