Doctor of Philosophy
School of Earth and Environmental Sciences - Faculty of Science
Heemstra, Simon, Bush fire patchiness, PhD thesis, School of Earth and Environmental Sciences, University of Wollongong, 2007. http://ro.uow.edu.au/theses/1
All bush fires are patchy at some scale and to some degree. In their most obvious form, bush fire patches are unburnt islands of vegetation within a fire boundary. Bush fire patchiness is an important part of the fire regime, because it can affect ecological processes such as local extinction and recolonisation and can contribute to the post-fire resilience of populations of animals that need vegetation cover. During a fire, the survival of plants and animals can often depend on the degree of patchiness. Unburnt patches may therefore affect the post-fire dynamics of plant and animal populations. Continuity of fuel loads is a major contributor to bush fire behaviour including the intensity and rate spread of the fire. Further, the degree of patchiness is one measure of the effectiveness of a hazard-reduction fire in reducing fuel and thus modifying subsequent wildfire behaviour. An understanding of the patchiness of fires is of critical importance for landscape bush fire risk management. Quantifying patchiness is no simple task, partly because of the continuous range of scales at which patchiness occurs and partly because patches cannot always be readily detected by remote sensing techniques, especially in forested systems in which canopy scorch is not closely aligned with burning of the understorey. Moreover, remotely sensed data are expensive at the fine resolution required to detect the small scale patterns that are ecologically relevant (i.e. of the order of metres). In such systems, a method of rapidly assessing patchiness in the field is much needed. The first aim of this study was to develop a field technique for the quantification of patchiness. The second aim was to determine the effectiveness of the technique in estimating overall patchiness in a site, using a simulation model based on a range of known patch sizes and distributions. The third was to use field estimation of patchiness to investigate the environmental parameters that influence the patchiness of fires. A range of methods was assessed and transect sampling the post bush fire landscape using Global Positioning Systems (GPS) to produce a spatial model within a Geographic Information System (GIS) was found to be an accurate and simple method for assessing the patchiness of a fire. This method was time efficient and cost effective. It produced an output that provided a clear description of the overall patchiness of a site as well as the relative distribution of the sizes of the patches. The spatial model can display the distribution of proportion of areas burnt within the landscape. I tested the statistical effectiveness of this transect sampling method using spatial modelling within a GIS. I wrote a program to simulate thousands of transect samples on a range of different spatial arrangements of patches within a landscape. This simulation revealed that the minimum number of transects required to provide an estimate of the actual 'patchiness' is dependent on the type of fire and range between 5 and 10, to obtain 90% accuracy of the estimate of the patchiness. The empirical transect data from each site was used using GIS to develop a spatial model assessing the relative influences of a range of topographic and environmental variables on the patchiness of a fire. The relative influence of these variables was different for each fire. From this analysis, a map was produced from a GIS that described the spatial distribution of patchiness within a given fire. This transect data was graphed to indicate the relative distribution and abundance of unburnt patches. The studies performed indicate these methods have important applications for managing biodiversity and wildfires.