基于三种机器学习算法的山洪灾害风险评价

被引：25

作者：

周超

方秀琴

吴小君

王雨晨

机构：

[1] 河海大学地球科学与工程学院

来源：

地球信息科学学报 | 2019年 / 21卷 / 11期

基金：

国家重点研发计划;

关键词：

随机森林机器学习算法; AdaBoost机器学习算法; ROC曲线; Boruta算法; 洪灾风险评价; 江西省;

D O I：

暂无

中图分类号：

TV87 [防洪工程]; P426.616 [降水引起的灾害];

学科分类号：

摘要：

依据洪灾风险概念模型,从触发因子、孕灾环境和承灾体3方面选取江西省的12个洪灾风险指标,采用k近邻、随机森林、AdaBoost 3种机器学习算法构建洪灾风险评价模型。利用精度、Kappa系数、ROC曲线(AUC值)3种定量评估指标评价洪灾风险模型,基于随机森林和Boruta特征提取算法共同分析指标重要性,最后对比3种模型绘制的江西省山洪灾害风险分区图并分析山洪灾害分布特征。结果表明:①AdaBoost模型的精度、Kappa系数和AUC值的平均值为别为0.902、0.870和0.826,精度和Kappa系数略优于随机森林,AUC值与随机森林相当,而k近邻模型的3种性能指标均低于前2种算法;②农田生产潜力、年最大6 h暴雨均值、年最大1 h暴雨均值、归一化差值植被指数、年降雨量均值这5个指标对最终的洪灾风险形成具有非常重要作用;③江西省较高风险区与最高风险区的面积和约占江西省总面积的34.4%,且主要分布于高降雨量、高暴雨量、农田生产潜力大的山区。

引用

页码：1679 / 1688

页数：10

共 23 条

[11] Flood probability quantification for road infrastructure: Data-driven spatial-statistical approach and case study applications[J] . Zahra Kalantari,Marco Cavalli,Carolina Cantone,Stefano Crema,Georgia Destouni.Science of the Total Environment . 2016
[12] Applying the K-nearest neighbor technique to the classification of workers according to their risk of suffering musculoskeletal disorders[J] . A. Suárez Sánchez,F.J. Iglesias-Rodríguez,P. Riesgo Fernández,F.J. de Cos Juez.International Journal of Industrial Ergonomics . 2015
[13] Efficient k NN classification algorithm for big data[J] . Shichao Zhang,Zhenyun Deng,Debo Cheng,Ming Zong,Xiaoshu Zhu.Neurocomputing . 2015
[14] Flood susceptibility assessment using GIS-based support vector machine model with different kernel types[J] . Mahyat Shafapour Tehrany,Biswajeet Pradhan,Shattri Mansor,Noordin Ahmad.Catena . 2015
[15] Efficient multi-camera vehicle detection, tracking, and identification in a tunnel surveillance application[J] . Reyes Rios-Cabrera,Tinne Tuytelaars,Luc Van Gool.Computer Vision and Image Understanding . 2012 (6)
[16] An artificial neural network model for flood simulation using GIS: Johor River Basin, Malaysia[J] . Masoud Kia,Saied Pirasteh,Biswajeet Pradhan,Ahmad Mahmud,Wan Sulaiman,Abbas Moradi.Environmental Earth Sciences . 2012 (1)
[17] An improved K -nearest-neighbor algorithm for text categorization[J] . Shengyi Jiang,Guansong Pang,Meiling Wu,Limin Kuang.Expert Systems With Applications . 2011 (1)
[18] Flood-hazard assessment and risk-based zoning of a tropical flood plain: case study of the Yom River, Thailand[J] . Tawatchai Tingsanchali,Fazlul Karim.Hydrological Sciences Journal . 2010 (2)
[19] Risk assessment and validation of flood disaster based on fuzzy mathematics
Weiguo Jiang a
[J]. Progress in Natural Science, 2009, 19 (10) : 1419 - 1425
[20] Random forests
Breiman, L
[J]. MACHINE LEARNING, 2001, 45 (01) : 5 - 32

← 1 2 3 →