Friday, August 8, 2008
Exhibit Hall CD, Midwest Airlines Center
Background/Question/Methods
Wildland fire is a widespread and important ecological process that affects and is affected by landscape heterogeneity. Fire ecologists and land managers need to understand why the propensity to burn varies among and within fire-prone landscapes. Burn probability (BP) modeling techniques combine the stochastic components of fire regimes (ignitions and weather) with sophisticated fire growth algorithms to produce high-resolution spatial predictions of the likelihood of burning. Previous applications of BP models in highly simplified artificial landscapes have isolated the importance of key environmental factors on BP patterns. We built upon this previous work to disentangle the environmental factors influencing BP patterns in a complex (i.e., real) and large (4M hectares) boreal landscape in and around Wood Buffalo National Park, Alberta, Canada. This area comprises a mosaic of coniferous (most flammable) and deciduous (less flammable) forest stands, wetlands (least flammable), and nonflammable features (lakes and exposed rock), and is dominated by large and intense crown fires that typically occur during droughts and high-wind events. A jackknife procedure was used, where BP patterns produced using the full set of variables were compared to those produced with “homogenized” individual and combinations of variables, to extract the relative contribution of the variables and their interactions.
Wildland fire is a widespread and important ecological process that affects and is affected by landscape heterogeneity. Fire ecologists and land managers need to understand why the propensity to burn varies among and within fire-prone landscapes. Burn probability (BP) modeling techniques combine the stochastic components of fire regimes (ignitions and weather) with sophisticated fire growth algorithms to produce high-resolution spatial predictions of the likelihood of burning. Previous applications of BP models in highly simplified artificial landscapes have isolated the importance of key environmental factors on BP patterns. We built upon this previous work to disentangle the environmental factors influencing BP patterns in a complex (i.e., real) and large (4M hectares) boreal landscape in and around Wood Buffalo National Park, Alberta, Canada. This area comprises a mosaic of coniferous (most flammable) and deciduous (less flammable) forest stands, wetlands (least flammable), and nonflammable features (lakes and exposed rock), and is dominated by large and intense crown fires that typically occur during droughts and high-wind events. A jackknife procedure was used, where BP patterns produced using the full set of variables were compared to those produced with “homogenized” individual and combinations of variables, to extract the relative contribution of the variables and their interactions.
Results/Conclusions
Results demonstrated that temporal and spatial ignitions patterns, types and configurations of flammable vegetation (fuels), and variability in daily fire weather conditions all contributed significantly to landscape-level BP patterning. Moreover, the strong interplay between endogenous (fuels) and exogenous (weather and ignitions) factors suggest that the boreal forest fire regime is neither ‘weather-dominated’ nor ‘fuels-dominated’. We acknowledge that our results are highly contingent on the particular landscape and fire environment under study; therefore, the next step of our research consists of undertaking the same exercise for other landscapes comparison purpose.
