Applying fire spread simulation over two study sites in California lessons learned and future plans

There is an increasing application for bushfire spread models in planning for prescribed burning and for the generation of fire risk assessment maps in fire prone communities. An evaluation of FARSITE and FlamMap bushfire spread models developed by the Fire Sciences Laboratory at Missoula involved a comparison of fire simulator models over two Californian landscapes representing different terrain and vegetation regimes. The paper includes a discussion on the models, the assumptions and limitations resulting from their application, and also the assembly of data to build landscape files and model outputs over the two test areas. The spatial datasets used in this study are sourced from the USGS EROS Data Center’s LANDFIRE database at 30- metre pixel resolution. Information about fuel which has been derived from satellite imagery, terrain modelling and biophysical and local field knowledge was used to build Anderson’s 13 Fire Behavior Fuel Models (FBFM13) and Scott and Burganvs 40 Fire Behavior Fuel Models (FBFM40). These were ingested into the FARSITE fire growth simulation model and FlamMap fire potential simulator. The FBFM40 provides a better representation of fuel across the landscape, leading to an improvement in surface fire intensity prediction and increased precision in determining crown fire behaviour. The FARSITE/FlamMap were used to model fire behaviour, and WindWizard simulated wind speed and direction scenarios across the Woodacre and Glen Ellen regions near San Francisco, California. FARSITE and FlamMap are two separate fire simulation models that use the same input datasets (vegetation/ground cover type, crown stand height, crown base height, crown bulk density, temperature, humidity, precipitation, slope, aspect, elevation, wind speed and direction). In this study, the actual fire perimeters were not available to compare the overestimated and underestimated fire growth perimeters/areas after and before using gridded wind data into the fire simulation. However, previous studies