摘要
贵州省思南县境内崩塌地质灾害较为发育,通过对县域崩塌易发性进行预测,可准确获取崩塌地质灾害分布规律,为国土部门开展崩塌防治提供科学指导。因此,首先采用遥感和地理信息系统等技术,对思南县的自然地理和地质条件等因素进行分析;再采用频率比分析和相关系数分析法,建立崩塌地质灾害与基础影响因子之间的非线性响应关系;最后,首次提出一种典型的机器学习:径向基神经网络模型,对思南县崩塌易发性进行预测并绘图。结果表明:径向基神经网络预测思南县的崩塌易发性的准确率(AUC曲线)达到0.945,非常准确地预测出了思南县崩塌地灾的分布规律。且崩塌易发性分布图显示极高、高、中等、低和极低易发区面积占县域总面积的比值分别为13.06%、14.08%、25.41%、23.68%和23.77%。
The collapse geological disasters are widely distributed in the Sinan County of Guizhou Province,China.As a result,the local economic development and the safety of people and properties in Sinan County are seriously threatened by collapse disasters.It is very necessary to explore the development laws of collapse disasters in depth.In recent years,the collapse susceptibility assessment(CSA)has been developed as an effective tool for collapses prediction and prevention,the development laws and distribution features of collapses in Sinan County can be revealed through CSA.Therefore,the remote sensing(RS)and geographic information system(GIS)technologies are adopted in this study to analysis the physic-geographical environment and geological conditions of Sinan County;And then the frequency ratio and correlation analysis methods are used to build the nonlinear dynamic corresponding relationships between the basic condition factors and collapse disasters.Finally,a typical machine learning model namly radial basis function neural network(RBFNN)is adopted to predict and map the collapse susceptibility based on GIS.Results show that the area under the receiver operating characteristic curve(ROC)of RBFNN is 0.945,suggesting that the distribution rules of collapse disasters in Sinan County are predicted very accurately using the RBFNN model;meanwhile,the proportions of very high,high,moderate,low and very low susceptible areas to the whole area of Sinan County are 13.06%,14.08%,25.41%,23.68%and 23.77%,respectively.
作者
胡涛
樊鑫
王硕
冷信风
王建伟
HU Tao;FAN Xin;WANG Shuo;LENG Xin-feng;WANG Jian-wei(Faculty of Engineering, China University of Geosciences, Wuhan 430074,China;Geological Engineering Exploration Institute of Guizhou Province, Guiyang 550002, China;Land and Resources Bureau of Tongren City, Tongren 554300, China;Land and Resources Bureau of Xiushui County, Xiushui 332400, China;Science and Technology Research Center of Safety Production of Jiangxi Province, Nanchang 330030, China)
出处
《科学技术与工程》
北大核心
2019年第35期61-69,共9页
Science Technology and Engineering
基金
国家自然科学基金(41572292)资助
关键词
崩塌
易发性评价
径向基神经网络
遥感
地理信息系统
collapse disaster
susceptibility assessment
radial basis function neural network
remote sensing
geographic information system
