Field Studies to Support Model Validation

All numerical models incorporate significant assumptions and approximations, and their predictions must always be regarded as estimates. Solution of the RANS equations, for example, requires some form of closure assumption dealing with the Reynolds stress terms. Since the Reynolds stress terms and the mean flow terms are coupled by the equations, inaccuracies in the closure approximations can affect the predicted mean flow fleld. Furthermore, the boundary conditions imposed on the model require the assumption of velocity profiles and momentum transport rates, which may themselves be approximated. Similar approximations are inherent in any of the various techniques used to compute the wind fleld, with further assumptions being present in each of the dispersion models. 

In recent years, a series of experiments has been undertaken in specific urban environments within the United States, with the specific objective of providing datasets for the evaluation of dispersion models being developed specifically for the urban environment. The first experiment of this series, Urban 2000, consisted of 

The Department of Homeland Security (DHS) has established a multiyear Urban Dispersion Program to understand the flow and dispersion in the deep street-canyon environment of New York City. The hrst experiment under this program, MSG 05, dealt with a limited number of releases in the vicinity of Madison Square Garden distribution of the results is currently restricted, though plume footprints without quantitative information are available (Allwine and Flaherty 2006). The experiment has also provided the basis for a comparison of the performance of CFD models in predicting flow patterns in this unique environment (Hanna et al. 2006). 

---