HR: 14:40h
AN: B43E-05    [Abstracts]
TI: Scaling Fire Regimes in Space and Time.
AU: * Falk, D A
EM: dafalk@u.arizona.edu
AF: Laboratory of Tree-Ring Research, University of Arizona, Tucson, AZ 85721 United States
AB: Spatial and temporal variability are important properties of the forest fire regimes of coniferous forests of southwestern North America. We use a variety of analytical techniques to examine scaling in a surface fire regime in the Jemez Mountains of northern New Mexico, USA, based on an original data set collected from Monument Canyon Research Natural Area (MCN). Spatio-temporal scale dependence in the fire regime can be analyzed quantitatively using statistical descriptors of the fire regime, such as fire frequency and mean fire interval. We describe a theory of the event-area (EA) relationship, an extension of the species-area relationship for events distributed in space and time; the interval-area (IA) relationship, is a related form for fire intervals. We use the EA and IA to demonstrate scale dependence in the MCN fire regime. The slope and intercept of these functions are influenced by fire size, frequency, and spatial distribution, and thus are potentially useful metrics of spatio-temporal synchrony of events in the paleofire record. Second, we outline a theory of fire interval probability, working from first principles in fire ecology and statistics. Fires are conditional events resulting from the interaction of multiple contingent factors that must be satisfied for an event to occur. Outcomes of this kind represent a multiplicative process for which a lognormal model is the limiting distribution. We examine the application of this framework to two probability models, the Weibull and lognormal distributions, which can be used to characterize the distribution of fire intervals over time. Lastly, we present a general model for the collector's curve, with application to the theory and effects of sample size in fire history. Sources of uncertainty in fire history can be partitioned into an error typology; analytical methods used in fire history (particularly the formation of composite fire records) are designed to minimize certain types of error in inference. We describe a theory of the collector's curve based on accumulation of sets of discrete events and the serial probability of recording a fire as a function of sample size. Using the Monument Canyon data set, we develop a nonlinear regression method to correct for differences in sample size among composite fire records. All measures of the fire regime in the MCN fire record reflected sensitivity to sample size, but these differences can be corrected at least in part by applying the regression correction, which can increase confidence in quantitative estimates of the fire regime.
DE: 9350 North America
DE: 1615 Biogeochemical processes (4805)
DE: 0400 Biogeosciences
SC: Biogeosciences [B]
MN: 2004 AGU Fall Meeting