期刊
ENVIRONMENTAL EARTH SCIENCES
卷 72, 期 6, 页码 1925-1938出版社
SPRINGER
DOI: 10.1007/s12665-014-3100-5
关键词
Three Gorges Reservoir; Landslide susceptibility; Landslide causative factors; Artificial Neural Network; ArcGIS
资金
- German Federal Ministry of Education and Research (BMBF)
A landslide susceptibility map is very important and necessary to efficiently prevent and mitigate the losses brought by natural hazard for a large area. For the purpose of landslide susceptibility analysis for the whole Xiangxi catchment (3,209 km(2)), Artificial Neural Network (ANN) analysis was applied as the main method. The whole catchment was divided into two parts: the training area and the implementation area. The backwater area (559 km(2)) of Xiangxi catchment was used as the training area for the ANN method. In the training area the correlations between the landslide distribution and its causative factors, which includes lithology, slope angle, slope curvature and river network, have been analyzed based on the geological map and digital elevation model (DEM). The back-propagation training algorithm in ANN was selected to train the sample data from the training area, which were composed of input data (causative factors) and target output data (landslide occurrence), in order to find the correlations between them. Based on these correlations and input data in the implementation area (causative factors), the network output data were obtained for the implementation area. In the end, a map of landslide susceptibility, which was established by network output data, was presented for Xiangxi catchment. ArcGIS was applied to extract and quantify input information from a DEM for susceptibility analysis and also to present the result visually. As a result, a landslide susceptibility map, in which 70 % of all landslides are rightly classified in the training area (backwater area), was created for Xiangxi catchment.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据