期刊
NATURAL HAZARDS AND EARTH SYSTEM SCIENCES
卷 9, 期 3, 页码 673-686出版社
COPERNICUS GESELLSCHAFT MBH
DOI: 10.5194/nhess-9-673-2009
关键词
-
资金
- NASA [NNG05GN31H]
- Lamont-Doherty Earth Observatory, Columbia University
Most landslide hazard assessment algorithms in common use are applied to small regions, where high-resolution, in situ, observables are available. A preliminary global landslide hazard algorithm has been developed to estimate areas of potential landslide occurrence in near real-time by combining a calculation of landslide susceptibility with satellite derived rainfall estimates to forecast areas with increased potential for landslide conditions. This paper presents a stochastic methodology to compare this new, landslide hazard algorithm for rainfall-triggered landslides with a newly available inventory of global landslide events, in order to determine the predictive skill and limitations of such a global estimation technique. Additionally, we test the sensitivity of the global algorithm to its input observables, including precipitation, topography, land cover and soil variables. Our analysis indicates that the current algorithm is limited by issues related to both the surface-based susceptibility map and the temporal resolution of rainfall information, but shows skill in determining general geographic and seasonal distributions of landslides. We find that the global susceptibility model has inadequate performance in certain locations, due to improper weighting of surface observables in the susceptibility map. This suggests that the relative contributions of topographic slope and soil conditions to landslide susceptibility must be considered regionally. The current, initial forecast system, although showing some overall skill, must be improved considerably if it is to be used for hazard warning or detailed studies. Surface and remote sensing observations at higher spatial resolution, together with improved landslide event catalogues, are required if global landslide hazard forecasts are to become an operational reality.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据