Using high-resolution weather data to predict fire spread using the FARSITE simulator—a case study in California chaparral

In October 2006, a wildfire burned 16137 hectares near Riverside, CA. The dominant fuel types within the fire perimeter were desert scrub, chamise chaparral, and interior sage scrub. High-resolution thermal imagery of the fire were collected by the aircraft-based FireMapper® within 12 hours of ignition providing a unique opportunity to compare predicted fire perimeters with actual fire perimeters in chaparral. Fire spread predictions were made with the FARSITE simulator using several different types of weather information. Observed weather data from a remote automated weather station and from an air quality monitoring station provided the lowest resolution weather data. Predicted weather data from the UC Santa Barbara MM5 (4 km resolution) model and the Regional Spectral Model (1 km) were used with FARSITE version 3 which supports gridded temperature, relative humidity, wind, and precipitation data and version 4 which only supports gridded wind data. Wind Wizard, a computational fluid dynamics (CFD) tool which models wind using CFD methods, provided the highest resolution wind data (100 m). While this fire was actively suppressed almost from the time of ignition, suppression effectiveness was limited due to the Santa Ana winds and the fact that suppression by aircraft was not possible until sunrise several hours after the 0100 ignition. Results of comparing the predicted fire perimeter locations produced by FARSITE with actual fire location determined by FireMapper will be presented.