The model evaluation study was designed to determine the effectiveness of each institution's modeling system in terms of the ability to respond appropriately to an urban dispersion concern in the recovery phases of an event. The study was conducted such that an evaluation team, independent of the six modeling teams, provided all of the input data (e.g., building data, meteorological data and tracer release rates) and run conditions for each of four experimental periods simulated. Tracer concentration data for two of the four experimental periods were provided to the modeling teams for their own evaluation of their respective models to ensure proper setup and operation. Tracer data were not provided for the second two experimental periods to provide for an independent evaluation of the models.
The results of this study revealed that all models were imperfect, and that the best simulations had a 70% overlap with the plume and 40% of the concentrations simulated within a factor of two of the measurements. It appeared that differences in model performance generally resulted from differences in approach flow conditions prescribed by the modelers. Future model development efforts needed to address modeling gaps identified from this study will also be discussed.