Value of Information Analysis to Support Data Collection for Characterizing Radionuclide Transport at the Nevada Test Site

To prioritize data collection for characterizing transport of radionuclides in groundwater away from the Rainier Mesa/Shoshone Mountain area of the Nevada Test Site, a value of information analysis (VOIA) was conducted. A VOIA is an approach to evaluating data-collection alternatives that uses Bayes' theorem to estimate how the collected data would update our understanding of the uncertainties important to decision makers. With this approach, current uncertainties are described by prior probability distributions that the data collection activities modify into posterior probability distributions. To conduct the analysis, a generic pathline transport model was created from the Nevada Test Site regional groundwater flow and transport model. Statistical sensitivity analysis techniques were used to identify key uncertain parameters based on the results of probabilistic simulations. An expert panel was convened to assess prior probability distributions for the key uncertain parameters, identify characterization activities, and estimate the ability of those activities and groups to reduce uncertainty over the key parameters of the transport model. A Monte Carlo analysis using the transport model was used to develop prior and posterior probability distributions for the contaminant boundary. The activities and groups of activities were prioritized based on several Bayesian metrics summarizing the amount of learning that would occur in the contaminant boundary probability distribution vis-à-vis the cost of these activities. This paper summarizes the results from the VOIA presented in detail by Deshler et al.