The fire management policy at Glacier National Park strives to allow fire to play its natural role to the maximum extent possible while protecting human life and property. Various fire models have greatly contributed to the success of the program by serving as decision aids for Park managers in selecting the appropriate management response for ongoing fires, planning prescribed fires, and analyzing effects of proposed fuels treatment activities. While such models have been successfully utilized, their application and accuracy has been hampered by limited detailed fuels information.

In this study, surface and crown fuels in Glacier National Park were analyzed and initially stratified based on an unsupervised classification of Landsat Thematic Mapper multispectral imagery. This classification generated statistical clusters that were grouped by spectral similarity for fuels inventory sampling. Over 720 fuel data collection sample plots were installed. Inventory techniques at each plot included Brown’s downed woody fuels transects, Burgan and Rothermel Fuel Inventories, and crown fuel data characteristics collection. Fuel loading and fire characteristics for each plot were computed by spreadsheet. Resulting fuel loading and fire characteristic groupings were associated with spectral groupings and a reclassification was conducted to generate a fuel model stratification based upon the best fit. These fuel models were then compared to the existing standard 13 Fire Behavior Prediction System Models and custom models were developed where significant differences existed. The stratification results are contained in a detailed database of fuel characteristics and parameters and displayed as a fuel model map. The resulting fuel model stratification will be useful in a variety of predictive models for fire behavior, fire effects, smoke modeling, and fuels treatment analysis.