Comparative analysis of landslide susceptibility evaluation models based on coefficient of determination method：A case study in the surrounding area of Baoshan basin, Yunnan province

As urbanization continues to progress in the BaoShan Basin, geological conditions have deteriorated and leading to frequent landslide natural hazards. To provide prevention and control recommendations, landslide susceptibility assessment was carried out in the study area. Nine evaluation factors, including elevation, slope, slope direction, normalized vegetation cover (NDVI), engineering geological rock group, distance from roads, distance from faults, distance from water systems, and density of disaster sites, were selected based on the regional geological background, human activities and disaster development characteristics. Four models were established: deterministic coefficient method (CF), the coupled model of deterministic coefficient method and hierarchical analysis method (CF-AHP), the coupled model of deterministic coefficient method and index of entropy method (CF-IOE), and the coupled model of deterministic coefficient method and distance function-combined weights method (CF-AHP-IOE). The results show that the hazard density and frequency ratios of CF, CF-AHP, CF-IOE, and CF-AHP-IOE models landslide susceptibility grading have significantly increased, effectively assessing the landslide susceptibility in the study area. The AUC values of the three models were 0.890, 0.911, 0.921, and 0.916, respectively, indicating higher evaluation accuracy for the coupled models. Among them, the CF-AHP-IOE model has a ratio of disaster frequency and disaster quantity exceeding 90% in the extremely high susceptibility zone, demonstrating a more reasonable weighting approach. The study can provide preemptive recommendations for the urbanization process in the BaoShan Basin.