Session 11.2 Application of a Vegetation Canopy Parameterization to Wildland Fire Modeling

Thursday, 5 August 2010: 1:45 PM
Crestone Peak III & IV (Keystone Resort)
Michael T. Kiefer, Michigan State University, East Lansing, MI; and S. Zhong, R. Shadbolt, W. E. Heilman, J. J. Charney, and X. Bian

Presentation PDF (948.8 kB)

Prediction of smoke dispersion from low-intensity fires is a particularly challenging subject due to the effect on dispersion of critical factors such as near-surface meteorological conditions, local topography, vegetation, and atmospheric turbulence within and above vegetation layers. In this study, we examine the atmospheric response of a weakly stable atmosphere to a surface heat source representative of a low-intensity fire, and then evaluate the impact of a forest canopy on the response. A modified version of the Advanced Regional Prediction System (ARPS) is utilized for this study, wherein the fire is parameterized in the model by prescribing an isolated region of steady upward surface heat flux. As has been performed in other large-eddy simulation studies, a vegetation canopy is parameterized by adding a pressure and viscous drag force term to the momentum equation, and by adding a sink term to the turbulent kinetic energy (TKE) equation, to account for the more efficient dissipation of turbulent energy within the canopy.

A two-step strategy is adopted wherein first, mean and turbulent fields from simulations with a parameterized isolated forest (but no fire) are compared to previous large-eddy simulation results and observational data. In the second stage, simulations with an isolated surface heat flux (i.e., a fire) with and without a parameterized forest are compared in order to understand how drag within the forested region impacts the atmospheric convection and underlying processes. Our results are found to compare favorably to those of other large-eddy simulation studies of canopy turbulence; notable features include a strong vertical gradient of horizontal wind speed near the top of the canopy, a vertical velocity maximum near the upstream edge of the forest region, and a maximum in TKE above the forest. Results are also shown to compare favorably with data from wind tunnel and field experiments.

Results from the second stage of work illustrate the notable differences between simulations with and without a parameterized forest, but with identical fire heat flux intensity and location. It is found that the main impact of the forest is to yield much more irregular development of new convective cell development, although the magnitude of vertical velocities and the spatial scale of the convection are largely unaffected. Additionally, the heated layer of air downstream of the fire remains closer to the surface in the simulation with both fire and forest. The possible implications of these results to smoke dispersion motivate future work examining the trajectories of particles in simulations with and without forest layers. It is also found that the inclusion of a forested layer yields maximum perturbation potential temperatures about 2.5 times larger then with the fire alone. The stronger heating of surface air is the result of the reduction in turbulence kinetic energy due to the forest canopy: turbulent eddies are weaker, due to the loss of energy to highly dissipative wake turbulence, and thus are less able to transport the heated air upward away from the source. Thus, much larger potential temperature perturbations are able to develop when the fire coexists with a forest canopy. Ongoing work involves the integration of the canopy parameterization into real data simulations with the full suite of model physics parameterizations available in ARPS.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner