Evaluating Australian forest fire rate of spread models using VIIRS satellite observations

Abstract

Accurate prediction of head-fire rate of spread is essential to fire management decisions during wildfires, however, evaluation of existing models is limited. Acquisition of reliable rate of spread observations for model evaluation is a key challenge, since wildfires are typically rare and difficult to monitor. We applied recent advances in satellite active fire remote sensing to generate a novel set of inferred rate of spread observations. Using these observations, we evaluated four commonly used Australian forest fire behaviour models. The Project Vesta Mk1 and Mk2 models provided the best agreement with satellite observations, although these models overpredicted at lower rates of spread. Model prediction error was mostly attributed to windspeed, suggesting that wind characteristics at the fire grounds were not fully characterised under some circumstances using station or gridded observations. We suggest that ongoing advancements in satellite active fire detection provide opportunities to evaluate and develop fire behaviour models. Show more