The Department of Defense is in the process of assessing and remediating closed, transferred, and transferring ranges. It is estimated that over 20 million acres of land in the United States potentially contain unexploded ordnance (UXO). Many of these sites involve very large geographical areas such that it is often impractical and/or cost prohibitive to perform 100 percent surveys of the entire site of interest looking for potential targets or individual UXO. Moreover, once a target area is identified, statistical sampling and analysis methods are needed to map the area in support of remediation and verification activities. Statistical sampling methods and tools were developed for several phases of the characterization and decision making process including optimal sampling designs and data analysis tools for target detection and delineation, anomaly/UXO density estimation, and post-remediation verification.
The Pacific Northwest National Laboratory and Sandia National Laboratory have developed statistical methods and tools for UXO characterization and mapping under SERDP-sponsored projects. Consistent with a Data Quality Objectives approach, methods are capable of devising an optimal survey scheme that ensures a high probability of detecting a target area of a specified size, shape, and anomaly density of concern. Methods for evaluating the performance of a meandering pathway also have been developed. Once targets are identified, other geostatistical and Bayesian methods have been established to support mapping of anomaly density or to obtain probability maps depicting the probability of at least one UXO or anomaly at all locations. Other statistical methods for determining the number of geophysical transects required to confidently demonstrate no or very few UXO remain at a site after remediation also have been developed. These methods have been programmed into Visual Sample Plan, a multi-agency sponsored tool for statistical sampling design and analysis. This project will support enhancements, validation, and testing of the methods against simulated data and at actual sites where UXO is an issue.
User-friendly software tools for evaluating the trade-offs between increased risk of not finding UXO versus the cost of more detailed characterization will be provided. These evaluations will support site-specific assessments of cost savings. The statistical sampling and geostatistical analysis methods will provide essential support for decisions relative to optimal sampling designs for target detection and delineation, anomaly/UXO density estimation, and post-remediation verification. Adequate defense for a less than 100 percent survey of very large areas could be achieved along with optimal survey strategies at all stages in the characterization, remediation, and verification cycle, thereby significantly reducing costs and optimizing resource allocation.