以仙游县为例,探讨了将地理信息系统技术(GIS)和支持向量机(SVM)算法应用于滑坡灾害危险性评价的基本思路和技术路线。主要内容包括SVM的基本原理和方法、滑坡灾害危险性评价指标的选取和量化、SVM模型的建立以及具体的实现过程。实践证明该方法是一种较好的滑坡灾害危险性评价方法。
The development of Geographic Information System (GIS) technology provides a new technical method for the evaluation of landslide risk. Support Vector Machine is the hotspot of machine-learning industry research, and has been applied successfully in many areas. By taking Xianyou County as an example, a new method for landslide hazard evaluation based on GIS and Support Vector Machines (SVM) is presented in this paper. It includes the basic principles and methods of SVM, selection and quantification of landslide hazard evaluation index, foundation of SVM model and the way to realize it. According to the actual situation of the research area, the quantification method has been stipulated separately for each of six selected appraisal indexes including the elevation, the gradient, the slope, the gneiss, the rainfall and the vegetation. The system of landslide geology disaster risk appraisal has been established, the special chart of each appraisal index has been obtained by the use of the geographic information system spatial analysis function. From the results of the appraisal, the extremely high-risk danger and the high-risk danger areas are basically located in the central and northwestern parts of the study area; the secondary risk in both sides of them. This distribution result has basically reflected the present situation of geological disaster in the research area. This method can be put in practice in geology hazard investigation.
[1] 李天斌,陈明东.滑坡实时跟踪预报[M].成都:成都科技大学出版社,1999.
[2] 闫满存,王光谦,刘家宏. GIS支持的澜沧江下游区泥石流灾害危险性评价[J].地理科学, 2001, 21 (4) : 334~339.
[3] 唐 川,朱大奎. 基于GIS技术的泥石流风险评价研究[J].地理科学, 2002, 22 (3): 300~304.
[4] 闫满存,王光谦,刘家宏. 基于GIS的澜沧江下游区滑坡灾害危险性分析[J]. 地理科学, 2007, 27 (3): 365~369.
[5] 阮沈勇,黄润秋.基于GIS的信息量法模型在地质灾害危险性区划中的应用.成都理工学院学报, 2001, 28 (1): 89~92.
[6] 王文俊,向喜琼,黄润秋,等.区域崩塌滑坡的易发性评价[J]. 中国地质灾害与防治学报, 2003, 14 (2): 31~38.
[7] 计会凤, 蒋 涛 ,张 军.基于GIS 的滑坡地质灾害损失评估建库方法研究[J]. 遥感信息, 2005, 1: 50~55.
[8] 张像源 ,周 萌.基于专家评分模型和GIS的滑坡预警分析开发研究[J]. 中国地质灾害与防治学报, 2006, 17 (2): 107~110.
[9] Vapnik V N. The Nature of Statistical Learning theory[M]. New York: Springer Verlag,1995.
[10] Vapnik V N. Statistical Learning theory[M]. New York: Wiley,1998.
[11] 马 勇,丁晓青.基于层次型支持向量机的人脸检测[J].清华大学学报(自然科学版),2003,43(1):35~38
[12] 李 侃,高春晓,刘玉树.基于SVM的空间数据库的层次聚类分析[J].北京理工大学学报, 2002, 22(4): 485~488
[13] 向喜琼,黄润秋.基于GIS的人工神经网络模型在地质灾害危险性区划中的应用[J].中国地质灾害与防治学报,2000,11(3):23~27.
[14] 陈宁生,张 飞.2003年中国西南山区典型灾害性暴雨泥石流运动堆积特征[J].地理科学,2006,26(6):701~705.
[15] 原立峰.基于SVM的泥石流危险度评价研究[J].地理科学,2007,27(2):296~300.
[16] Vapnik V N. (张学工译).统计学习理论的本质[M].北京:清华大学出版社,2000.
