作者: Jie Liu , Zhao Duan
DOI: 10.3390/E20110868
关键词:
摘要: In this study, a comparative analysis of the statistical index (SI), entropy (IOE) and weights evidence (WOE) models was introduced to landslide susceptibility mapping, performance three validated systematically compared. As one most landslide-prone areas in Shaanxi Province, China, Shangnan County selected as study area. Firstly, series reports, remote sensing images geological maps were collected, field surveys carried out prepare inventory map. A total 348 landslides identified area, they reclassified training dataset (70% = 244 landslides) testing (30% 104 by random selection. Thirteen conditioning factors then employed. Corresponding thematic data layers generated based on ArcGIS software. Finally, area under curve (AUC) values calculated for order validate compare models. For dataset, AUC plots showed that WOE model had highest accuracy rate 76.05%, followed SI (74.67%) IOE (71.12%). case prediction rates SI, 73.75%, 63.89%, 75.10%, respectively. It can be concluded best capacity mapping County. The map produced profound engineering significance terms hazard prevention control other similar areas.