核密度分析及自举法估计曹妃甸口岸进口铁矿石汞含量的代表值

Estimation of representative value of mercury content in iron ore imported from Caofeidian port by kernel density method coupled with Bootstrap method

对曹妃甸口岸进口的641个铁矿石样品的汞含量进行了测定,采用正态概率分布函数验证数据的正态性,利用核密度分布估计法分析了数据的多态性。用自举法(Bootstrap)分别对整体样本和子样本进行重复模拟取样,估计了汞含量的代表值及置信区间。结果表明,整体样本的汞含量呈非正态分布,核密度分布图能直观反应汞含量分布的多态性。整体样本汞含量的代表值为0.066 3μg/g,澳大利亚、巴西、秘鲁、日本、南非、乌克兰、智利、印度的铁矿石汞含量代表值依次为0.057 3、0.077 4、0.056 8、0.003 0、0.007 8、0.013 0、0.002 9、0.047 8μg/g。自举法模拟重新取样估计有限单次样本的代表值是稳健、有效的,这为有害元素风险评估中元素含量的估计提供了一种新方法。

The mercury content in 641 iron ore samples imported from Caofeidian port was determined.The normality of data was verified by normal probability distribution function,and the polymorphism of data analyzed by kernel density distribution estimation.The repetitive simulation sampling of entire sample and subsample was conducted by Bootstrap method,respectively.The representative value and confidence interval of mercury content were estimated.The results indicated that the mercury content in entire sample showed abnormal distribution,while the kernel density distribution diagram could intuitively reflect the polymorphism of mercury content distribution.The representative value of mercury content in entire sample was 0.066 3μg/g.The representative value of iron ore from Australia,Brazil,Peru,Japan,South Africa,Ukraine,Chile and India was 0.057 3,0.077 4,0.056 8,0.003 0,0.007 8,0.013 0,0.002 9and0.047 8μg/g.The estimation of representative mercury content in finite samples through repetitive simulation sampling by Bootstrap method was robust and effective.It provided a new method for the estimation of elemental content in risk assessment of harmful elements.