The Influence of aspect and slope on the spatial distribution of bushfire

It is commonly assumed that northwest aspects are driest and most prone to bushfires and that steeper slopes bring higher fire risks. This study examined bushfires in the Sydney region to determine whether northwest slopes are more prone to fire and whether they experience more severe burning, and to analyse the relationship between slope and burn severity. Bushfire data were obtained from multiple sources. Remotely sensed imagery was used to derive fire severity using ‘before’ and ‘after’ NDVI data and a difference method using TM5/TM4 ratios. Severity classes were created using an ISODATA classification to correspond to: no burn, and low, medium, high and extreme intensity. Aspect and slope data were derived from DTM datasets at 25-metre resolution. Aspect was grouped into eight classes and a “northwest score” calculated using a sinusoidal function. The aspect classes associated with unburnt and low to extreme burn severity were compared using a Chi-squared test. An ANOVA analysis was used to test the trends in mean slope values associated with low to extreme fire severity. Relationships between environmental data are influenced by resolution and 25-metre resolution may introduce variations that are smaller than a fire’s operational scale, therefore analyses were repeated with different levels of aggregation. The study found statistically significant differences between the aspect of fire affected areas and those of the fire environs, but no aspect was preferred in all fires. The influence of aspect on severity was not simple: southerly aspects were associated with low intensity fires, high intensity fires were associated with northwest aspects, and extreme intensity burning was found on all aspects. There was a strong inverse relationship between slope and burn severity for all fires. The aggregated data did not reveal new or altered relationships; generally these analyses strengthened the statistical significance of findings with un-aggregated data.