含未检出值的环境监测数据集的处理方法

Introduction to the methods of parameter estimation for environmental monitoring data set with truncated data below a detection limit

环境样品分析过程中经常遇到包含未检出值的数据集,这在对含有微量金属和微量有机污染物样本的分析研究中更为常见.目前处理未检出数据的方法通常是忽略或者用0、检出下限的1/2或1/3进行简单替代.然而,当未检出数据比例较高时,忽略或简单替代可能给数据表征带来较大偏差,也不利于更深入的统计分析.为改善对包含未检出数据的分析效果,目前已经建立了多种统计方法.本文系统介绍了这些方法,包括总体均值和方差估计的参数和非参数方法以及关于均值的95%置信区间上界的估计方法.此外,本文利用虚拟数据和实测数据给出了主要方法的应用实例.

Data sets containing nondetects that are below detection limits, are inevitable in analytical results for environmental samples, especially for trace elements and trace organic pollutants. Commonly, the nondeteets were either ignored or simply substituted by zero, half or one third of the detection limits. However, these approaches would result in biased representative statistics and difficulty in further statistical analysis, especially when there was a large percentage of nondetects. In fact, a number of methods have been developed to handle the data and to provide the best estimation of the population parameters based on the data above the detection limits. These methods, including both parametric and distribution-free are introduced in this article. The methods for calculating the upper limits of the 95% confidence intervals of population means are also presented. The method application is illustrated based on both virtual and field data sets.