Estimation of Spatial Patterns and Inventories of Environmental Contaminants Using Kriging

摘要

Kriging is a relatively new statistical approach to spatial estimation. The kriging estimator is a weighted average which is the best linear unbiased estimator. The derivation of the kriging weights takes into account the proximity of the observations to the point (or area) of interest, the structure of the observations (the relationship of the squared differences between pairs of observations and the intervening distance between them) and any systematic trend (or drift) in the observations. Additionally, kriging provides a variance estimate that can be used to construct a confidence interval for the kriging estimate. This paper will discuss the assumptions made in kriging and the derivation of the kriging estimator and variance. The application of kriging is demonstrated with lead measurements in soil cores from two sites near lead smelters and a third site in a control area. 21 references, 22 figures. (ERA citation 10:019069)