Short-term fire front spread prediction using inverse modelling and airborne infrared images

Abstract

A wildfire forecasting tool capable of estimating the fire perimeter position sufficiently in advance of the actual fire arrival will assist firefighting operations and optimise available resources. However, owing to limited knowledge of fire event characteristics (e.g. fuel distribution and characteristics, weather variability) and the short time available to deliver a forecast, most of the current models only provide a rough approximation of the forthcoming fire positions and dynamics. The problem can be tackled by coupling data assimilation and inverse modelling techniques. We present an inverse modelling-based algorithm that uses infrared airborne images to forecast short-term wildfire dynamics with a positive lead time. The algorithm is applied to two real-scale mallee-heath shrubland fire experiments, of 9 and 25 ha, successfully forecasting the fire perimeter shape and position in the short term. Forecast dependency on the assimilation windows is explored to prepare the system to meet real scenario constraints. It is envisaged the system will be applied at larger time and space scales.