Optimization of vulnerability assessment through socio-economic data spatialization for enhanced landslide risk evaluation

In response to issues in regional landslide risk evaluation (LRE), such as the over-reliance on administrative unit statistics for vulnerability characterization and the difficulty in capturing the fine-scale spatial heterogeneity of socio-economic factors, this study takes the landslide-prone southern Anhui mountainous area, China as a case study to establish a comprehensive “susceptibility-hazard-vulnerability-risk” evaluation framework. First, landslide susceptibility assessment (LSA) was conducted by coupling frequency ratio (FR) with random forest (RF), support vector machine (SVM), and extreme gradient boosting (XGBoost). Through accuracy comparison, the FR-RF model was identified as the optimal one (AUC = 0.836). Subsequently, annual cumulative precipitation and rainfall erosivity factor were introduced, and the analytic hierarchy process (AHP) was applied to achieve a spatiotemporal landslide hazard assessment (LHA). Innovatively, based on land use data and multiple linear regression (MLR), population density and economic density by industry were spatialized at a 100 m × 100 m grid scale, and a spatial landslide vulnerability assessment (LVA) model was established. Finally, a regional landslide risk zoning map was generated through weighted overlay (hazard weight 0.6, vulnerability weight 0.4). The results show that high and very high hazard zones account for 38.74 % of the study area. Vulnerability exhibits significant spatial differentiation, with high and very high vulnerability zones covering only 3.83 %, primarily distributed in areas of intensive human activity. The comprehensive LRE reveals that high and very high risk zones occupy 34.67 % of the total area. Notably, 32.93 % of originally low-hazard areas were elevated to high risk due to the overlay of high vulnerability. By employing spatialization techniques for socio-economic data, this study achieves grid-level refined characterization of landslide vulnerability. It provides a methodological reference for regional LRE and offers practical implications for disaster risk management and territorial spatial planning in mountainous regions.