科研之友
微信
新浪微博
Facebook
分享链接
摘要
The main objective of this study was to assess spatial prediction of slopes movement susceptibility in the Bousselam upstream basin, northeast of Algeria, using a linear indexing model and Geographic Information Systems. First, the locations of 1109 slope instabilities, which occurred in the last three decades, were mapped upon data from various sources such as follows: remote sensing, aerial photographs interpretation, and internal reports compilation. This slope movement inventory was randomly segmented into training and validation datasets (75% of the known events locations were used for training and building the model and the remaining 25% for its validation). Second, nine natural and anthropogenic causing factors were mapped as independent variables: geological factors (lithology and faults density), morphometric factors (slope, aspect, and elevations), environmental factors (precipitations, seism, and stream network density), and the land use factor (roads and rail network density). Third, the relative value of each categorical variable involved in the slope movements emergence was assessed (categorization of evaluation criteria, standardization of factors, and weighting of variables). Then, a global index value of slopes movement susceptibility was calculated for each cell in the study area by using a linear indexing model. Finally, the slopes movement susceptibility map was categorized into five hierarchic classes and validated using the validation dataset that was not used in the model building. The area under the curve was included to assess prediction capability of the adopted model (sensitivity = 0.83 and 1-specificity = 0.74). The resulted susceptibility map may be used for preliminary land planning purposes.
- 出版日期2016-3
