基于集成学习的山区中小流域滑坡易发区早期识别优化试验  

Optimization Experiment of Early Identification of Landslides Susceptibility Areas in Medium and Small Mountainous Catchment Based on Ensemble Learning

作　　者：刘海知 徐辉 包红军[1,2] 鲁恒 宋巧云[1,2] 狄靖月[1,2] 王蒙 曹爽 LIU Haizhi;XU Hui;BAO Hongjun;LU Heng;SONG Qiaoyun;DI Jingyue;WANG Meng;CAO Shuang(National Meteorological Center,Beijing 100081,China;CMA-HHU Joint Lab.for HydroMeteorological Studies,Beijing 100081,China;College of Water Resource&Hydropower,Sichuan Univ.,Chengdu 610065,China;State Key Lab.of Hydraulics and Mountain River Eng.,Sichuan Univ.,Chengdu 610065,China)

机构地区：[1]国家气象中心,北京100081 [2]中国气象局–河海大学水文气象研究联合实验室,北京100081 [3]四川大学水利水电学院,四川成都610065 [4]四川大学水力学与山区河流开发保护国家重点实验室,四川成都610065

出　　处：《工程科学与技术》2022年第6期12-20,共9页Advanced Engineering Sciences

基　　金：国家重点研发计划项目(2019YFC1510702);国家气象中心预报员专项课题(Y202105);中国气象局创新发展专项(CXFZ2022J019)。

摘　　要：滑坡作为山洪水沙耦合运动的物源和动力基础,其易发区的识别是山洪水沙灾害预报预警和风险评估的重要前提。以往的山洪水沙灾害防治研究主要关注洪水的影响,而忽视了固体物源的作用。为完善山区中小流域山洪水沙灾害防控体系,提出基于集成学习的山区中小流域滑坡易发区早期识别方法,并对数据样本构建和影响因子选取过程进行优化试验。将滑坡单元下垫面环境因子频率比作为无监督学习算法数据样本进行聚类分析;根据聚类算法易发性分区结果选取非滑坡单元,并结合滑坡单元构建集成学习分类算法数据样本集,比较单体算法和融合算法的易发性分区结果准确率和覆盖度。选取研究区域高分卫星遥感影像建立松散堆积物直接解译标志,基于目视解译识别松散堆积物面积,通过回归分析构建松散堆积物面积–体积幂律关系,形成研究区域松散堆积物空间分布图。将固体物源作为下垫面环境因子,比较引入物源因子前后的滑坡易发性分区结果准确率和覆盖度。结果表明:K-Means–RF、K-Means–AdaBoost融合算法输出的高易发区覆盖率相对于K-Means单体算法分别提高9.3%、12.1%。两类融合算法的易发性分区准确率和泛化能力比较接近,K-Means–AdaBoost融合算法对于滑坡点的预测效果更优。考虑物源因子后的K-Means–RF和K-Means–AdaBoos融合算法易发性分区中的高易发区覆盖率分别提高14.2%和17.7%,召回率均提高12.1%。Landslides are the source and dynamic basis of the coupled movement of flash flood and sediment disaster in mountainous,the identification of landslide susceptibility areas is an important prerequisite for flash flood and sediment disasters prediction-prewarning and risk assessment.In the past,research about flash floods and sediment disaster prevention and control paid attention to the flood’s role while ignoring the effect of mass sources.To improve the prevention and control system of flash flood and sediment disasters in the medium and small mountainous catchments,a landslide susceptibility area early identification method based on ensemble learning was proposed,and an optimization experiment for data sample construction and influence factor selection process was conducted.The frequency ratio of factors on the underlying surface of landslide units was used as unsupervised learning algorithm data samples for clustering analysis, and non-landslide units are selected based on clustering algorithm susceptibility partitioning, which constituted ensemble learning algorithm data samples for landslide susceptibility partition-ing with landslide units. Accuracy and coverage of the results of landslide susceptibility partitioning for medium and small mountainous catch-ment was compared between the simplex algorithms and fusion algorithms. The accuracy and coverage of landslide susceptibility identification were compared before and after the introduction of the mass-source as the underlying surface factor. Direct interpretation signs of loose deposits in the study area was established through high-resolution satellite remote sensing images, loose deposits area in the study area was identified through visual interpretation, area-volume power law relationship of loose deposits was established through regressive analysis and the distribu-tion of the loose deposits in the study area was obtained. The mass source was regarded as the underlying surface factor, and the accuracy and coverage of landslide susceptibility areas

关 键 词：滑坡 易发性 影响因子 集成学习 

分 类 号：P642[天文地球—工程地质学]