Landslide susceptibility assessment using weights-of-evidence model and cluster analysis along the highways in the Hubei section of the Three Gorges Reservoir Area. (November 2021)

• Record Type: Journal Article
• Authors: Chen, Linfei; Guo, Haixiang; Gong, Peisong; Yang, Yuying; Zuo, Zhili; Gu, Mingyun
• Is Part Of: Computers & geosciences. Volume 156(2021)
• Publication Date: 2021-11
• Subjects: Highway -- Landslide susceptibility -- GIS -- Weights-of-evidence model -- Cluster analysis -- Three gorges reservoir area; Environmental policy -- Periodicals; 550.5
• DOI: 10.1016/j.cageo.2021.104899
• Languages: English
• ISSNs: 0098-3004

Abstract

Abstract: Landslide susceptibility assessment has become the focus of geological disaster research to strengthen disaster prevention and mitigation. Landslide disasters frequently occur in the Hubei section of the Three Gorges Reservoir Area (TGRA), with some potential landslides located along the highway, which brings risks to highway engineering, maintenance and transportation. In this paper, a comprehensive landslide susceptibility evaluation indicator framework with three dimensions and 12 factors was established, and an integrated approach was applied to evaluate the landslide susceptibility level, which combined weights-of-evidence model, seven clustering algorithms, three quality evaluation indices and the elbow method. To validate the effectiveness of the methods, five objective measures were employed for evaluation. The 69 samples along the highway were used for training, and another 30 samples were collected for validation. The results showed that the landslide susceptibility level of potential landslides can be effectively predicted by K-means algorithm. It was found that the landslide susceptibility for each cluster had significant differences, which were mainly reflected in natural induced factors, followed by human induced factors, while the slope structure showed little differences; the areas with low landslide susceptibility appeared sheet distribution, while the areas with high landslide susceptibility showed zonal distribution along the Yangtze River and its tributaries. This study developed a comprehensive indicator system and method for landslide susceptibility assessment along highways and provided a reference for the risk evaluation and prevention management. Highlights: The landslide susceptibility evaluation indicators along the highways are provided. The methods based on GIS and machine learning are proposed. The landslide susceptibility in the study area has obvious regional characteristics. Specific landslide prevention suggestions are put forward.