|A method for examining patterns in mapped fire histories: identification of homogeneous fire landscapes
|Year of Publication
|Moritz, MA, Davis, FW
|Fire in California Ecosystems: Integrating Ecology, Prevention, and Management
In examining the mapped fire history of a large region, one may need to separate a study area into sub-regions that are homogeneous in terms of fire regime (i.e., to identify homogeneous "fire landscapes"). Because a fire regime is the result of complex interactions between fuel distributions, weather, and the cause and spatio-temporal patterns of ignitions, identification of fire landscapes may not be an easy task. Simplification of a fire regime (e.g., to fire frequency) or the use of surrogates (e.g., climatic regions or fuel maps) is often used, but this approach may ignore important aspects of how a fire regime manifests itself in a particular area. Inclusion of all available information, such as the fire size range, seasonality, and unusual intervals between fires, can provide a much better view of how fire landscapes differ from each other in significant ways. We propose an objective and repeatable method using variables generated from a mapped fire history, and we apply it to Los Padres National Forest in central coastal California. Fire variables were calculated on a regular grid spacing and incorporate the following: seasonality and cause of fire starts, number of times burned, longest and shortest interval between fires, and largest and smallest fires to pass over a site. Results indicate that this method captures the vast majority of variation in fire variables and their spatial pattern, providing mapped fire landscapes for use in fire planning or for further statistical analysis. Fire is the primary ecological disturbance structuring many of the world's terrestrial ecosystems, and spatio-temporal patterns of fires can provide insights into how these systems have developed and how we should manage them. In analyzing the fire history of a specific region, one is concerned that the fire regime be stationary (i.e., not containing mixed distributions) over space and time, but the scale and timing of dominant mechanisms are often poorly understood. Temporal shifts in fire regime can be caused by changes in climate or fire suppression, and different methods have been developed for dealing with mixed distributions over time (Clark 1989, Johnson and Gutsell 1994). Identification of areas that are spatially homogeneous in terms of fire history has received some attention, but many studies are performed using the spatial unit for which data were collected (e.g., at the scale of a specific county or forest). This scale may be appropriate, particularly if the goal is to characterize a process at a regional scale. Conversely, one may need to separate a study area into spatial units that are homogeneous in terms of fire history to examine the importance of local factors. A notable example of this is Baker (1989), in which homogeneous regions were sought by fitting fire-interval distributions to fire history data. Although characterization of fire frequency is a well established approach (Heinselman 1973, Johnson and Gutsell 1994, Johnson and Van Wagner 1985), the focus on fire intervals can omit important aspects of a fire regime that many mapped fire histories contain. Chou and others (1990) used a fire history to examine the distribution of fires and their spatial neighborhood effects, but the study area had been simplified to a binary variable (i.e., burned versus unburned). Inclusion of all available information, such as the fire size range, seasonality, and unusual intervals between fires, can provide a more complete view of how regions differ from each other in important ways. We propose here a methodology that is flexible, yet quantitative and repeatable, for identifying "fire landscapes" that are homogeneous in terms of several fire-related attributes. As a demonstration of its usefulness, we quantify and compare the vegetation composition of resulting fire landscapes on Los Padres National Forest (LPNF) in central coastal California to test whether analysis of fuel dependency at the scale of the entire study area is appropriate.