Geostatistics Applied to Groundwater Pollution. III: Global Estimates

Abstract

Kriging, a geostatistical technique, can be used to obtain the best (minimum estimation error), linear, unbiased estimates of contaminant concentrations at points within a contaminant plume from a set of measured contaminant concentrations. The present paper presents the theoretical and numerical procedures needed to combine these local estimates to obtain global estimates for the mean contaminant concentration in any specified portion of the plume. Procedures for computing global estimation errors are also presented. Global estimates and estimation errors can then be used to compute standard confidence intervals. The use of the developed methods is illustrated by a set of measured groundwater contaminant concentrations from a hazardous waste site.