Morphological Characteristics Of Heavy Metals In Cultivated Soil Of Pingshui Village,Zhaoping County,Hezhou And Risk Assessment Based On Support Vector Machine

The random mining of the historical gold mines has caused great pollution to the surrounding soil in Pingshui Village,Wujiang Town,Zhaoping County,Hezhou City.Among which the problem of heavy metal pollution is more prominent.Therefore,the cultivated soil was selected as research object in this study in Pingshui Village,Wujiang Town,Zhaoping County,Hezhou City.A total of 34 soil samples were collected systematically.The total amount of heavy metal elements which Cr,Ni,Cu,Zn,As,Cd,Pb and Hg were determined.And p H value was determined in soil,too.The BCR method was used to extract the forms of heavy metals in the soil.And the content characteristics and morphological distribution of heavy metal elements in the soil were analyzed.The traditional evaluation method and the soil heavy metal risk assessment method based on support vector machine were used to evaluate the ecological risk.The soil heavy metal pollution was discussed and suggested by the research in the study area.On the basis of risk assessment,the prediction model of support vector machine was used to predict the available content of soil heavy metals in Cd and As elements,which provided reference for local pollution control.The results showed that:(1)The average contents of Cr,Ni,Cu,Zn,As,Cd,Pb and Hg were 2.36,1.43,1.92,1.95,5.97,8.77,2.89 and 8.89 times of the background values of Guangxi soil respectively in the study area.The average contents of Cr,As,Cd and Hg exceeded the screening values of agricultural land soil,which were 1.29,3.06,7.90 and 1.02times of the screening values respectively.(2)There was little difference in the distribution characteristics of heavy metals in soil in the study area.The Cr,Ni,Cu,Zn,As,Cd,Pb and Hg were residual state mainly.The proportions of heavy metals were 94.20%,95.01%,84.00%,84.35%,83.85%,93.66%,81.99%and 96.69%respectively.The proportion of weak acid extractable Cd and Zn in individual sampling points is relatively high,which needs attention.(3)The geo-accumulation index analysis showed that Hg and Cd were in a moderate-high pollution state.As was in a moderate pollution state,and the other heavy metal elements were in a mild or no pollution state.This potential ecological danger index showed that Hg and Cd had high ecological risk.The As is a medium ecological risk.The remaining heavy metal elements are minor ecological risks.the comprehensive potential ecological risk is high.The comprehensive potential ecological risk is high in the study area.The results of risk assessment coding method showed that As,Hg and Cr had no pollution risk.Cd,Cu,Ni and Pb had slight pollution risk,and Zn had moderate pollution risk.The analysis of heavy metal pollution evaluation method based on support vector machine shows that the high-risk points account for a large proportion and the comprehensive ecological risk is higher in the study area.(4)Support vector machine prediction models SVM-L and SVM-R with linear kernel functions and Gaussian kernel functions were constructed to predict the effective state contents of soil heavy metals Cd and As in the remaining sampling sites using the total and effective state contents of soil heavy metals Cd and As in some sampling sites and soil p H values.Both models showed good prediction ability on the training and test sets,with SVM-R having a higher coefficient of determination(R~2),lower mean absolute error(MAE)and root mean square error(RMSE),and a prediction value/actual value(P/O)closer to 1 than SVM-L,resulting in more satisfactory prediction results.