|
Abstract:
|
We present new results from a research program aimed to efficiently simulate fire regimes on forested landscapes, while taking into account fire's interactions with vegetation, fire management, and changing climate. The intended application is to the fire component of integrated spatial dynamic models, or "landscape models".. Our approach is based on adding some structure to neutral percolation models of fire spread. Specifically, we make the parameters that control the probabilities of fire arrival and spread vary in space and time in response to changing landcover and annual fire weather conditions. This allows fire-model behaviour to be coupled with vegetation dynamics modes, and to climate change scenarios, or to historical fire weather data. Fire arrival is modeled as spatially inhomogeneous Poisson process; the model parameter estimates can easily be scaled to drive any grid-based simulation environment, of any spatial resolution. Fire sizes are modelled as a tapered Pareto distribution. We consider some aspects of joint modelling with correlated random effects to account for annual fire weather effects.
|
