Coupled atmosphere-fire behavior model sensitivity to spatial fuels characterization

Recent improvements in the spatial characterization of fuels for localized regions from remote sensing offer the prospect of improving the accuracy of fire behavior models in simulating real wildfire events. This is especially true in the emerging era of coupled weather-fire models, which are run at very high spatial resolution and need to represent the fuel as accurately as possible.

In this paper we use three independent sets of spatial fuels data to compare the fire behavior generated for the Corral Canyon component of the Calabasas fire that occurred on 22 Oct 1996 in the Santa Monica Mountains of Los Angeles County. The fire simulation area encompasses a canyon with a narrow riparian zone, resistant to burning, and steep slopes covered with a variety of hard and soft chaparral species and senesced grasses. The fire simulation area is approximately 1.5 km by 1.5 km. The fuels data bases are 1) pure chaparral - as would be inferred from 1 km by 1km resolution AVHRR data; 2) a data set derived for the Santa Monica Mountains by the National Park Service, with representation of fuel variations by polygons; and 3) a 20 m raster fuels data base derived from the AVIRIS hyperspectral infrared sensor.

The HIGRAD/FIRETEC coupled atmosphere-fire modeling system is used to test the sensitivity of fire behavior to these different fuel representations. In particular , we are concerned with the sensitivity of the fire spread rate up the steep Malibu Bowl, where a firefighter was severely burned in the actual event, and with the overall fire intensity and spatial coherence using the different fuels data bases. As a result of this study, we hope to develop a better understanding of fuel specification requirements for coupled atmosphere-fire models.