地统计学方法在环境污染研究中的应用

空间自相关性是环境变量普遍具有的一个统计特征.地统计学方法是描述、模拟和预测这一类变量空间变异特征的有效工具.本文概述地统计学方法在环境污染研究中的应用情况,主要包括:通过变异函数分析来描述和模拟环境变量的空间分布结构;由克立格插值技术来估算环境变量在未采样点处的取值;采用多元地统计学方法研究多个环境变量间的空间相关结构;通过指示地统计学方法评价空间插值的局部不确定性以及进行环境变量空间分布的随机模拟等.同时结合实例分析,简要探讨了地统计学方法在描述、预测环境变量空间结构特征和空间相关分析方面的优势.

Geostatistical Methods for Research on Environmental Pollution

Autocorrelation is a general statistical property of environmental variables. Geostatistics provides useful tools to describe, model and predict spatial heterogeneity of environmental variables.The main applications of geostatistics to the research on environmental pollution include introducing semivariance analysis to characterize and model the spatial structures of environmental variables; using kriging methods to estimate values of environmental variables at unsampled locations; adopting multivariate geotatistics to detect the spatial correlation structures between different environmental variables; assessing the local uncertainty of predicted unsampled values by indicator geostatistics, and performing stochastic simulation for the spatial distributions of environmental variables.The advantages of geotatistics analysis in characterizing and predicting spatial structures and spatial correlations of environmental variables are briefly discussed using a case study.