Abstract: |
Economic analyses of wildfire management activities require good information about the characteristics of wildfire regimes over adequate temporal and spatial scales. The primary statistical models used to describe wildfire regimes are reviewed (Pareto, truncated Pareto, lognormal, negative exponential, extreme value) and the salient characteristics are described. The literature review consistently indicates that, in most cases, a few catastrophic fires account for most of the area burned. Attention is focused on two modeling approaches that we have found to be useful in characterizing catastrophic fires: negative exponential distributions and extreme value distributions. Data were collected for 20 years of wildfires on private lands in Florida, 30 years of wildfires on public lands in the West, and nearly a century of wildfires on public lands in California. Statistical methods used to estimate the parameters of these models are described, and hypotheses are tested linking model parameters with climatic and management variables. Management and policy implications of modeling extreme wildfire regimes are discussed. |
