Computing wildfire behaviour metrics from CFD simulation data

Abstract

In this article, we demonstrate a new post-processing methodology which can be used to analyse CFD wildfire simulation outputs in a model-independent manner. CFD models produce a great deal of quantitative output but require additional post-processing to calculate commonly used wildfire behaviour metrics. Such post-processing has so far been model specific. Our method takes advantage of the 3D renderings that are a common output from such models and provides a means of calculating important fire metrics such as rate of spread and flame height using image processing techniques. This approach can be applied similarly to different models and to real world fire behaviour datasets, thus providing a new framework for model validation. Furthermore, obtained information is not limited to average values over the complete domain but spatially and temporally explicit metric distributions are provided. This feature supports posterior statistical analyses, ultimately contributing to more detailed and rigorous fire behaviour studies.