Disaster risk assessment of collapses and landslides in a hilly coastal city: the role of rainfall triggers and the disaster-inducing environment

In this study, we developed a dual-driven framework of ‘environmental and precipitation triggers’ using the maximum entropy (MaxEnt) model, optimized by the Kuenm package in R, to assess susceptibility to collapses and landslides in the Jinpu New Area, Dalian, China. Factors representing the disaster-inducing environment include elevation, slope, aspect, topographic position index (TPI), surface roughness, topographic wetness index (TWI), lithology, distance from faults, distance from rivers, river power index (RPI), land use types, normalized difference vegetation index (NDVI), and distance from roads. Following 10 repetitions of cross-validation, the model achieved an area under the curve (AUC) value of 0.83, demonstrating excellent predictive performance. The study classified susceptibility to collapses and landslides into five categories: very low, low, medium, high, and very high and assessed potential disasters under extreme precipitation scenarios corresponding to 5-, 10-, 20-, and 40-year return periods. The results revealed a marked spatial expansion of disaster areas as extreme rainfall increased, with the very high, high, medium, and low risk areas growing by 9.9%, 12.5%, 16.9%, and 9.8%, respectively, while the very low risk areas decreased by 49%. In addition, the single-factor detection of the Geodetector identified slope as the primary driving factor of collapse and landslide disasters. The interaction detection revealed that the nonlinear interaction between TWI and surface roughness was the strongest. These findings provide a scientific foundation for regional geological disaster prevention and management.