Large eddy simulation of canopy-structure effects on smoke dispersion from low-burning prescribed fires

We presents results from high-spatial resolution modeling of air quality and smoke dispersion from a low-intensity prescribed burns, simulating typical burn conditions in the New Jersey Pine Barrens. We used the Regional-Atmospheric-Modeling-System (RAMS)-Based Large Eddy Simulation (RAFLES) model to simulate the effects of canopy structure on within-canopy and above-canopy smoke dispersion. The RAFLES model incorporates 3-D heterogeneous effects of tree canopies on wind flow and turbulence in the atmospheric boundary layer. It runs at high spatial resolution (5x5x3 m3) over a domain of 1.5x1.5x1.5 km3. It includes dispersion of prescribed scalars. The shaved grid cell method is used to represent physical obstructions to the flow from tree stems. Model parameters include leaf density, tree stems, the vertical distribution of leaf density, and the horizontal differences between individual tree crowns. The detailed 3D canopy is generated by the Virtual-Canopy Generator (V-CaGe), which generates canopies based on remote sensing and ground observations, and species-specific allometric equations. For the present study we used an airborne lidar dataset of the simulated forest to develop the canopy geometry in the simulation domain. An intensive observation campaign provided atmospheric, canopy, fire and smoke data to parameterize the forcing of our simulations. Smoke emission and heat were prescribed as a dynamic spatially heterogeneous forcing, which was derived from observed fire behavior and aerosol concentrations. Fire spread and emission rates were based on empirical equations based on fire line intensity and ignition time. Initial ignition location and time were prescribed and the simulated fire then spread to neighboring grid-cells as a function of fire intensity, wind speed and direction and the neighboring fuel density. Virtual experiments with the same forcing but different (e.g., sparser or denser) canopy structures determined the interactions between canopy structure, ejection height, and near-source concentrations of the smoke plumes. We tested the effects of canopy structure by simulating a virtual homogeneous canopy and comparing the resulting plume ejection height and concentration dynamics with simulations of a realistically heterogeneous canopy. The fuel distribution, which control the fire spread rate and emission strength was either homogeneous, or heterogeneous and positively or negatively correlated with the canopy leaf area index. This simulated cases where fuel accumulation is larger in gaps due to grasses (negative correlation with canopy) and cases where fuel accumulation is due to leaf litter under trees (positive correlation with canopy). These simulations tested the effects of the interaction between the spatial structures of fuel and canopy on the plume dynamics. Our results indicate that canopy shape and fuel distribution both affect the rate and ejection of smoke plumes from low-burning prescribed fires.