Application of geostatistics to expedited site characterization
John M. Pitt
This thesis develops geostatistical methods for use in the USDOE Expedited Site Characterization (ESC) process with focus on application to sites with contaminated soil and groundwater. Statistical methods for on-site sample location selection for geological and environmental sampling, characterizing uncertainty in the geologic and contaminant models, modeling the spatial distribution of contamination in the presence of non-detect data, determining when sufficient data have been collected, and post investigation analysis of geological and environmental data are given. These statistical methods are applied to data from an ESC demonstration at a former manufactured gas plant where soils were contaminated with polynuclear aromatic hydrocarbons and an ESC project at the Savannah River Site in South Carolina to characterize metal, pesticide, and volatile organic contamination in groundwater. A recommended approach to the use of geophysical methods and direct push technologies, in conjunction with geostatistical methods, for the characterization of the geologic environment and contaminant spatial distribution is developed. A summary of the USEPA Data Quality Objectives process, the USDOE Streamlined Approach for Environmental Restoration, and the USEPA Site Accelerated Cleanup Model is given, with focus on how ESC fits into the environmental restoration process.