Computational Models for the Prediction of Aerosol Dispersion

Since the dispersion of aerosols is driven primarily by advection, the sophistication and accuracy of the wind-field characterization provided to a model is the principal factor in determining the accuracy of the dispersion predictions provided by the model. This may range from  

Basing estimates on a wind field described by a single mean wind vector and an estimate of the atmospheric stability (in effect, the modeler in this situation would simply be using a computer to implement a Pasquill-Gifford-style prediction). 

Using live sensor data from the vicinity of the release to provide a characterization of the actual wind field in the area. In practice, there are currently very few locations in which a dense meteorological data sensor network is available to provide this data. 

More often, three-dimensional gridded meteorological forecast data provided by mesoscale meteorological models such as MM5 will be available. The resolution of the grid will typically be on the order of 1 km, which is too large to capture microscale surface features. Mesoscale model results can be coupled with fine-scale models by enforcing conservation of mass to provide enhanced fine-scale detail. 

Using detailed CFD simulations to capture the microscale wind-field features associated with obstacles and local terrain. Detailed CFD simulations presently take too long to be useful in rapid emergency response situations but can be useful in advance preparation and to develop understanding of local dispersion patterns, especially within urban areas (Hanna et al. 2006). 

---