Smoke modeling in a coupled fire-atmosphere framework

A wide suite of tools exists that can be used to help assess smoke dispersion. They range from simple Gaussian smoke models which aim to asses the area affected by smoke based on fuel type, fire area and wind conditions, to complex multi-model systems predicting the emissions, dispersion and air quality effects associated with wildland fires. The latter ones generally use a set of specialized sub-models, designed to attack specific tasks associated with the fire emission forecasting. The multi-model systems typically consist of a fuel consumption model, providing an estimate of the amount of burnt fuel, an emission model computing fluxes of chemical species, a plume rise model assessing the injection height, and a chemical and transport model computing how the species react and disperse in the atmosphere. The fire emission and dispersion may be linked to meteorological conditions, by feeding these models with weather data provided by a separate numerical weather prediction model. There are two major limitations of this approach. First, this modular approach generally does not allow for capturing the two-way interactions between the system components. For instance, even if the fire spread and fuel consumption are computed based on the simulated weather forecast, the fire itself is not represented in the weather model, so the weather prediction used for fire progression may not represent well the local, fire-affected weather conditions. As a consequence, the inaccuracies in the prediction of the local weather conditions may adversely affect estimates of the plume height and dispersion which both heavily depend on the weather input. The second drawback is the limited fidelity of such a system in terms of realistic representation of the fire progression, fire emissions, plume rise and plume dispersion. For example, a typical plume rise model assumes an existence of just one ideal plume (in most cases just rising vertically) which is placed in the center of the fire, while in reality, the emission pattern is much more complex. The regions of high fire activity may induce strong updraft cores, ingesting fire emissions much higher than its estimated by the plume rise models, while strong winds may limit the vertical plume to much lower elevation than predicted by an idealized plume rise model. In this paper we present a new integrated approach to the problem of smoke simulations. We show an integrated modeling framework, which renders the fire progression together taking into account the fire-weather and fuel-weather feedbacks, explicitly resolves the fire-induced convective plumes, as well as simulates transport and dispersion of the fire-emitted species in the atmosphere. We attempt model validation using MISR plume height as well as the surface particulate matter observations.