|
|||||
|
|
Robust landslide susceptibility analysis by combination of frequency ratio,heuristic GIS-methods and ground truth evaluation for a mountainous study area with poor data availability in the Three Gorges Reservoir area,PR China |
|
| Authors: | Markus Schleier Renneng Bi Joachim Rohn Dominik Ehret Wei Xiang |
| Institution: | 1. GeoZentrum Nordbayern, Department of Applied Geology, University Erlangen-Nuremberg, Schlossgarten 5, 91054, Erlangen, Germany 3. Faculty of Engineering, China University of Geosciences, Wuhan, Hubei, 430074, China 2. State Office for Geology, Resources and Mining of Baden-Wuerttemberg, Albertstrasse 5, 79104, Freiburg i. Br., Germany |
| Abstract: | Increasing number of geohazards, like mass movements, is one of the main environmental impacts following the impoundment of the Yangtze River and its tributaries due to the inventory of the Three Gorges Dam hydroelectric power plant. Although many cities and settlements are endangered, no detailed hazard mapping is possible because of the huge size of the affected area. Due to strongly limited data availability, a robust landslide susceptibility model was established exemplarily for the Xiangxi catchment as one of the main tributaries. The analyses were limited to translational, rotational and combined landslides in soft rock sediments because these represent the main types of mass movement activity in this area. The qualitative landslide susceptibility analysis was carried out by a combination of frequency ratio analyses and a heuristic iterative index-based method using a Geographical Information System. As conditioning factors, the parameters lithology, slope angle, -aspect, -curvature, drainage buffer distance and land use were applied. To improve the objectivity of the index-based method, the results of frequency ratio analyses were taken into consideration to assess the importance of each factor. Model verification and evaluation by ground truth enable to improve the model by iterative calculations and to identify the best performance model. Results indicate that 89 % of all known landslides are located within areas showing high susceptibility according to the best performance model. The study demonstrates that a rather simple but robust model achieves good results and is applicable for regional landslide susceptibility analyses in mountainous areas with poor data availability. |
| Keywords: | |
| 本文献已被 SpringerLink 等数据库收录！ | |