Integrating information value method and extreme gradient boosting for debris flow susceptibility assessment on the piedmont clinoplain considering watershed erosion

Assessing debris flow susceptibility is crucial for predicting their occurrence and associated hazards. However, traditional methods rely on relatively static environmental conditions, which exposes the lag in the temporal dynamic changes of debris flow susceptibility assessment. This study focuses on the transition from low to mid mountain terrain located at the southern piedmont of the Yinshan Mountains to a plain within the piedmont clinoplain of the Inner Mongolia Plateau. By integrating the Information Value Method (IVM) and Extreme Gradient Boosting (XGBoost) and utilizing high spatiotemporal surface change data derived from interferometric synthetic aperture radar (InSAR), a new evaluation factor, namely the "basin erosion degree, " is introduced to enhance debris flow susceptibility assessment. The contribution analysis of evaluation results and evaluation factors based on two models shows that IVM and XGBoost identified 90.48% and 95.23% of debris flow gullies in the study area, respectively, and both performed well. XGBoost also performed well in identifying non-debris flow gullies, demonstrating its extremely high accuracy and practicality in assessing debris flow activity in small data scenarios, while IVM misjudged a large number of non-debris flow gullies due to not considering the weight of evaluation factors. Moreover, the newly introduced watershed erosion evaluation factor ranks fourth in the contribution to debris flow, which is of great significance for activity evaluation. The dynamic characteristics of typical debris flow gullies in the study area were simulated using Flo-2D numerical simulation software. The results indicate that mudslides in the area can invade roads, residential buildings, and other facilities, posing a threat to production and daily life. This study accurately assessed the susceptibility of debris flows in the study area by introducing dynamic evaluation factors, making progress in debris flow susceptibility modeling, and promoting the construction of active disaster management systems.