%0 Journal Article
%T Landslide Susceptibility Assessment Based on Slope Unit and BP Neural Network
%A Miao Wang
%A Caiyan Wu
%A Xiao Liu
%A Qihao Wei
%J Open Access Library Journal
%V 10
%N 3
%P 1-20
%@ 2333-9721
%D 2023
%I Open Access Library
%R 10.4236/oalib.1109390
%X The result of landslide susceptibility assessment (LSA) can be used for landslide hazard management and emergency rescue. Lixian County in Sichuan Province is selected to be the study area for its frequent landslides. Based on the correlation analysis and geological environment interpretation, a total of 9 controlling factors of landslide were selected (i.e., lithology, elevation, slope curvature, slope angle, aspect, normalized difference vegetation index (NDVI), slope length, distance to river (DTR), distance to fault (DTF)). GIS-based back propagation neural network (BPNN) method is applied based on the slope unit in this paper. The landslide susceptibility maps of Lixian county were classified into five zones: very low, low, moderate, high and very high susceptibility classes. The results show that most of the historical landslides are located in the region with high landslide susceptibility. There are 3.3%, 6.6%, and 6.1% of the historical landslides distributed in very low, low, and moderate susceptibility classes of the study area, respectively. The remaining 29.8% and 54.2% of historical landslides are located in high and very high landslide susceptibility classes. Most landslides distribute along the Zagunao River and its influent in very high susceptibility area. The assessment result is validated by the ROC curve with the area under the curve (AUC) are 95.3%, which indicates the method of this research is good for LSA.
%K Slope Unit
%K Back Propagation Neural Network
%K Landslide Susceptibility Assessment
%K Lixian County
%U http://www.oalib.com/paper/6783218