ABSTRACT
In this chapter, our objective was to determine the relationship between topographical elements (i.e., elevation, slope, and aspect) and forest fire occurrences over the nine fire-dominant natural subregions in the Canadian province of Alberta. We employed 20,588 historical fire events over the period 2001–2014 in conjunction with Shuttle Radar Topography Mission (SRTM)-derived digital elevation model. For finding the relations, we used the fire events during the period 2001–2008 and generated relative frequency distribution (also known as observed probabilities). Then, we found that both the elevation and slope at the location of fire events demonstrated distinct patterns; those modeled using log-logistic distribution for elevation and generalized extreme value (GEV) for slope probability density functions (PDFs) were demonstrated as the best fitting. However, only the GEV PDF was found to be statistically significant per the Kolmogorov–Smirnov (K-S) test (i.e., K-S statistics ~0.01127, while the critical value was 0.01275 at 95% confidence level). Thus, we applied the GEV PDF over an independent data available during the period 2009–2014 as a function of slope and found that the results were also statistically significant (i.e., K-S statistics ~0.01449, while the critical value was 0.01454 at 95% confidence level). Thus, we applied the GEV PDF as a function of the slope map to generate a slope-derived probability of forest fire occurrences at 30 m spatial resolution over the study area of interest. The generated map could be used for forest fire management purposes.
