Landslide hazard assessment in Dadu River Basin based on GIS and BP neural network

Landslide is one of the common natural disasters. Its strong suddenness and high destructiveness bring serious threat to people's life and property safety. Landslide hazard risk assessment can help us effectively prevent and deal with landslide disasters. In this paper, rainfall, DEM, slope, curvature, vegetation coverage, distance from river and distance from fault are taken as input variables, and the recorded landslide disaster points and their surrounding areas are defined as output data. BP neural network algorithm and different activation functions are used to classify and regression forecast the landslide disaster risk in Dadu River basin. The assessment results show that the results of classification and regression calculation of landslide affected areas by BP neural network in Dadu River Basin are not different, about 0.78% of the area may be affected by Grade 1 landslide, about 0.18% of the area may be affected by grade 2 landslide, and about 0.72% of the area may be affected by grade 3 landslide.