Role of Groundwater Monitoring for Closure of Underground Nuclear Tests on the Nevada Test Site

Over 800 underground nuclear tests were conducted at the Nevada Test Site in a variety of hydrogeologic environments. As of the 1996 Environmental Impact Statement for the site, more than 100 million curies of radioactivity remained in the subsurface from these tests, much of it near or below the water table. The U.S. Department of Energy Environmental Management program is implementing a closure strategy for these sites that anticipates closure-in-place, natural attenuation, and institutional controls. Groundwater monitoring is a key component of this strategy, but its role is significantly evolved from that of a detection- or compliance-based monitoring concept. Indeed, monitoring is part of the integrated closure process itself, not an activity confined to a static post-closure period. The reasons for this evolution derive from the complex hydrogeologic conditions, the long time-frames of concern, and recognition that a significant degree of uncertainty is irreducible. The hundreds of test locations are grouped into Corrective Action Units that measure over 100 km2 in area and extend to depths in excess of 1000 m. Despite concerted data collection efforts, the technical basis for closure of these large regions relies heavily on complex numerical models of flow and transport. The inherent uncertainties in these models present challenges for reaching regulatory acceptance of closure, and challenges for confidently locating monitoring wells. The solution now being pursued for the NTS is to integrate model evaluation and monitoring. In addition to standard goals of contaminant detection and protection of human health, an explicit monitoring objective is to increase confidence in model results by assessing the reliability of model forecasts. The initial monitoring network is expected to eventually transition to a long-term closure design, with less emphasis on confidence-building as uncertainty in forecasts is reduced. The methodology for this iterative process of monitoring and model refinement will incorporate expert-judgment and Bayesian updating of model input parameters to provide a cost-beneficial monitoring network that is expected to reduce model prediction uncertainty. This approach to monitoring for these large and complex contaminant areas is consistent with the underlying reliance on model predictions and will ensure that water quality samples are collected in a manner and location that is consistent with the current understanding of contaminant flowpaths.