@article {bnh-8379, title = {Fire coalescence and mass spot fire dynamics - final project report}, number = {736}, year = {2023}, month = {04/2023}, institution = {Natural Hazards Research Australia}, abstract = {

The Fire coalescence and mass spot fire dynamics project (the Spotfire Project) was one of the core research projects within the Bushfire and Natural Hazards CRC{\textquoteright}s Bushfire Predictive Services research cluster. Specifically, the Spotfire Project was focused on enhancing our understanding of the physical processes involved in spot fire development and coalescence and developing computationally efficient mathematical models that can accurately account for these the patterns of bushfire propagation associated with mass spotting and other modes of dynamic fire behaviour. Current operational models and associated simulation platforms are predicated on the assumptions that fires spread in a quasi-steady manner and that different parts of a fire line evolve independently of other parts {\textendash} both of these assumptions are manifestly untrue during mass spotting events.

The research took a multifaceted approach, which drew upon experimentation, computer simulation and mathematical modelling to develop a new dynamic modelling framework that permits faster-than-real-time simulation of fire propagation incorporating dynamic drivers. Experimentation took place at both laboratory and field scales. Laboratory experimentation was conducted in the CSIRO Pyrotron and in collaboration with Portuguese colleagues at the Centre for the Study of Forest Fires. Field scale experiments were also conducted in Portugal. The experiments provided insights into the dynamic nature of fire behaviour and provided data for calibration and validation of the models under development.

Coupled fire-atmosphere models were used to perform idealized simulations of various scenarios involving dynamic fire propagation. The model output provided detailed quantitative insights into the physical processes driving dynamic fire propagation and were used to inform development of new mathematical models. The mathematical model developed during the project is essentially a two-dimensional version of a coupled fire-atmosphere model {\textendash} it incorporates the feedback between the fire and the atmosphere but within the two-dimensional Spark fire simulation platform.

The research has yielded many important and significant insights into the behaviour of coalescing spot fires, and these insights have enhanced our understanding of the processes driving fire propagation and the way we model dynamic fire behaviours. These have in turn provided new understanding of violent pyroconvective events and extreme bushfire development.

The most significant research outcome was the development of the pyrogenic potential model, which permits world{\textquoteright}s first capability to model dynamic modes of fire propagation (e.g., vorticity-driven lateral spread) using a two-dimensional simulation framework. This means that explicitly modelling such effects in operational timeframes is now a feasible option. The research has also examined various other issues related to the spotting process and dynamic fire propagation more generally. These include: the effects of wind-terrain interaction on ember trajectories and the likely distribution of spot fires downwind of complex terrain, the influence of spot fires on the overall rate of spread of a fire, the influence of terminal velocity assumptions on ember trajectory modelling, the influence of fuel characteristics (e.g., bulk density) on spot fire development and pyroconvective feedback, the role of fine scale vorticity effects on dynamic fire behaviour, and development of simple measures of pyroconvective potential.

There is considerable utilisation potential for the research findings. The advances in fire behaviour modelling are easily incorporated within the Spark simulator platform. Given that Spark has been formally chosen by AFAC to be the new national bushfire prediction platform, the project{\textquoteright}s research findings will eventually be available to all involved in operational prediction of bushfire propagation. Moreover, the insights into dynamic fire behaviours provided by the project will form the basis for new firefighter training and education materials that will equip operational personnel with better knowledge of the full spectrum of possible fire behaviours in given scenarios. This will improve situational awareness, with benefits for firefighter safety.

The project has also developed a new mapping product that identifies regions prone to mass spotting in association with vorticity-driven lateral spread. This will be available to fire management agencies and other relevant organisations, to assist with fire behaviour prediction, with special relevance to anticipating blow-up fire events and extreme bushfire development.

Overall, the Spotfire Project has presented a new paradigm for understanding fire behaviour and has made the first significant advances towards the next generation of operational models. The insights gained from the research complements and extends existing fire behaviour knowledge in a way that enhances our ability to deal with the increasing bushfire threat into the future.

}, issn = {736}, author = {Jason J. Sharples and James Hilton and Sullivan, Andrew and Badlan, R} } @article {bnh-8343, title = {Spotfire utilisation project: implementation of the VLS filter}, number = {730}, year = {2022}, month = {05/2022}, institution = {Bushfire and Natural Hazards CRC}, address = {Melbourne}, abstract = {

The aim of this research was to produce an operational tool to assist fire agency personnel in assessing the potential for extreme bushfire development. Specifically, this work{\textquoteright}s intention was to aid in the identification of regions of the landscape that are prone to mass spotting and other dynamic fire behaviour associated with vorticity-driven lateral spread (VLS). The research was divided into three phases: 1) calibration of the slope scale to each DEM, 2) creation of the filter (first and second order) and finally refinement of the filter and, 3) develop training material and ensure filters and related documents are distributed to relevant personnel.

