COS 187-2 - Exogenously constrained dynamic percolation shows a phase transition in landscape controls of low-severity fire regimes

Friday, August 10, 2012: 8:20 AM
E144, Oregon Convention Center
Maureen C. Kennedy, School of Environmental and Forest Sciences, University of Washington, Seattle, WA and Donald McKenzie, Pacific Wildland Fire Sciences Lab, US Forest Service, Seattle, WA
Background/Question/Methods

It is difficult to understand the dominant controls of historical wildfires because we cannot reliably reconstruct weather or fuel composition, yet understanding of historical fire dynamics will inform management of fire-prone landscapes into the future.  Scaling laws can be used to infer differences in the major drivers of fire spread.  We use the Sørensen’s distance as a measure of fire co-occurrence between pairs of recorder trees at six fire-history sites in the Pacific Northwest, and we see power-law scaling in the relationship between Sørensen’s distance and the Euclidean distance between pairs of recorder trees (the Sørensen variogram).  We modify a dynamic percolation model with an exogenous constraint on fire size (exogenously constrained dynamic percolation, ECDP) and use a Monte Carlo goodness-of-fit inference procedure to identify domains of model behavior able to replicate the observed patterns. 

Results/Conclusions

Simulations of ECDP show that the power-law in the Sørensen variogram occurs in a limited domain of parameter values that corresponds to the percolation threshold for dynamic percolation.  Monte Carlo comparison of simulations to observed patterns reveals that this critical domain corresponds with the fire histories at sites with the most complex topography, whereas sites with the least complex topography correspond to ECDP runs well outside the critical domain.  This threshold represents a phase transition from sites for which endogenous and exogenous controls are in balance to sites for which endogenous controls have a stronger effect.  When we perform the same analysis for a self-organized criticality model we find correspondence to fire-history sites only within this critical domain.  Self-organized criticality cannot identify a similar phase transition because it only accounts for endogenous controls on fire spread, whereas fire dynamics (and the ECDP model) are shaped by both endogenous and exogenous controls.  By quantifying the transition between endogenous and exogenous controls on landscape fire we should improve predictions of the response of fire regimes to a rapidly changing climate.