Landslide hazard assessment in the middle reach area of the Dadu River based on the GDIV model

Regional geological hazard assessment is an important non-engineering approach for disaster reduction and prevention. Constructing a regional landslide hazard assessment model is of great significance in improving the accuracy of geological hazard evaluation and the efficiency of prevention. This study focuses on the frequent landslide occurrence in the middle reach area of the Dadu River and selects 13 primary factors, including elevation, slope, aspect, seismic parameters, soil type, engineering geological lithology, annual average rainfall, and topographic wetness index (TWI), to establish a primary evaluation index system for landslide hazard. Considering the varying contributions of each factor to landslide formation and the impact of the commonly used weighted raster superposition methods on assessment accuracy, the geographic detector and variable weight raster overlay techniques are introduced, leading to the development of the GDIV model. Using data from 313 landslide hazard points identified in the 2021 geological hazard risk survey at a scale of 1∶50,000 in Sichuan Province, the landslide hazard assessment in the middle reach area of the Dadu River basin is conducted based on the GDIV model, and the evaluation results are compared with those of the LRI model. The results show that the study area is predominantly characterized by middle and lower risk areas, accounting for 78.3% of the total area. The extremely high and high-risk areas are primarily located in the low-elevation regions along the banks of Dadu River, Geshizha River, and Donggu River. Compared to the LRI model, the evaluation results based on the GDIV model exhibit higher accuracy, with an area under the receiver operating characteristics (ROC) curve of 0.917. The GDIV model proposed in this paper improves the accuracy of regional Landslide hazards assessment, and serves as a valuable reference for similar geological disaster evaluations in other areas.