Regression Of Influencing Factors And Spatial Prediction Of Heavy Metal Content In Suburban Agricultural Land

An accurate understanding of the distribution of heavy metals in the soil of agricultural land in the suburbs and various sources of pollution is of great significance for preventing risks to the soil environment of surrounding agricultural land during the process of urbanization,scientifically planning the industrial layout,and maintaining human health and safety.This paper uses agricultural land in the suburbs of a southern county as the study area.The study area is about 122.08 km~2.There are a large number of industrial enterprises in the area.There are obvious differences in human activities between the north and the south.A total of 86 samples were collected to test and analyze the five heavy metals in the soil(Cd,Pb,Zn,Cr,Ni)and soil p H.At the same time,the environmental factors(Soil p H,elevation,river system,traffic road,four types of industrial enterprises,residential areas)what may affect the spatial distribution of heavy metals in the study area.Based on the above data,this paper systematically studied the content distribution,spatial structure characteristics of heavy metals,and integrated multivariate statistical methods,PMF with Geodetector to analyze and judge their main sources and influence factors,so as to use heavy metals.Environmental factors with significant spatial distribution are used as auxiliary variables,combined with geographic weighted regression kriging(GWRK)to perform quantitative regression prediction in order to more accurately describe the spatial distribution characteristics of heavy metals in the study area.The conclusions of this study are as follows:(1)The average values of the five heavy metals(Cd,Pb,Zn,Cr,Ni)of the agricultural land in the study area are 2.08,110.02,191.79,88.41,27.96 mg/kg,and the median values are 1.76,90.25,168.50,88.10,27.95 mg/kg,the coefficients of variation were 53.03,52.29,36.40,15.87,16.08%,respectively.The average and median values of Cd,Pb,and Zn are much higher than the local soil environmental background value.There are no points where Cr exceeds the background value of the local soil environment,and the average value of Ni is slightly higher than the background value of the local soil environment.The heavy metals Cd,Pb,and Zn are significantly enriched in the agricultural land in the suburbs of the city,which has the typical characteristics of soil pollution in the suburbs.Integrating the calculation results of Moran’s I index and semi-variance function model,it is found that Cd,Pb,Zn,and Cr in the study area have local spatial heterogeneity;Ni has no local spatial heterogeneity,and Cd,Pb,Zn,Cr,and Ni With certain spatial structure and randomness,Cd,Pb,Zn,and Cr can be predicted and estimated by using geographic weighted regression model combined with Kriging interpolation,while Ni element is not suitable for geographic weighted regression model fitting.(2)According to the statistical analysis of the heavy metal content under different gradients of the source analysis model and environmental impact factors,combined with the results of the geographic detector model,the spatial distribution of Cd,Pb,and Zn is affected by industrial enterprises,transportation activities and elevation,and the specific environmental impact factors.The order of importance is:Cd:Electrical machinery industry enterprises>metal products industry enterprise-s>non-ferrous smelting and processing industry enterprises>elevation>other industry e-nterprises>roads.Pb:Electrical machinery industry companies>Metal products industry companie-s>Elevation>Roads>Non-ferrous smelting and processing industry companies>Other i-ndustry companies.Zn:Electrical machinery industry enterprises>metal products industry enterprise-s>elevation>non-ferrous smelting and processing industry enterprises>other industry e-nterprises>roads.The spatial distribution of Cr is only affected by topography(elevation);the spatial distribution of Ni has no obvious relationship with the environmental factors selected in the study.Soil p H did not significantly affect the spatial distribution of the five soil heavy metals.(3)The use of environmental factors that have a significant effect on the spatial distribution of heavy metals as auxiliary variables and based on geographically weighted regression kriging(GWRK),ordinary multiple linear regression kriging(MLRK),and ordinary kriging(OK)The combined interpolation method is used to construct Cd,Pb,and Zn spatial distribution prediction models for accuracy comparison.The GWRK model of Cd(R~2=0.521;K=1.005)has better prediction accuracy than MLRK(R~2=0.421;K=1.230)and OK(R~2=0.376;K=1.328)methods;The GWRK model of Pb(R~2=0.551;K=2340.096)has better prediction accuracy than MLRK(R~2=0.310;K=3294.829)and OK(R~2=0.203;K=3880.793)method;The GWRK model of Zn(R~2=0.628;K=3652.677)has better prediction accuracy than MLRK(R~2=0.557;K=4059.093)and OK(R~2=0.485;K=4807.313),the GWRK method has the best fitting accuracy for Cd,Pb,Zn distribution prediction.After using three prediction models to map the spatial distribution of Cd,Pb,and Zn in all 86 samples in the study area,it is found that the prediction range of the GWRK method is closest to the actual content,which reduces the inherent smoothing effect of ordinary kriging.The spatial prediction distribution map more in line with reality.