Research on Regional Crustal Stability Evaluation Based on Random Forest Weighting: A Case Study of Pumped Storage Power Station in Tibet

A large pumped storage power station is planned in Mangkam, Xizang, adjacent to the Lancang River Fault Zone, where the terrain is highly dissected. In recent years, seismic activity, extreme weather, and human interference have collectively caused frequent collapses and landslides. Based on comprehensive data collection and field geological investigations, this study evaluates regional crustal stability using a Random Forest (RF) model, incorporating seismicity, geological hazards, geophysical fields, and in situ stress conditions. The results show that the area of stable, relatively stable, relatively unstable and unstable crustal zones account for 31.2%, 34.09%, 16.53% and 18.16%, respectively. Fault structures are the dominant factor controlling crustal instability, while geological hazard susceptibility exerts a significant secondary influence. The RF and Analytic Hierarchy Process (AHP) yield broadly consistent zoning patterns, but RF demonstrates clearer spatial differentiation and a more pronounced response to key controlling factors, indicating its greater effectiveness for crustal stability assessment.