Abstract:The geological condition of the western mountainous area of Shijiazhuang city is poor, and the topography of this often forms heavy rainfall in the rainy season, so the geologic disasters are frequent, among which landslide disasters account for a relatively heavy proportion. Therefore, it is of great significance to analyze the susceptibility of landslide disasters in the western mountainous area of Shijiazhuang city. The existing single model has been widely used in landslide disaster susceptibility evaluation, but it still has some limitations. In order to improve the accuracy of the evaluation of regional landslide disaster susceptibility, multiple models such as single model and compound model were used to compare and analyze the landslide susceptibility in the study area, and an evaluation model with better evaluation accuracy was obtained. Firstly, a total of 12 evaluation factors including rainfall, slope, elevation, slope direction, curvature, geological lithology, distance from water system, distance from fault, seismic intensity, distance from road, vegetation coverage and land use type, were screened out by evaluating the evaluation factors related to forming disaster in the study area. Based on the informative method, the hierarchical analysis method, entropy weighting method and combined weighting method were used for comparative analysis, and the susceptibility zoning map was finally obtained, and its reliability was verified by ROC curve and density of disaster points. The results show that the AUC values of the above three analysis methods in the ROC curve are 0.732, 0.749, 0.766, respectively, and the evaluation results of the three models have a certain degree of reliability, among which the combined weighting method has higher accuracy and better reliability. Finally, the combined weighting model is determined to be adopted.