基于遥感影像的溃坝风险人口估算模型研究  

Study on Population Estimation Model of Dam Break Risk Based on Remote Sensing Images

作　　者：余再康 程井[2] 余雷[1] 向思瑶 李宗樾[1] YU Zaikang;CHENG Jing;YU Lei;XIANG Siyao;LI Zongyue(Guizhou Dam Safety Monitoring Center,Guiyang 550002,China;College of Water Conservancy and Hydropower Engineering,Hohai University,Nanjing 210098,China)

机构地区：[1]贵州省大坝安全监测中心,贵州贵阳550002 [2]河海大学水利水电学院,江苏南京210098

出　　处：《人民黄河》2025年第4期44-50,共7页Yellow River

基　　金：水利部重大科技项目(SKS-2022153);国家重点研发计划项目(2022YFC3005501);贵州省水利科技经费项目(KT202217,KT202218)。

摘　　要：针对小型水库生命损失风险评估中面临的下游人口分布资料缺乏的问题,提出了一种基于Sentinel-2A遥感影像的溃坝风险人口估算模型。以DYJ水库为例,采用监督分类法快速提取遥感影像中乡村居民房屋建筑信息,结合溃坝洪水演进结果,以洪水是否淹没到房屋建筑为判据,对最大似然法、支持向量机法、马氏距离法和最小距离法4种监督分类方法进行对比,结果表明:最小距离法总分类精度只有65.7%,其余3种方法总分类精度均在90%以上,其中最大似然法对遥感图像识别的精度最高,总分类精度达到97.55%,Kapps系数为0.93,能较为准确地提取研究区域内的居民建筑要素。采用最大似然法提取建筑信息,并结合溃坝风险人口估算模型,最终得到DYJ水库下游风险人口为1470人。该模型可快速估算小型水库下游风险人口,弥补了人口密度法和居民点累计估算法在精度和效率方面的不足。A risk population estimation model based on Sentinel-2A remote sensing images was proposed to address the lack of downstream population distribution data in the risk assessment of life loss for small reservoirs.Taking the DYJ reservoir as an example,supervised classification was used to quickly extract the building information of rural residents from remote sensing images,and combined with the evolution results of dam break flood,a risk population estimation model was built considering the flooded state of the buildings.Four supervised classification methods,including Maximum Likelihood Classification(MLC),Support Vector Machine(SVM),Mahalanobis Distance Method(MDM),and Minimum Distance Classification(MDC),were compared.The results show that the MDC has an overall classification accuracy of only 65.7%,while the other three methods achieve over 90%accuracy.Among them,the MLC provides the highest recognition accura-cy for remote sensing images,with an overall classification accuracy of 97.55%and a Kappa coefficient of 0.93,in extracting residential building elements in the study area.With the construction information extracted by MLC and the dam break risk population estimation model,the risk population downstream of the DYJ reservoir is estimated to be 1470 people.This model can rapidly estimate the risk population downstream of small reservoirs,addressing the shortcomings in accuracy and efficiency of population density methods and cumulative estimation methods for residential areas.

关 键 词：小型水库 遥感影像 风险人口 溃坝 洪水演进 

分 类 号：TV621[水利工程—水利水电工程] TP751[自动化与计算机技术—检测技术与自动化装置]