Geostatistical approach for managing uncertainty in environmental remediation of contaminated soils; case study

Sandia National Laboratories, Albuquerque, NM, United States

Variability in the spatial distribution of soil contaminants and limited (non-exhaustive) sampling create uncertainty in planned remediation activities at a site. Geostatistical simulation techniques can be applied to existing data to map the probability of exceeding a regulatory threshold (action level) across the site. This probability map can be used to devise a remediation plan based on an allowable probability of exceeding the regulatory action level that is acceptable to regulators and stakeholders. Compared with the traditional, deterministic approach to defining areas to be remediated, this probabilistic approach requires an acknowledgment of a nonzero probability of failure (failing to remove material actually exceeding the action level) that must be agreed upon by the regulators and stakeholders. Evaluation of costs for remediation alternatives and for potential regulatory failure can be combined with probability maps to provide site operators, regulators and stakeholders with cost-risk information from which key decisions can be effectively understood and negotiated. Approximately 300 samples were collected from a site in central New Mexico. These samples show concentrations of lead ranging from the 0.1 to 3,000 ppm. Initial negotiations with stakeholders and regulators resulted in two action levels being considered where each action level defined an acceptable human health risk for a residential or a industrial future land-use scenario. Probability-of-exceedence maps were created through geostatistical simulation at both action levels. Maps of areas to be remediated for several acceptable probabilities of exceedence, and the corresponding remediation costs as a function of proposed action level allowed the regulators and stakeholders to see clearly the economic consequences of their decisions. In the final decision, the regulators and stakeholders accepted a higher action level and a remediation map constructed with a five percent probability of exceedence. The cost difference between the selected action level, corresponding to an industrial future land-use scenario, and the alternative action level, corresponding to a residential future land-use scenario resulted in a projected savings in remediation costs of approximately $6.6 million.

This record provided courtesy of AGI/GeoRef.