UW Aquatic & Fishery Sciences Quantitative Seminar

Maureen Kennedy

SFR, UW and USFS Pacific Wildlands Fire Sciences Lab

Using simple models to understand complex behavior

Simulation models can be placed along a continuum of model abstraction, where the most abstract models are simpler and less mechanistic, and such models may be able to replicate aggregate properties of ecosystems. As increasing mechanism and realism are incorporated into a model, it becomes less abstract and more specific to a location, event, time, and/or ecosystem. As we move from more abstract to less abstract models, the information required to initialize, parameterize, and evaluate a model increases and becomes more detailed. It is imperative that when we use simulation models we match the level of model abstraction that we allow to the amount and quality of information available to initialize, parameterize, and evaluate the model. I show how a simple stochastic model of fire spread is able replicate the variability in the spatio-temporal fire history record of six watersheds throughout Washington state, and in this context such a model is preferred to more complex models of fire spread that are intended to replicate the spread of individual fires and fire events. From the simulation results we are able to infer that in watersheds with relatively complex topography the balance between bottom-up (e.g., topographic barriers, fuels) and top-down (climate, weather) controls on fire spread is shifted to bottom-up controls, whereas in watersheds with relatively simple topography the balance is shifted to top-down controls.