The computational fluid dynamics (CFD) approach has been increasingly applied to many atmospheric applications, including flow over buildings and complex terrain, and dispersion of hazardous releases. However there has been much less activity on the coupling of CFD with atmospheric chemistry. Most of the atmospheric chemistry applications have been focused on the modeling of chemistry on larger spatial scales, such as global or urban airshed scale. However, the increased attention to terrorism threats have stimulated the need of much more detailed simulations involving chemical releases within urban areas. This motivated us to develop a new CFD/coupled-chemistry capability as part of our modeling effort.

The model we have developed is based on the FEM3MP CFD model with a coupled chemistry submodel, which uses SMVGEAR II as the chemistry solver. The CFD/Chemistry model has initially been applied to chemically reactive releases around buildings. These local-scale flow computations are typically based on domain sizes of one to a few kilometers square with highly graded meshes to resolve the complicated flows associated with building wakes interactions. A selection of these simulations will be presented at this conference.

More recently, we have explored the use of the CFD approach to study air quality within urban areas. It is clear that some compromises in grid resolution have to be made due to the larger domain sizes that are of interest. However, even with the somewhat coarser grid sizes, the typical grid resolutions are approximately an order of magnitude finer than that used in mesoscale simulations. With the finer resolution, CFD calculations can resolve terrain and can even model building effects with higher fidelity. This should lead to more accurate predictions of wind fields and dispersion patterns. Several challenges nevertheless must still be overcome. For example, the limited area representation requires boundary conditions that must be driven from large-scale forecasts. These boundary conditions include not only the usual meteorological fields, but also the chemical pollutants that advect into the domain from sources outside of the computational zone. Another issue that needs to be addressed is that the inventory of sources within the domain should be consistent with the finer grid resolution. Due to the somewhat unreliable estimates of release rates from power plants and unpredictable meteorological conditions, it is often difficult to obtain very precise spatial inventories of source terms.

In the air quality study, we have focused on the high ozone level episode that occurred in the Livermore Valley on July 31, 2000. Some preliminary simulations of the July 31 episode using two widely used mesoscale forecast models and an urban airshed model have not been able to recreate the occurrence of high ozone condition in the appropriate locations within the Valley. In this meeting, we will present some results of our simulations of wind and dispersion patterns for this particular scenario. The results are based on using relatively fine grid resolutions and concomitantly better representation of the terrain upwind of the Livermore Valley. Initial results show that the wind patterns are highly influenced by local topography that contributed to the channeling of the northerly mean flow down the valley towards the Livermore Valley.

Acknowledgement. This work was performed under the auspices of the U.S. Department of Energy by the University of California, Lawrence Livermore National Laboratory under contract No. W-7405-Eng-48.