Risk Prediction Of Soil Heavy Metal Pollution Based On Two Classification LR Method And Indicator Kriging Method

In industrial and mining areas with strong human activities,soil heavy metals are not only affected by industry,life,transportation,mining activities,but also affected by regional geological background,topography,vegetation coverage,rainfall and other natural factors.Identification of risk factors and spatial prediction of soil heavy metal pollution in industrial and mining areas have become a hot topic in environmental science.At present,most of the researches on risk prediction of soil heavy metal pollution use parametric geostatistics to carry out relevant research.Parametric geostatistics not only does not pay attention to the uncertainty(probability)of soil heavy metal pollution,but also directly eliminates the specific value in order to meet the requirements of normal distribution of data,which leads to the omission of important information and affects the prediction effect.On the other hand,parametric geostatistics can not predict the pollution of heavy metals in soil under the condition of insufficient data.Therefore,based on the concept of risk management,according to different data conditions,different nonparametric soil heavy metal pollution risk prediction methods are proposed,which has practical significance for soil pollution risk management in industrial and mining areas.In this study,the regional soil in Zhejiang Province is strongly affected by human activities as the research object.The content characteristics of heavy metals in rock ore,tailings and irrigation water are studied.The leaching law of heavy metals in typical rock ore and tailings is analyzed.Combined with the investigation data of heavy metals in regional farmland soil,remote sensing technology is used to extract natural factors such as slope,aspect,elevation and vegetation coverage The spatial pollution probability of soil heavy metals was predicted based on two classification LR method and indicator Kriging method,and the differences of different prediction methods were compared.The research results are as follows(1)The content of as and Cd in the rock ore and tailing of iron mine area is high,while the content of Cu and Pb in the ore and tailing of copper mine area is high.The waste rock or tailing of mining area has the ability to release heavy metals such as Zn,Cu,Ni,CD,Pb and as to the environment after being affected by rainwater.(2)Based on the two classification LR method,the main factors leading to the risk of soil heavy metal Cu pollution are slope aspect,population density,non mining industry weighted distance,mining area distance and road distance;the main factors affecting the risk of Cd pollution are rainfall,mining area distance,non mining industry weighted distance and population density.(3)Based on the main influencing factors of soil pollution risk,the two classification model of Cu and Cd pollution risk prediction was established.The accuracy of soil Cu and Cd pollution risk prediction was 82.6% and 69.5% respectively.The two classification LR method was better than CD in soil Cu pollution risk prediction.The high-risk area of soil Cu pollution is mainly located in the copper mining area,the medium risk area is located in the broken zone of gold mine,iron ore area and the scope of transportation,the low-risk area is located in the low-density population and high-density population area,and the high-risk area of soil Cd pollution is mainly located in the densely populated area,followed by the mining area.(4)Based on the indicator Kriging method,the soil Cd pollution risk is high,and the pollution risk of as,Cu,Hg,Pb and Zn is different;the pollution risk of Ni and Cr is low,and the prediction accuracy of pollution risk is more than 90%,which is higher than the two classification LR method.(5)In the construction of soil heavy metal pollution risk prediction model,the binary LR method has certain requirements for the proportion of over standard samples.In industrial and mining areas strongly affected by human activities,the binary LR method uses independent variables with geographical attributes to establish a regression model,and the regression model can be used to predict the spatial pollution risk of soil heavy metals,which can provide a better management idea for the area where the survey data of soil heavy metal pollution is lack,especially in industrial and mining areas.