Published works

In some operational fire prediction models such as the McArthur model [1] or Vesta model [2], a static (constant) wind reduction factor (WRF) is used to account for subcanopy wind profile. However, the WRF is far more complicated than what can be described by a constant value. A dynamic WRF depends on wind velocity well above the forest canopy, canopy height and canopy density which may vary from location to location. In our earlier work, we developed a simpler model considering uniform vertical canopy density, known as the Harman-Finnigan model [3].  However, Moon et al. [4] and Sutherland et al. [5] demonstrated that vertically heterogonous leaf area density (LAD) can significantly change the WRF. Massman et al [6] presented a set of equations accounting for vertically heterogonous LAD,  plant area index (PAI) and above-canopy wind profiles towards calculating subcanopy wind adjustment factor (WAF). WAF is basically the inverse of WRF. In this study, we implement this LAD based WRF (by inversing Massman’s WAF) model in CSIRO’s Spark platform within its Vesta fire propagation model [2].

This research was initially tested using a synthetic dataset to apply canopy parameters that represent Australian vegetation properties. The preliminary Spark simulation results show some variation in fire rate of spread based on various vegetation types. Further, we assess the implementation using three case studies based on three fire incidents, Mount Cooke Fire 2003 of Western Australia, Kilmore Fire 2009 of Victoria, and Lithgow Fire (State Mine) 2013 of New South Wales. Taking PAI as half of leaf area index (LAI), testing four different LAD configurations, it is found that for Tall Open Eucalyptus LAD, best results are obtained in relation to actual final fire perimeters