An ignition probability layer (fire density) was derived using an interpolation process (ArcInfo; pointinterp) of the fire start data. 2-km cells were used for the interpolation process because fire locations were commonly only reported to the nearest Public Land Survey Section (PLSS), approximately 1 square mile. Interpolation resulted in 2-km cells being attributed with the count of ignitions. Ignition probability was derived by classifying the density data into 5 classes (low to high) using the "natural break" algorithm included in ArcMap.
The values represent relative values that have been standardized between 0.0 and 1.0
Originally, fire occurrence point data were obtained for the period 1983 to 2002 from firefighter report forms (DI-1202 for Department of Interior agencies, 5100-29 for FS, and DL-850B from IDL) that had been compiled by Bureau of Land Management (BLM), Bureau of Indian Affairs (BIA), US Fish and Wildlife Service (USFW), National Park Service (NPS), US Forest Service (USFS), Idaho Department of Lands (IDL), and the Montana Department of Natural Resources (MDNR). This period of time was selected to promote consistency among the various federal and state agencies.
These data were problematic due to data input errors, especially with respect to location; fires with erroneous locations were discarded. Rural and volunteer fire department data were not used in this analysis because no consistent fire reporting system and data source was available. Upon review of these data, the Idaho State Fire Plan Working Group (ISFPWG) Technical Team decided to use a compilation that was conducted by the BLM National Science and Technology Center (NSTC) for fires recorded on Federal lands. The compiler of these data also reported that these data contained many errors. These data were clipped to within a 10-k buffer of the Idaho state line. Fires reported by the states of Idaho, IDL and Montana, DNRC, within 10-k buffer of Idaho were then added to the BLM NSTC compiled data.
Ignition probability was derived using all ignitions recorded from federal and state lands, both human and lightning caused. Ignition probability in this analysis describes the number of fires per 4KM2/20 years. A high rating indicates areas where relatively more fires occur per unit area.
Two methods were used to derive ignition probability: (1) using all ignitions; and (2) using only ignitions resulting in fires that exceeded 10-ac. This was to assess whether there was a difference in ignition patterns in respect to fire size. Ignition patterns seemed to be almost the inverse of each other. In general, these data suggested that northern Idaho has many fires that do not burn substantial areas. Conversely, southern Idaho has fewer fires, but substantially more area is burned. The ISFPWG technical committee decided to use the ignition probability data derived from all fires to derive ignition probability. Fire weather and fuels were evaluated independently.
Ignition Probability data were standarized between 0.0 and 1.0 prior to their use in diriving Relative Wildland Fire Risk.