## Iterative modeling strategy and uncertainty

The second change in the strategy was designed to better represent the pragmatism of an iterative modeling approach focused on quantifying and attempting to reduce uncertainty sufficient to support regulatory decisions. Multiple decision points were added between NSO and NDEP at critical steps in the overall progression of UGTA studies. Each decision represents a juncture between continuing forward in the strategy progression or looping back (iterating) through studies. For example, a decision point was added at the end of development of a flow and transport model in the second stage to assess whether the data and model results are adequate. If both are judged adequate, the studies proceed to an external peer review. If either the data or model results are judged inadequate, the studies return to additional site characterization, refined modeling studies, and sensitivity and uncertainty analysis, all essential parts of an iterative modeling cycle.

Uncertainty in UGTA studies, particularly modeling uncertainty, was also reassessed in the strategy revisions. The multiple components of uncertainty in modeling studies are divided into statistical and structural uncertainty following guidelines established in the uncertainty literature (Morgan and Henrion, 1990; Krupnick et al., 2006). Statistical uncertainty includes knowledge uncertainty and variability as a subset of knowledge uncertainty; structural uncertainty refers to model, conceptual model, and decision and regulatory uncertainty.

Reassessment of uncertainty in UGTA studies led to changes in both the approach and output of modeling studies. Modeling under the original strategy emphasized development of a preferred model of groundwater flow and radionuclide transport. However, the external peer review of the modeling studies for Frenchman Flat (IT Corporation, 1999) concluded that a single model result did not adequately represent the full range of potential model responses. The revisions in the UGTA strategy are designed to emphasize development of multiple alternative model responses that represent a spectrum of permissive combinations of model output using multiple alternative models of the hydrologic conditions and geologic setting of flow and transport in the NNSS.