Abstract: 【Objective】To develop a high-precision geological hazard assessment model for township-scale applications in the Qinba Mountain area, reveal the spatial distribution characteristics of geological hazard under different rainfall scenarios in Guanyuan Town, and provide a scientific basis for regional geological hazard prevention and mitigation.【Methods】Guanyuan Town, Langao County, was selected as the study area. Eight conditioning factors, including elevation, slope, aspect, curvature, lithology, normalized difference vegetation index , distance to roads, and distance to rivers, were selected to establish an Information Value–Random Forest (IVM-RF) coupled model for geological hazard susceptibility assessment. The SHapley Additive exPlanations method was employed to quantify the contribution of each conditioning factor to the model output. Based on historical rainfall records, four rainfall scenarios (moderate rain, heavy rain, rainstorm, and severe rainstorm) were designed to conduct geological hazard assessment under different rainfall conditions.【Results】The IVM-RF coupled model achieved higher assessment accuracy than the individual models. SHAP analysis indicated that NDVI, elevation, distance to roads, and slope were the dominant conditioning factors controlling the model output. With increasing rainfall intensity, the areas classified as very high and high hazard expanded continuously, and were mainly distributed in low mountainous areas and along major roads within the study area.【Conclusion】The IVM-RF coupled model effectively improves the accuracy of township-scale geological hazard assessment in the Qinba Mountain area. Hazard assessment under different rainfall scenarios successfully reveals the influence of rainfall intensity on the spatial distribution of geological hazards. The proposed approach provides technical support for geological hazard monitoring, early warning, and disaster prevention and mitigation in geomorphologically similar mountainous regions.