Probabilistic Risk Models for Multiple Disturbances: An Example of Bark Beetles and Wildfire

Haiganoush K. Preisler, Alan A. Ager, Charles  McHugh, Jane L. Hayes, and David R. Brillinger

USDA Forest Service Pacific Southwest Research Station (1), Pacific Northwest Research Station (2,4), and Rocky Mountain Research Station (3), and University of California Statistics Department (5)

Building probabilistic risk models for highly stochastic forest disturbances like wildfire and forest insect epidemics is a challenging problem.  Even more difficult is modeling the synergistic relationships that often exist among disturbances.  In the case of bark beetles and wildfire, of interest is the conditional probabilities of a fire given a beetle outbreak in some prior year, and the probability of of post-fire beetle outbreaks following wildfire. This study used a probabilistic model of wildfire occurrence built by the authors (HKP, DRB) and considers the additional influence of prior bark beetle infestations.  The study used historic data (1980 – 2005) on fire occurrence and bark beetle outbreaks collected in Oregon and Washington, USA.  Spatial data on bark beetle activity was obtained from aerial sketch maps created by the Forest Service Forest Pest Management program.  Historical federal wildland fire occurrence data were obtained from the National Interagency Fire Management Information Database (NIFMID) which included information on the date, location, and size of the fire in addition to fire weather and drought indices The methods employed provide an objective tool for modeling complex hybrid processes generally, and for estimating associated probabilities specifically.

Fire Session - Thursday Afternoon

corresponding author:

Alan Ager
USDA Forest Service
Pacific Northwest Research Station
1401 Gekeler Lane
La Grande, OR 97850-3368
541-278-3740
aager@fs.fed.us