基于二分类Logistic回归模型土壤Cu污染风险预测  

Predicting the Pollution Risk of Cu in Soil Using Binary Logistic Regression

作　　者：张子健 李湘凌[1,3] 卢新哲 徐景烨 魏迎春 ZHANG Zi-jian;LI Xiang-ling;LU Xin-zhe;XU Jing-ye;WEI Ying-chun(School of Resources and Environmental Engineering,Hefei University of Technology,Hefei 230009,China;Technology Innovation Center of Ecological Evaluation and Remediation of Agricultural Land in Plain Area,Ministry of Natural Resources,Hangzhou 311200,China;Anhui Province Engineering Research Center for Mineral Resources and Mine Environments,Hefei 230009,China;Zhejiang Institute of Geological Survey,Hangzhou 311200,China)

机构地区：[1]合肥工业大学资源与环境工程学院,安徽合肥230009 [2]自然资源部平原区农用地生态评价与修复工程技术创新中心,浙江杭州311200 [3]安徽省矿产资源与矿山环境工程技术研究中心,安徽合肥230009 [4]浙江省地质调查院,浙江杭州311200

出　　处：《土壤通报》2021年第6期1418-1426,共9页Chinese Journal of Soil Science

基　　金：浙江省自然资源厅科技项目(No.2020-45);浙江省地勘资金项目(【省资】No.2019007,No.2020006);浙江省重点研发计划项目(No.2021C04020)资助。

摘　　要：土壤重金属来源广泛,其污染风险受到多种因素综合影响,常与影响因素存在非线性关系。Logistic回归法可克服普通线性回归法的应用限制,对土壤污染风险的影响因素识别和污染风险预测中具有良好的应用潜力。以浙江某人类活动强烈影响区域土壤为研究对象,选择坡度、坡向、海拔、人口密度、道路距离、非矿工业加权距离和到矿区距离为自变量,以土壤Cu污染风险分类为因变量,采用二分类Logistic回归模型,识别土壤Cu污染风险影响因素并预测污染风险。结果表明:土壤Cu污染风险受到人口密度、矿点距离、道路距离和非矿工业加权距离的影响显著,但土壤Cu污染风险与人口密度、矿点距离和道路距离均无简单线性关系,密集矿点的叠加作用显著改变了距离对土壤污染风险的影响。区内土壤Cu污染高风险区主要位于铜矿采选及其尾矿库影响区域,中风险区位于破碎带金矿、铁矿区及交通运输影响范围内,低风险区位于低密度人口和高密度人口区域。Soil heavy metals come from a wide range of sources,whose pollution risks are influenced by varieties of factors.Usually,there is a nonlinear relationship between the pollution risk and the influencing factor.Logistic regression method could overcome the limitation of ordinary linear regression method,and has good application potential in the identification of influencing factors and the prediction of soil pollution risk.Taking the soil in a certain region strongly influenced by humans in Zhejiang Province as the object,the influencing factors of soil pollution risk and predicting the pollution risk were investigated based on the binary logistic regression model,with the independent variables of slope,aspect,altitude,population density,road distance,weighted distance of non-mining industry and distance to mining area,and the dependent variable of risk classification of soil Cu pollution.The results showed that the risk of soil Cu pollution was significantly affected by population density,distance to mining area,road distance,and weighted distance of non-mining industry,but there was no simple linear relationship between it and population density,distance to mining area and road distance.The additive effect of intensive mines dramatically changed the distance effect on the soil pollution risk.The high risk area of soil Cu pollution was mainly located in the Cu mining area and its influence area of the tailings pond,the medium risk area were located in the gold deposit of fracture zones,iron producing area and transportation influenced zone,and the low risk area was located in the areas of high and low density population.

关 键 词：土壤 污染风险 铜 二分类 LOGISTIC回归 预测 

分 类 号：X825[环境科学与工程—环境工程]