Application of User-Oriented MOE to HPAC Probabilistic Predictions of Prairie Grass Field Trials

摘要

The goal of this task is to improve the verification, validation, and accreditation (VV&A) of hazard prediction and assessment models. These studies are part of a larger joint VV&A effort that DTRA and the Department of Energy, via the Lawrence Livermore National Laboratory (LLNL), are conducting. The Institute for Defense Analyses role includes conducting comparisons between models, providing analysis and discussions associated with these examinations, and exploring and developing measures of effectiveness (MOE) that can aid hazard prediction model validation and accreditation. This paper develops and demonstrates concepts that can ultimately aid validation and user accreditation of transport and dispersion models. The paper applies a user-oriented MOE to the probabilistic outputs that are obtained from the Hazard Prediction and Assessment Capability (HPAC) model predictions of the short-range Prairie Grass field trials. This MOE allows for the assessment of the probabilistic predictions in terms of the false negative and false positive fractions. Quantitative descriptions of a user s potential risk tolerance are also described in this paper. This study develops these quantitative descriptions and uses them to evaluate the goodness of HPAC predictions of the Prairie Grass field trials from the perspective of a few notional users.