The use of Computational Fluid Dynamics (CFD) models for studying transport and dispersion problems in complex urban environments is becoming increasingly popular, due to the rapidly improving ability of models to simulate large urban regions. Los Alamos National Laboratory has been performing high resolution CFD simulations of major metropolitan areas in support of DOE's Chemical and Biological Non-Proliferation (CBNP) Program. The primary goal of such simulations is the support of rapid response planning, and impact assessment for the release of chemical and/or biological agents.

Our CFD model (HIGRAD) is capable of simulating a wide variety of atmospheric physical phenomena. HIGRAD uses state-of-the-art numerical techniques to accurately predict the evolution of atmospheric phenomena from micro-scale to meso-scale flows. The code can simulate flows at the relatively low-energy scales characterized in urban dispersion studies, to the flows characterized by more violent atmospheric phenomena that are encountered in forest fires, natural and man-made disaster scenarios, and explosions.

HIGRAD uses a novel grid system that incorporates the features and advantages of both a generalized coordinates system, and a terrain following coordinate system. The model grid can "zoom in" at high-resolution over regions of interest, while using coarser resolution elsewhere in a simulation. The model is second-order accurate in space and time, and uses either a Smagorinsky type, or one-equation turbulent kinetic energy based subgrid closure. Advection of model variables is done using a non-oscillatory forward-in-time advection scheme (MPDATA) that can accurately model regions of strong shear. The model can be run in an anelastic mode, or, alternatively, it can solve the fully compressible Navier-Stokes equations. Recent improvements to the model numerics include the incorporation of a "Method of Averages" approach to the solution of the fully compressible Navier-Stokes equations, and a fully implicit time integration of the model variables using a sophisticated implicit Newton-Krylov scheme.

The model includes a physics package that allows for the consideration of radiative heating and cooling effects, including shading, and a surface energy budget. This package also allows for the input of land use information in order to more realistically account for surface radiative and thermodynamic properties.

Finally, HIGRAD is designed to run on computers using massively parallel architecture. Therefore, it can be used to simulate high resolution three-dimensional flows that were not practically possible only a few years ago.

One important aspect of such CFD modeling is the validation of the model against available data sets. We present here, a series of validation studies that compare HIGRAD model simulations to USEPA wind-tunnel data, tow-tank data, and data obtained from the Vertical Transport and Mixing Experiment (VTMX) recently conducted in Salt Lake City.