基于非参数核密度估计模型的尾矿库事故后沉积物中Pb和Hg的水生生物风险评估

汞(Hg)和铅(Pb)是2种典型的重金属污染物,对水生生态系统具有较强的毒害作用,是环境管理的重要指标。2015年11月的陇星锑业尾矿库泄露事件,造成嘉陵江流域约346 km长河段受到重金属污染,尾矿砂中高含量Hg和Pb主要归宿在污染团经过的河道沉积物中,引起的嘉陵江流域生态风险尚未被全面评估。为解决沉积物中污染物对水生生物的毒性效应数据较少的问题,采用相平衡分配法,利用大量的水生生物毒理试验数据,将其转化为相应的沉积物毒性效应数据;采用基于非参数核密度估计的物种敏感度分布(SSD)法,对嘉陵江沉积物中2种重金属(Hg和Pb)进行生态风险评价,并与其他分布模型(Normal、Logistic和Weibull)进行对比。结果表明:Hg的非参数核密度估计模型的K-S检验统计量、均方根误差(RMSE)和误差平方和(SSE)分别为0.111 1、0.025 04和0.000 627,相较其他分布模型为最小;Pb的非参数核密度估计模型的K-S检验统计量为0.125 0,相较其他分布模型为最小,RMSE和SSE分别为0.028 42和0.000 807,为较优。非参数核密度估计模型对2种重金属毒性数据有很好的适应性,可获得较优的模拟效果。嘉陵江流域15个采样点沉积物中Pb浓度显著高于Hg,但沉积物中Hg的生态风险水平远高于Pb。

Aquatic biological risk assessment of Pb and Hg in sediments after tailings reservoir accidents based on non-parametric kernel density estimation model

Mercury (Hg) and lead (Pb) are two typical heavy metal pollutants, which have strong toxic effects on aquatic ecosystems and are important indicators of environmental management. The leakage of Longxing Antimony Industry Tailings Pond in November 2015 had caused heavy metal pollution in a river section of about 346 km. High content of Hg and Pb in the tailing sands mainly end up in the river sediments that the pollutant passed through and the ecological risks of the basin caused by the high content of Hg and Pb in the tailings sands had not been comprehensively assessed scientifically. In order to solve the problem of less data on the toxic effects of pollutants in sediments on aquatic organisms, the phase equilibrium distribution method was adopted to use a large number of aquatic organisms toxicological test results and converted them into corresponding sediment toxic effect data. The non-parametric kernel density model was used to evaluate the ecological risks of Hg and Pb and compare it with three distribution models of Normal, Logistic and Weibull. The K-S test statistics, root mean square error (RMSE) and sum of square error (SSE) of Hg non-parametric kernel density model were 0.111 1, 0.025 04, and 0.000 627, respectively, which were the smallest compared to other models. The K-S test statistics of the non-parametric kernel density model of Pb was 0.125 0, which was the smallest amount, and RMSE and SSE were 0.028 42 and 0.000 807, respectively, which was a better result. The results showed that the non-parametric kernel density estimation model had good adaptability to the toxicity data of the two heavy metals and could obtain better simulation results; Pb concentration in the sediment was significantly higher than that of Hg, but the ecological risk level of Hg in the sediment was far higher than that of Pb.