Existing fire behaviour models under-predict the rate of spread of summer fires in open jarrah (Eucalyptus marginata) forest

Summary Fire behaviour data from Project Vesta experimental fires were used to evaluate the fire behaviour relationship on the Forest Fire Danger Meter (FFDM), the Forest Fire Behaviour Tables (FFBT) for Western Australia and the fire spread model of Burrows (1999) for predicting the rate of spread of fires in jarrah (Eucalyptus marginata) forest under dry summer conditions. Experimental fires were conducted at two sites in south-western Australia under conditions of moderate to high forest fire danger with fine fuel moisture contents of 6–9% oven dry weight and wind speeds up to 20 km h−1 (at 10 m in the open). Fuels ranged in age from 2 to 22 y since the last fire and had understorey shrub layers from 0.5 to 2.0 m tall. Mean rates of spread for experimental fires were 392 m h−1 and 405 m h−1 at the low and tall shrub sites respectively, with the fastest fires spreading at close to 1200 m h−1. Fires spread two to three times faster than predicted by the FFDM and FFBT fire spread models, and up to five times faster than predicted by the spread model of Burrows. Both the FFDM and FFBT models predicted fire spread better at the low shrub fuel site and when wind speeds were <12.5 km h−1. All models under-predicted fire spread at the tall shrub site and grossly under-predicted fire spread when wind speeds were >12.5 km h−1. We argue that existing fire behaviour models under-predict because they were developed using data from fires that had not achieved a quasi-steady rate of spread, whereas the line ignition technique used during Project Vesta experiments resulted in fires travelling at a speed closer to their potential rate of spread for the prevailing conditions. Existing guides remain valid for predicting spread during the early stages of fire growth when fires are developing from a point ignition and the headfire remains narrow.