Spatial-temporal landslide hazard analysis considering ecological and hydrological factors

Ecological factors, such as vegetation roots, can enhance soil shear strength, reduce surface runoff erosion, and improve slope stability. Conversely, heavy rainfall can rapidly saturate unsaturated soil, weakening shear strength. As such, both ecological and hydrological processes play critical roles in shallow slope failures. In this study, we propose a spatial-temporal landslide hazard model considering ecological and hydrological factors. Specifically, ecological factors are incorporated in convolution neural network as a spatial probability module, and hydrological factors are involved in infinite slope hydrological model as a temporal probability module, finally we integrated the spatial-temporal probabilities using integration module. A case study in the Hengduan Mountain Region demonstrated that the proposed model outperforms traditional models by achieving higher prediction accuracy, with an increase in the landslide hit index, demonstrating the effectiveness of involving ecological and hydrological factors in landslide hazard analysis, offering practical suggestions for landslide risk management.