Landslide susceptibility assessment based on ConvNext in Longyang district of Baoshan city, China

Landslide is a common type of geological disaster in the world, which causes great harm to people's life and financial security. The evaluation of landslide susceptibility plays an important role in the prevention of landslide disaster, and has always been the focus of research in this field. In order to establish an accurate landslide susceptibility evaluation model, this paper takes Longyang District, Baoshan City, Yunnan Province, China as the research area, and comprehensively considers 10 evaluation factors such as elevation, slope, slope direction, lithology, water system, residential area, highway, terrain, Normalized Difference Vegetation Index(NDVI) and rainfall to establish a landslide susceptibility evaluation model based on ConvNeXt. More than 27000 sample data were used for model training, and the spatial distribution map of landslide susceptibility was obtained by using the trained model to predict the study area. Comparing this model with three typical landslide susceptibility evaluation models based on ResNet and Support Vector Machine(SVM), the performance indicators Area Under Curve(AUC) values of the three models were 0.97, 0.77, and 0.65, respectively, with accuracies of 0.92, 0.70, and 0.69, indicating that the landslide susceptibility evaluation model based on ConvNeXt proposed in this paper has significant advantages.The landslide susceptibility can be accurately evaluated, and the scientific basis for the prevention and control of landslide disasters can be provided.