节点文献
基于多变量统计分析的大型滑坡敏感性评价:以汶川地震影响的陇南地区为例
Large landslide susceptibility assessment by multivariate statistical analysis in the Longnan area affected by the Wenchuan earthquake
【摘要】 以"5.12"汶川地震影响较严重的甘肃东部的陇南震区为研究区,分析地震前后影响历史大型深层滑坡的空间要素,以专家经验分析确定的成因要素作为滑坡敏感性评价的初选方案。为了避免专家经验的客观性和要素相互重叠对滑坡敏感性评价的影响,执行主成分(PCA)多变量分析法以确定新的评价方案。2个方案分别通过逻辑回归模型(LR)执行,并绘制滑坡危险等级地图。采用成功率曲线和累积面积曲线对2种评价结果的有效性和性能进行检验和对比。检验结果表明:采用新方案所得的滑坡敏感性地图具有更好的性能和更高的精度,因此,新方案下由LR绘制的滑坡危险等级地图可作为指导研究区震后重建、基础设施建设和土地规划的依据。
【Abstract】 Taking Longnan area of the east part of Gansu province,which was seriously affected by the "5.12" Wenchuan earthquake,as study area,a comprehensive analysis was conducted based on field investigation of the front and back of the Wenchuan earthquake.The caused factors were determined as the first scenario by expert experiences.To avoid the expert subjectivity and reduce the interaction between the caused factors,principal component analysis was conducted to determine a new scenario.The two scenarios were assessed by logical regression analysis.The landslide susceptibility maps produced by logical regression analysis were validated by the success rate curve and the cumulative area curve.The results show that the landslide susceptibility map in the new scenario is better than that of the original one.Therefore,the landslide susceptibility map with the new scenario can be recommended as the consultation of the post-quake reconstruction,infrastructure construction and land planning.
【Key words】 logistic regression; principal components analysis; large landslide; susceptibility assessment; Langnan earthquake area; Wenchuan earthquake;
- 【文献出处】 中南大学学报(自然科学版) ,Journal of Central South University(Science and Technology) , 编辑部邮箱 ,2012年09期
- 【分类号】P642.22
- 【被引频次】7
- 【下载频次】330