Spatial Prediction Of Regional Heavy Metals In Soils By Using The Bayesian Geostatistical Approach

With the rapid economic development,the environment was impacted by human activities.The concentrations of heavy metals in soils were increased by industrial,agricultural,and traffic emissions.If the contents exceeded the limit of the bearing capacity of the environment,the soil ecosystem would be threatened.Meanwhile,heavy metals could be absorbed by crops and accumulated through the food chain to influence human health.Therefore,it was necessary to investigate the concentrations of soil heavy metals using better spatial prediction approaches in the areas affected by human inputs,which could provide references for soil environment control and industrial and agriculture sustainability.There were many industrial factories in Zichuan,Boshan,and Zhoucun,including coal-fired power plants,refractory plants,ceramic manufacturers,mechanical processing plants,and petrochemical enterprises.Heavy metals accumulated in soils because of industrial activities.In addition,selenium-enriched agriculture was developed in recent years based on the resources of selenium-enriched land,but the application of pesticides and fertilizers might increase the level of heavy metals.Therefore,Zichuan,Boshan,and Zhoucun were selected as the study area in this work.A total of 265 samples were collected from surface soil(0-20cm)and subsurface soil(20-40cm)and analyzed the concentrations of ten heavy metals and Se,as well as soil’s physical and chemical properties,including organic matter,grain sizes of soil and p H,were tested.The descriptive information was calculated by using classical statistics,and the spatial patterns of heavy metals and Se were predicted by the Integrated nested Laplace approximations-stochastic partial differential equation(INLA-SPDE)approach.Meanwhile,the INLA-SPDE approach was compared with Ordinary Kriging(OK)and sequential Gaussian simulation(SGS)to verify the validity in the spatial prediction of the soil heavy metals.Besides,the land of the study area was divided into three levels according to the contents of Se and the selenium-enriched land with low heavy metal pollution risk was delimited based on the spatial distribution of the Se and Nemerow index.The results derived from this study were shown as follows.1)In the surface soils of the study area,the mean concentrations of 10 heavy metals were higher than the background value of Shandong Province.The enrichment factor of Cd,Cu,Hg,Pb and Zn were over 1,suggesting that they might accumulate in the surface soils.The results of the positive matrix factor(PMF)showed that As,Co,Cr,Mn and Ni were the natural sources,and Cd,Cu,Hg,Pb and Zn were influenced by human activities.The average contents of 10 heavy metals in the subsurface soils were higher than the background values of Shandong Province and the enrichment factor suggested that Cd,Hg,Pb and Zn accumulated in soils.According to the results of PMF,As,Co,Cr,Mn and Ni were the natural sources and Cd,Cu,Hg,Mn,Pb and Zn were influenced by industrial,agricultural activities and traffic emissions.2)According to the results of the INLA-SPDE approach,the distributions of As,Co,Cr,Mn,and Ni were associated with soil grain sizes or distributed randomly in the surface soils,suggesting that these five heavy metals were mainly controlled by natural factors.The concentrations of Cd,Cu,Hg,Pb and Zn were related to human activities because high-value areas of these five heavy metals were close to the agricultural,industrial parts,and traffic lines.In the subsurface soils,Co,Cr,and Ni were associated with natural factors.The high-value areas of Cd,Cu,Hg,Pb,and Zn were close to the industrial parks,towns with intensive human activities,suggesting the contents of these five heavy metals were related to the human inputs.Based on the results of standard deviation and 95%highest posterior density credible intervals(HPD CI),the samples density and natural environment factors were the major reason leading to the uncertainty of the spatial prediction.3)INLA-SPDE approach was an effective spatial prediction tool of soil heavy metals by comparing with OK and SGS,and it had faster computing speed,higher prediction accuracy,and better spatial expression.In addition,the INLA-SPDE approach could provide references for source identification of heavy metals by utilizing the fix effect paraments.Consequently,the INLA-SPDE approach could be regarded as a method providing support for soil heavy metal pollution control.4)The mean concentration of Se in surface soils was 0.34mg/kg and was 2.63 times its background value of Shandong Province in surface soils.The spatial distribution calculated by the INLA-SPDE approach showed that the content of Se was affected by human activities and the high-value areas were concentrated in the northwest,central,and southeast parts of the study area.In this work,0.3mg/kg was regarded as the cutoff value to divide the selenium-enriched land and non-selenium-enriched land.The result of spatial prediction showed that the area of selenium-enriched land was 423.49km~2 and the area of selenium-rich land with low heavy metal pollution risk was 411.98km~2 based on the method of Nemero comprehensive pollution index.