Assessment of landslide susceptibility in watersheds during extreme rainfall using a complex network of slope units

Rainfall-induced landslides present significant challenges in regional landslide prediction and management. Traditional regional landslide susceptibility assessment models often evaluate individual units in isolation, neglecting the hydrological connections between slope units within a watershed. This approach fails to account for the occurrence of landslides in groups. To address this limitation, we propose the "Network-based Landslide Susceptibility Assessment Model" (NLSAM). This model incorporates the impact of water transfer between slope units using a complex network and integrates a physically-based model to account for interactions between slopes. In this study, we applied the NLSAM to a watershed in Fuyang District, Zhejiang Province, China. Experimental results show that extreme rainfall increases water transfer between slope units, destabilizing more slopes and elevating landslide susceptibility. Validation results demonstrate that the recall of NLSAM is 0.93, confirming the model's ability to identify group-occurring landslides. NLSAM captures rainfall propagation paths and quantifies their impacts, assisting decision-makers in formulating more effective management strategies.