Treatment of the smoke as a mixture of the chemically active species that undergo chemical and
physical processes in the atmosphere is very computationally intensive, since each additional
chemical species leads to an additional 3D scalar equation, plus the complex mathematical
representation of the chemical reactions. In order to reduce this computational cost, WRF-SFIRE
offers also a simplified smoke representation trough a passive scalar. In this case, instead of
resolving the concentration of all the spices listed in Figure 2, only one variable is added to the
WRF computations (tr_8), which is a scalar that does not react in the atmosphere. If the
simplified version of the smoke treatment is chosen, the execution of WRF-Chem is bypassed
5
and the dynamical WRF core handles the transport of the smoke directly. In this simplified
mode, the tracer flux is proportional to the fuel consumption rate.
3. Model setup
The two Santa Ana fires (2007) simulated in this study were driven by strong westerly Santa Ana
winds induced by a high-pressure system located over northern Nevada. As the pressure built up
and the high pressure system moved eastward, very strong and gusty Santa Ana winds bringing
very warm and dry air from the Nevada desert affected the San Diego area. Wildland fires are
directly driven by the local winds, often of a different characteristic than the main synoptic flow.
Regional topography and land use mosaic may interact with the large-scale flow, creating
specific local weather conditions that may be crucial for wildfire behavior. In order to resolve the
development and movement of this large-scale weather system driving the Santa Ana winds,
together with the local circulation affected strongly by the complex topography of southern
California, WRF was configured with four nested domains: D01, D02, D03, and D04, of
horizontal-grid sizes 32km, 8km, 2km, and 500m, respectively. The domain setup used in this
study is shown in Figure 3. The fire model uses 30m-resolution elevation and fuel dataset, while
the atmospheric model ~1.5km resolution MODIS land use representation.