Data Sources

Most spatial data used in this study came from the LANDFIRE prototype project mapping effort (Table: Definition of data inputs, Rollins and others 2006). The LANDFIRE project created spatial data layers of topography, biophysical environments, vegetation, and fuels at 30-m resolution for two map zones in the Rocky Mountains (map zones 16 and 19). All layers were available at the www.landfire.gov Web site.

The fuels layers used in this study included two surface fuel classifications: (1) the 13 fire behavior fuel models (FBFM) of Albini (1976), defined by Anderson (1982), and mapped using methods described by Keane and others (1998, 2000, 2007); and (2) the default fuel characterization classes defined in the Fuel Characterization Classification System (FCCS) described by Sandberg and others (2001) (http://www.fs.fed.us/pnw/fera) and mapped using methods described by Keane and others (2007). The FBFMs, which do not represent actual surface fuels, provided an indication of the expected surface fire behavior whereas the FCCS classes indicated the characteristics of the actual surface fuelbed, information useful for fire effects simulation. In the next update of our fire danger model, we will incorporate the expanded set of 40 recently derived fire behavior fuel models of Scott and Burgan (2005). Note that when we refer to “fire behavior” we are referring to the physical characteristics of the combustion process (Rothermel 1972). When we refer to "fire effects" we are referring to the direct and indirect consequences of the combustion process (DeBano and others 1998).

The canopy fuels layers used were the LANDFIRE canopy bulk density and canopy base-height layers. Canopy bulk density (CBD) represents the mass of available canopy fuel per unit volume of canopy in a stand (Scott and Reinhardt 2002), and it is defined as the dry weight of available canopy fuel per unit volume of the canopy including the spaces between the tree crowns (Scott and Reinhardt 2001). Canopy base height (CBH) represents the level above the ground at which there is enough aerial fuel to carry the fire into the canopy, and it is defined as the height from the ground to the bottom of the live canopy (Scott and Reinhardt 2001) but may also include dense, dead crown material that can carry a fire. These two map layers were developed for the forested lands of map zone 16 using a predictive landscape modeling approach that integrated remotely sensed data, biophysical gradients, and field reference data (Keane and others 2007). The canopy fuel characteristics were calculated for numerous plots distributed throughout the map zone using the FUELCALC model (Scott and Reinhardt 2001), and each plot was described from a set of predictor variables computed and mapped specifically for the LANDFIRE project. The predictor variables were related to CBD and CBH using a classification and regression tree (CART) approach.

Fire behavior was simulated with these surface and canopy fuels layers assuming 90^th percentile weather conditions using the FIREHARM (Keane and others 2004) program to estimate surface fire spread rate, flame length, and fireline intensity based on the Rothermel (1972) fire spread model and crown fire intensity and spread based on the Rothermel (1991) and the Scott (1999) crown fire algorithms. FIREHARM is a computer program that calculates four fire behavior variables (fireline intensity, spread rate, flame length, crown fire potential), five fire danger variables (spread component, burning index, energy release component, Keetch-Byram drought index (Burgan 1993), ignition component), and five fire effects variables (smoke emissions, fuel consumption, soil heating, tree mortality, scorch height) for each day across an 18-year climate record (6,574 days), and for every polygon in a user-specified landscape. Daily values across the 18-year period can be used to estimate probabilities that fire behavior, fire danger, or fire effects variables may exceed important thresholds. These probabilities can be mapped onto the landscape in a GIS, and maps can be used to prioritize, plan, and implement fuel or fire treatments.

In addition, LANDFIRE provided a fire regime condition class (FRCC) digital map created by simulating historical landscape conditions and comparing these simulations with current vegetation conditions derived from satellite images. FRCC is an ordinal index with three categories that describe how far the current landscape has departed from presettlement-era conditions (Hann 2004) (see www.frcc.gov for complete details).

Several other data layers were used to derive ignition risk. Relative plant greenness was estimated from an AVHRR image from June 1, 2004 (Burgan and Hartford 1993). These data were obtained from the USDA Forest Service, Rocky Mountain Research Station, Missoula Fire Sciences Laboratory. The effects of long-term drought were estimated from Palmer Drought Severity Index data obtained from the National Climate Data Center (Table: Definition of data inputs). Available PDSI data represented a span of 20 years (1971-1990), and data were derived from a 2.5-degree continental scale grid of PDSI reconstructed by Cook and others (2004). Lightning strike data were obtained from the National Lightning Detection Network (Table: Definition of data inputs). Data made available for map zone 16 will ultimately be available for all 66 map zones of the continental United States. Map zones of the Western United States to eastern Montana and New Mexico (map zones 1-24, 28) were anticipated to be completed in 2006 and the entire continental United States by the end of 2008 (http://www.landfire.gov/schedule_map.php).