High-resolution observations of combustion in heterogeneous surface fuels

In ecosystems with frequent surface fires, fire and fuel heterogeneity at relevant scales have been largely ignored. This could be because complete burns give an impression of homogeneity, or due to the difficulty in capturing fine-scale variation in fuel characteristics and fire behaviour. Fire movement between patches of fuel can have implications for modelling fire spread and understanding ecological effects. We collected high resolution (0.8 0.8-cm pixels) visual and thermal imaging data during fire passage over 4 4-m plots of mixed fuel beds consisting of pine litter and grass during two prescribed burns within the longleaf pine forests of Eglin Air Force Base, FL in February 2011. Fuel types were identified by passing multi-spectral digital images through a colour recognition algorithm in ‘Rabbit Rules,’ an experimental coupled fire-atmosphere fire spread model. Image fuel types were validated against field fuel types. Relationships between fuel characteristics and fire behaviour measurements at multiple resolutions (0.8 0.8 cm to 33 33 cm) were analysed using a regression tree approach. There were strong relationships between fire behaviour and fuels, especially at the 33 33-cm scale (R1⁄4 0.40–0.69), where image-to-image overlap error was reduced and fuels were well characterised. Distinct signatures were found for individual and coupled fuel types for determining fire behaviour, illustrating the importance of understanding fire-fuel heterogeneity at fine-scales. Simulating fire spread at this fine-scalemay be critical for understanding fire effects, such as understorey plant community assembly. Additional keywords: fire heterogeneity, fuel type, image recognition, IR imagery, longleaf pine, regression tree. Received 19 September 2013, accepted 28 March 2014, published online 11 September 2014