Infrared observations and numerical modelling of grassland fires in the Northern Territory, Australia

This paper reports on a small-scale pilot experiment held early in the dry season near Darwin, Australia, in which fine-scale observations of several prescribed fires were made using infrared digital video. Infrared imaging is used routinely to locate fires as infrared radiation suffers little attenuation as it propagates through the smoke that normally obscures visible imagery. However, until now, little use has been made of digital video imagery in analyzing the convective-scale structure of prescribed (or wild) fires. The advantage of digital video imagery is that the individual frames can be objectively analyzed to determine the convective motion in the plane viewed by the camera. The infrared imagery shows mostly rising plumes, much like convective clouds. The flow is highly convective, and the vertical transport of heat is confined to relatively narrow thermals. The updrafts range from a few In s(-1) to around 15 m s(-1). A numerical model is used to simulate one of the prescribed fires at very high-resolution. For the most part, the model predictions compare well to the observations. The model produces plumes that are around 7 m high, and spaced around 5 m apart, which is similar to that observed. The model correctly predicts the mean rate of spread of the fire to be 1.3 m s(-1). Perhaps the most serious limitations to using infrared observations of the type presented here are the difficulties in interpreting precisely the relationship between the observed infrared temperature field and the air temperature calculated by the model, and the exact connection between the infrared camera derived flow field and that calculated by the model.