Using Geostatistics To Evaluate Cleanup Goals

Michael F. Marcon and Loren P. Hopkins Abstract Geostatistical analysis is a powerful predictive tool typically used to define spatial variability in environmental data. The information from a geostatistical analysis using kriging, a geostatistical tool, can be taken a step further to optimize sampling location and frequency and help quantify sampling uncertainty in both the remedial investigation and remedial design at a hazardous waste site. Geostatistics were used to quantify sampling uncertainty in attainment of a risk-based cleanup goal and determine the optimal sampling frequency necessary to delineate the horizontal extent of impacted soils at a Gulf Coast waste site. Introduction Statistical assessment of pre- and post remediation data is a critical step in evaluating contaminant characterization and attainment of cleanup standards. This paper discusses the incorporation of geostatistical methods in sampling strategies at hazardous waste sites. A Phase I Contaminant Assessment (conducted in two stages) completed at a chemical blending facility in the Gulf Coast indicated that eight compounds had significantly impacted superficial soils. A second phase sampling plan was developed using geostatistical tools and the Data Quality Objective (DQO) Process (1) to define the horizontal extent of contamination which exceeded preliminary remediation goals. The purpose of the Phase II sampling plan was to identify all one half acre plots on the site which potentially posed an unacceptable risk to human health and the environment. The Phase II program was based on: preliminary remediation goals (PRGs) established during the risk assessment; the future land use for the property; uncertainty levels acceptable to both the Environmental Protection Agency (EPA) and the property owner; and the amount of expected concentration variability observed at the site. These inputs were used in a geostatistical analysis to develop a final sampling program to delineate the horizontal extent of contamination and eliminate the need for confirmatory sampling after remediation was completed. P. 19