The importance of resolving small-scale processes and their impacts on large-scale smoke plume dynamics

Despite significant advancements within coupled fire-atmosphere models over the past decade, forecasting fire growth and smoke impacts on air quality remains challenging for operational numerical weather prediction (NWP) models. Wildland fires dynamically interact with the atmosphere, which can impact fire growth, fire plume rises, and downwind smoke dispersion. For example, larger scale smoke transport can be impacted by fire growth rates, which are driven by fine-scale winds near the surface and fuel moisture. Winds within forest canopies can be particularly challenging to resolve for NWP models. Typically, these winds are either (1) explicitly computed by the NWP model, (2) scaled using a logarithmic wind profile or (3) parameterized using a canopy model. Each of these methods have underlying assumptions that may not necessarily be appropriate for determining winds at the fire mid-flame height within forest canopies. To test the sensitivity of canopy winds on downwind smoke transport, a fire-atmosphere coupled model (WRF-SFIRE) was used to simulate a prescribed burn within a forested plot with simple terrain during the RxCADRE experiment. Results presented here found that WRF-SFIRE significantly overestimated fire growth rates and fire plume rise heights by a factor of 2 when extrapolating above canopy winds to the fire mid-flame height using the logarithmic wind profile assumption. When winds within WRF-SFIRE were compared to nearby wind measurements, a systematic bias of +3.9 m s^-1was found. A wind canopy parameterization specifically adopted for fires was implemented within the WRF-SFIRE that scales above canopy winds using a non-dimensional wind profile. Implementing a canopy wind model within WRF-SFIRE significantly improved sub canopy winds (bias = -0.2 m s^-1), fire growth rates and downwind smoke dispersion when compared to observations.