Skip to main content
USGS - science for a changing world

Modeling Large-scale Fire Effects: Concepts and Applications


Publication Date
Time Period


Climatic changes anticipated for the next centuryare expected to alter the effects of fire onlarge-scale vegetation patterns. It is unlikelythat future interactions between fire andvegetation can be predicted from knowledge ofcurrent and historic patterns. Thus, there is aneed for simulation models that will producerealistic large-scale projections. Three topicswere addressed in this paper: 1) the difficultiesin applying current fine-scale models acrosscoarse scales, 2) qualitative modeling atcontinental scales, and 3) semi-qualitativemodeling at regional scales.A review of extrapolation problems revealed that avariety of methods have been developed by modelers;each has its advantages and disadvantages. Acontinental-scale model of [...]


Attached Files

Click on title to download individual files attached to this item or download all files listed below as a compressed file.

Original FGDC Metadata

47.07 KB


Because climatic changes are expected to occurmuch more rapidly than the historical average(Schneider 1989), it is unlikely that futureinteractions between fire and vegetation can bepredicted from knowledge of current and historicalpatterns (Brubaker 1988, Payette et al. 1989).Thus, there is a need for simulation models thatwill produce realistic large-scale projections toassist resource managers, and society at large, inunderstanding the long-term effects of currentpolicies and decisions. Fire scientists haveconcluded that the "ideal" simulation model offire effects on vegetation would be (Schmoldt etal. 1998):process-based, rather than statistically-basedspatially explicitapplicable over broad and fine scalesmodular, integrating fire behavior, fire effects, andsuccessioncapable of incorporating climatic variabilityIn most cases, practical andtheoretical considerations preclude the attainmentof this ideal. For example, no currentmechanistic model of fire behavior (e.g. Finney1995) is applicable over broad geographic areas,primarily because the requisite data forinitialization are lacking. Similarly, there aretheoretical issues involved in applying modelsbuilt at fine scales across spatiallyheterogeneous landscapes (King et al. 1991,Rastetter et al. 1992). Decisions will continue tobe made based on existing information andtechniques, however, and "compromise" models willbe used that are subject to different types oferror magnification when they are applied at broadscales. Understanding the limitations of thesemodels is a key to assessing the reliability oftheir predictions and to estimating confidenceintervals around them.There are three keyquestions whose answers can guide both thedevelopment of broad-scale models of fire effectson vegetation and an accurate assessment of theirlimitations:1. What are the difficulties/dangersin applying current fine-scale models acrosscoarse scales? How can the appropriate level ofaggregation of data and model algorithms bedetermined?2. Can methods be developed formodeling at continental scales? Are theynecessarily qualitative or subjective?3. What are the broadest scales at which quantitative methodscan be used effectively? Can they be integratedwith qualitative methods to improve predictions?In the following chapters, I address each of thesequestions in turn. Chapter One examines currentstrategies for "scaling up" fire-effects models.I develop a classification of existing strategies,suggest potential sources of error in each, andpropose an adaptive strategy for applying modelsto meet specific objectives. Chapter Twodescribes a continental-scale, rule-based modelfor predicting transitions between vegetationtypes in response to altered fire frequency. Iuse the fire-effects literature as a source ofexpert knowledge for building the model. ChapterThree describes a regional-scale model to predictfire frequency in forested areas of the InteriorColumbia River Basin. I use disparate databases,developed at different scales, to develop a modelthat integrates statistical methods and expertknowledge. I conclude by evaluating how thepartial answers given to these questions can helpto refine existing models and to suggest futuredirections for modeling the interactions of fireand vegetation.

Item Actions

View Item as ...

Save Item as ...