Linking complex forest fuel structure and fire behaviour at fine scales

Improved firemanagement of savannas and openwoodlands requires better understanding of the fundamental connection between fuel heterogeneity, variation in fire behaviour and the influence of fire variation on vegetation feedbacks. In this study, we introduce a novel approach to predicting fire behaviour at the submetre scale, including measurements of forest understorey fuels using ground-based LIDAR (light detection and ranging) coupled with infrared thermography for recording precise fire temperatures.We used ensemble classification and regression trees to examine the relationships between fuel characteristics and fire temperature dynamics. Fire behaviour was best predicted by characterising fuelbed heterogeneity and continuity across multiple plots of similar fire intensity, where impacts from plot-to-plot variation in fuel, fire and weather did not overwhelm the effects of fuels. The individual plot-level results revealed the significance of specific fuel types (e.g. bare soil, pine leaf litter) as well as the spatial configuration of fire. This was the first known study to link the importance of fuelbed continuity and the heterogeneity associatedwith fuel types to fire behaviour at metre to submetre scales and provides the next step in understanding the complex responses of vegetation to fire behaviour. Additional keywords: fuel heterogeneity, IR imagery, LIDAR, longleaf pine, regression tree, savanna. Received 18 October 2010, accepted 21 February 2012, published online 26 July 2012