Modelling the dynamic behaviour of junction fires with a coupled atmosphere-fire model.

Thomas, C.M., J.J. Sharples and J.P. Evans
International Journal of Wildland Fire, 26(4), 331-344, doi: 10.1071/WF16079, 2017.


Dynamic fire behaviour involves rapid changes in fire behaviour without significant changes in ambient conditions, and can compromise firefighter and community safety. Dynamic fire behaviour cannot be captured using spatial implementations of empirical fire-spread models predicated on the assumption of an equilibrium, or quasi-steady, rate of spread. In this study, a coupled atmosphere–fire model is used to model the dynamic propagation of junction fires, i.e. when two firelines merge at an oblique angle. This involves very rapid initial rates of spread, even with no ambient wind. The simulations are in good qualitative agreement with a previous experimental study, and indicate that pyro- convective interaction between the fire and the atmosphere is the key mechanism driving the dynamic fire propagation. An examination of the vertical vorticity in the simulations, and its relationship to the fireline geometry, gives insight into this mechanism. Junction fires have been modelled previously using curvature-dependent rates of spread. In this study, however, although fireline geometry clearly influences rate of spread, no relationship is found between local fireline curvature and the simulated instantaneous local rate of spread. It is possible that such a relationship may be found at larger scales.

Key Figure

Fig. 6. Surface wind at 1 m above ground level (AGL) (as input into the fire-spread model, but thinned to 28-m resolution for plotting) and vertical component of vorticity 10 m AGL for one 458 ensemble member (a) 5 min; and (b) 10 min after ignition. The solid black line shows the location of the fireline. To indicate rate of spread, the grey lines (partially obscured) show the fireline 30 s earlier and later. The vorticity data have been smoothed to aid visualisation. The transects used in Fig. 5 are shown in panel (a).

UNSW    This page is maintaind by Jason Evans | Last updated 29 November 2013