Source Apportionment And Spatial Distribution Of The Heavy Metals In The Agricultural Soil And Rice In A Regional Scale

With rapid urbanization in China in recent years,heavy metals(HMs)polluted agricultural soil has exacerbated.Thus,soil and agricultural product HM pollution has been considered a challenging global issue.For this,in this study,we regarded a county in southern China where soil and crop HMs pollution were serious as the study area.Fristly,analyzed the pollution status of soil HMs,revealed the main sources of soil HMs paddy fields in a southern county and in Tai'an City in northern China by principal component analysis/absolute principal component scores(PCA/APCS)and positive definite matrix factor(PMF)receptor models,and the sources of the two regions were compared.The spatial structural features of soil HMs were obtained using geostatistics,and compared the spatial interpolation performance of the three methods:the multiple linear regression model(MLR),geographically weighted regression model(GWR),and geographically weighted regression kriging.Secondly,taking soil Cd with the most serious pollution in the county as a case,we used spatial simulated annealing(SSA)algorithm to optimize the layout of the samples.Thirdly,the pollution status,main sources and influence factors of rice HMs were analyzed.Then we used the GWR to establishe the prediction models of rice HMs,and compared with the MLR and BP artificial neural network(BP-ANN)prediction methods.Finally,we determined the rice non-production areas in the country.The main conclusions of this study are as follows:The average content of soil Cd of a county in southern China was 0.97 mg/kg,which exceeded the second-grade standardized critical value(SGSCV),and the excessive rate was 98.5%.The average concentrations of soil Pb,As,Cr,and Hg did not exceed their corresponding SGSCV,but the above four HMs exceed their corresponding SGSCV in some areas and the excessive rates of Pb,As,Cr,and Hg were 0.11%,3.39%,0.32%and 6.04%,respectively.PCA/APCS analysis revealed three main sources of soil HMs in a county in southern China:industrial activities,agricultural activities and soil parent materials.Soil Pb,As and Hg mainly originated from industrial pollution,which accounted for 6.110 and 123%of the total source,respectively.The main sources of soil Cd and Cr were agricultural activeties and soil parent material,respectively.PMF analysis revealed the following four potential sources of agricultural soil HMs in Tai'an city:industrial and mining activities,agricultural activities,residential living activities,and business activities.Soil Hg mainly originated from residential living activities,which accounted for 75.3%of the total source.The main sources of soil Ni were residential living activities,agricultural activities,and industrial and mining activities,which account for 38.2,27.50.and 25.1%of the total source,respectively.Soil Cu was mainly produced by agricultural activities(36.6%),followed by residential living activities(29.8%)and industrial and mining activities(25.8%).The main sources of Cd,Pb,and Hg in the southern paddy field and the northern dryland soil were similar,but the sources of Cr were different.The Cr in the southern paddy soil mainly originated from the parent material,while the main source of soil Cr in northern dryland was industrial and mining activities.The results of the different methods were different.Therefore,when analyzing the source of soil HMs,it is necessary to choose the best method base on the distribution characteristics and actual conditions of soil HMs in the study area.The highest soil Cr concentration was obtained in the southeast of the study area,and the highest Cd,Pb,As and Hg concentrations were mainly distributed in the northeast of the study area.Topographic factors(including elevation),soil factors(including soil type,soil texture,soil organic matter and soil pH)and human activities(including coal mines,steel plants,chemical plants,and zinc oxide plants)effected the concentration and distribution of soil HMs.Four interpolation methods[GWRK,GWR,MLR,and ordinary kriging method(OK)]were used to predict the spatial distribution of five HMs(Cd,Pb,As,Cr and Hg)in soil.Results showed that the root mean square error(RMSE)of the Cd,Pb,As,Cr,and Hg obtained by GWRK was the lowest,and the mean absolute error(MAEE)was also lowest except for Pb and Hg in OK,indicating that GWRK was the best model for prediction of soil Cd,Pb,As,Cr and Hg in the study area.The maps obtained from GWR and GWRK could capture more local details than that of OK method,and it reduced the smoothing effect problem of OK approach.The GWRK outperformed the MLR by explicitly addressing spatial dependency.Results showed that GWRK enhanced the prediction precision for soil HMs compared to the GWR.This phenomenon was because the former takes into account spatial autocorrelation of the residuals.Therefore,GWRK was a more promising spatial interpolation method compared with other three methods in predicting soil HMs.The optimal sampling number of soil Cd obtained by SSA in the study area was 57.Cross-validation results show that the predict standard deviation of SSA was 0.801,which was better than simple random sampling and stratified sampling(1.269 and 0.991 respectively).The average concentrations of five HMs Cd,As,Cr,and Hg in rice were 0.47,0.1,0.23,0.005,and 0.03 mg/kg,respectively.Rice Hg concentrations didn't exceed the standard in all samples,while the other rice HM concentrations exceeded their corresponding standard in some areas and the percentages of Cd,As,Cr,and Pb were 61.51,4.22,3.24,and 0.65%,respectively.The main sources of rice HMs were soil HMs and industrial activities.The rice Cd.Pb and As were significantly positively correlated with their corresponding soil HM content,indicating that the rice HM was mainly affected by soil HM,and the crop species,soil pH,soil organic matter and external sources of pollution(including coal mines,chemical plants,zinc oxide plants,and major arterials).Three methods(GWR,MLR,and BP-ANN)had been used to predict the five HMs(Cd,Pb,As,Cr and Hg)in rice.Results showed that the correlation coefficient(Cd,Pb,As,Cr and Hg were 0.390,0.448,0.357,0.582,and 0.625,respectively)of the five HMs obtained by GWR were the highest,and the RMSE(Cd,Pb,As,Cr and Hg were 0.465,0.055.253,0.002,and 0.028,respectively)and MAEE(Cd,Pb,As,Cr,and Hg were 0.311,0.038.,0.178,0.001,and 0.020,respectively)were the lowest,indicating that GWR was the best method for prediction of rice Cd,Pb,As,Cr and Hg in the study area.The safe use and strict control areas of Cd account for 33.61 and 30.28%of the total cultivated land area,respectively,while As and Pb with 0.29%and 0.04%of the total cultivated area respectively belong to safe use area.Taking the HM in soil and rice into account,the rice non-production areas were determined.The strict control area accounted for 30.27%of the total cultivated area in the study area.It's necessary to take some measures to remediate the contaminated soil in this area.