Regional dynamic early warning model for rainfall-induced landslide in Fujian, China

Abstract

Recent advances have significantly enhanced dynamic early warning systems for rainfall-induced landslides by integrating rainfall thresholds with susceptibility mapping. However, there is still room for optimizing these models for regional-scale applications. This study introduces a dynamic early warning model in Anxi County, Fujian Province, China—a region prone to rainfall-induced landslides. First, historical landslide data and their controlling factors were analyzed for the study area, and an optimal landslide susceptibility map was produced by integrating an information value model with a logistic regression model. Second, based on the theory of effective rainfall, the regional rainfall threshold model was established according to the indices of daily and hourly rainfall, and the model’s accuracy was evaluated. Third, the more effective E-D (daily index) and EE-D (hourly index) models were coupled with the landslide susceptibility map for landslide dynamic early warning. The model’s validation results show its significant predictive capabilities, with the hourly model proving more accurate for short-duration rainfall events. This study provides valuable insights for local authorities on dynamic early warning for rainfall-induced landslides and offers guidance for refining dynamic early warning systems in similar regions.

Keywords

• Landslide susceptibility
• rainfall-induced landslide
• rainfall threshold
• dynamic early warning model