Landslide susceptibility mapping and model performance assessment | |
Xiao, Chenchao ; Tian, Yuan ; Si, Kangping ; Li, Ting | |
2013 | |
关键词 | Landslide susceptibility mapping Performance assessment Logistic regression GAM SVM GIS CONDITIONAL ANALYSIS METHOD LOGISTIC-REGRESSION DISTRIBUTIONS PREDICTION |
英文摘要 | In this paper landslide susceptibility mapping and model performance assessment was conducted using three models, logistic regression, GAM, and SVM, in a study area in Shenzhen, China. Ten factors, slope angle, aspect, elevation, plan and profile curvature of the slope, lithology, NDVI, building density, the distance to the river, and the distance to the fault were selected as influencing factors for the landslide occurrences. All three models were trained and the resulting susceptibility maps were created. The performances of the three models were then assessed by AUC values through a 10-fold cross-validation. It could be concluded that in the study area GAM had the best overall performance among the three models, while SVM was better than logistic regression. Based on the derived DPR values, the optimum thresholds between stable areas and risky areas for all three models were also determined.; Engineering, Civil; Engineering, Geological; EI; CPCI-S(ISTP); 0 |
语种 | 英语 |
DOI标识 | 10.4028/www.scientific.net/AMM.353-356.3487 |
内容类型 | 其他 |
源URL | [http://ir.pku.edu.cn/handle/20.500.11897/405725] |
专题 | 地球与空间科学学院 |
推荐引用方式 GB/T 7714 | Xiao, Chenchao,Tian, Yuan,Si, Kangping,et al. Landslide susceptibility mapping and model performance assessment. 2013-01-01. |
个性服务 |
查看访问统计 |
相关权益政策 |
暂无数据 |
收藏/分享 |
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。
修改评论