Predicting woody fuel consumption in eucalypt forest fires in southern Australia

Consumption of woody fuel (diameter > 0.6 cm) during forest fires has important implications for fire suppression and firefighter safety, fire behaviour, ecological effects of fire and emissions of carbon and smoke. The ability to accurately predict woody fuel consumption is therefore important to support both forest and fire management decision making. However, information on woody fuel consumption in eucalypt forest fires in southern Australia is scant and the predictive capacity of existing models is unknown. Pre-fire woody fuel structure and its consumption by fire was measured in eucalypt forests at experimental/prescribed fires and high intensity wildfires as part of the Woody Fuel Consumption Project (WFCP) in south-western Western Australia and northern-central Victoria. Three additional datasets were also sourced to increase variation in forest type, fuel complex and fire characteristics. These comprised data from south-western Western Australia collected as part of Project Aquarius, the Warra Long Term Ecological Research site in Tasmania and Tumbarumba in south-eastern New South Wales. Data were combined to better understand and model the dynamics of woody fuel consumption and evaluate five existing models for the consumption of woody fuels for potential application in southern Australian eucalypt forest fires. The proportion of woody fuel consumed by forest fires was highly variable, ranging from 9 to 100%. Evaluation statistics for existing models were best for the Australian National Carbon Accounting System model which assumes 50% woody fuel consumption, however this assumption failed to capture large deviations where woody fuel consumption has been particularly high or low. CONSUME Activity and CONSUME Southern Woody models underpredicted observations while the CONSUME Western Woody model had very little bias and a better proportion of predictions within ±10% of the observed woody fuel consumption. The BURNUP model showed the greatest overall level of error when used with natural fuels; however its performance improved when applied to heavy, modified fuel loads resulting from timber harvesting operations. A generalised linear model (GLM) approach to model development identified the Forest Fire Danger Index (FFDI) as the best predictor of the proportion of woody fuel consumed. Fireline intensity also had a statistically significant, positive relationship with the proportion of woody fuel consumed. Analysis of combustion stages did not improve the ability of models to predict ignition, partial and full consumption of woody fuels. However, analysis showed that variation in fire behaviour potentially has a greater impact on woody fuel consumption, than variation in fuel characteristics