科研之友
微信
新浪微博
Facebook
分享链接
摘要
Rainfall-triggered shallow slope failures are very common in the western Southern Alps of New Zealand, causing widespread damage to property and infrastructure, injury and loss of life. This study develops a geographic information system (GIS)-based approach for shallow landslide/debris-flow susceptibility assessment. Since landslides are complex and their prediction involves many uncertainties, fuzzy logic is used to deal with uncertainties inherent in spatial analysis and limited knowledge on the relationship between conditioning factors and slope instability. A landslide inventory was compiled using data from existing catalogues, satellite imagery and field observations. Ten parameters were initially identified as the most important conditioning factors for rainfall-generated slope failures in the study area, and fuzzy memberships were established between each parameter and landslide occurrence based on both the landslide inventory and user-defined functions. Three output landslide susceptibility maps were developed and evaluated in a test area using an independent population of landslides. The models demonstrated satisfactory performance with area under the curve (AUC) varying from 0.708 to 0.727. Sensitivity analyses showed that a six-parameter model using slope angle, lithology, slope aspect, proximity to faults, soil induration, and proximity to drainage network had the highest predictive performance (AUC = 0.734). The runout path and distance of potential future landslides from the susceptible areas were also modelled based on a multiple flow direction algorithm and the topographic slope of existing debris-flow deposits. The final susceptibility map has the potential to inform regional-scale land-use planning and to prioritize areas where hazard mitigation measures are required.
- 出版日期2015-12