The first-order and second-order wind-terrain filter enables the user to identify parts of the terrain which are susceptible to VLS through its established association with steep or broken leeward facing terrain elements. This is achieved using terrain attributes such as topographic slope, aspect, and profile curvature, as well as environmental variables such as the wind speed and direction. This builds upon the research initially undertaken as part of the original Bushfire Cooperative Research Centre.

This report summarises the three phases of the project, during which time the researchers have gained familiarity and VLS mapping capability with the GIS software {\textquotedblleft}ArcGIS Pro{\textquotedblright}, which is used to create and validate the wind-terrain filters. This facilitated the two main research objectives {\textendash} to develop and then refine the first order and second order wind-terrain filters so that they may be applied using DEMs of any resolution.

Initially, this involved manual tuning of the first-order wind-terrain filter{\textquoteright}s parameters to fit a number of calibration events. This tuning was also confirmed through analyses of slope distributions over various landscape domains. These static cases were then validated with real case conditions and the filter compared to areas where VLS was known to have occurred. Scaling thresholds for the 30, 90 and 250 metre resolution DEMs were also evaluated explicitly, and a general linear model has been determined to allow estimation of parameter thresholds for more general DEM resolutions.

The second-order filter was then designed and tested in a similar manner to the first-order filter.\  The two filters were then merged into a combined operational product, which was also validated using known cases of VLS. \ Both filters are presented here - separately and as a combined product - for the operational consideration of end-users. \ End-user feedback was then evaluated to further tailor the operational product to specific operational needs, platforms, and formats. Due to travel restrictions static filters have been produced for immediate use. Development of a dynamic mapping tool is also continuing in which other factors such as the forecast wind speed and direction (and potentially other variables) can be incorporated automatically to provide heightened operational intelligence on extreme bushfire development.

}, issn = {730}, author = {Badlan, R and Jason J. Sharples} } @article {bnh-7495, title = {Fire coalescence and mass spot fire dynamics: experimentation, modelling and simulation {\textendash} annual project report 2019-2020}, number = {625}, year = {2020}, month = {11/2020}, institution = {Bushfire and Natural Hazards CRC}, address = {MELBOURNE}, abstract = {

This report outlines the progress of the Fire Coalescence and Mass Spot Fire Dynamics project, which is one of the projects within the Next Generation Fire Modelling cluster. Specifically, the report summarises progress of the first two years of the second phase of the project, which has been extended over 2018-2021.

All milestones from the 2015-2018 phase of the project have now been delivered, and the project is continuing to build upon this work in delivering important insights into the dynamics of fire behaviour and fire line interaction. Phase 2 of the experimental program is in progress, and will extend recently published work from Phase 1. The project continues to yield important and significant insights into the behaviour of coalescing fires, and these insights are enhancing our understanding of the processes driving fire propagation and the way we model dynamic fire behaviours. \ 

In particular, the research has continued to develop the pyrogenic potential model by incorporating firebrand dynamics. This has resulted in the world{\textquoteright}s first capability to model dynamic modes of fire propagation such as vorticity-driven lateral spread using a two-dimensional simulation framework with a spotting module. This means that explicitly modelling such effects in operational timeframes is now a feasible option. The research has also examined how wind-terrain interaction influences ember trajectories and the likely distribution of spot fires down wind of complex terrain.

In the past year, the project team have delivered a number of research outputs, including conference presentations and posters and journal publications. Still more publications are in preparation. The project team has also delivered on a number of key utilisation activities. These have mainly involved discussions with key end users about the prospects of the research being incorporated into education materials and training resources for firefighters and fire behaviour analysts. The project has also begun working on a dedicated utilisation project aimed at development of spot fire spatial mapping tools.

After providing some background information on the project{\textquoteright}s aims and methodology, this report provides details on the progress of the project to date. In particular, this includes:

At the time of writing, the project is on-schedule.

}, keywords = {experimentation, fire coalescence, mass spot fire dynamics, modelling, simulation}, issn = {625}, author = {Jason J. Sharples and James Hilton and Sullivan, Andrew and Badlan, R} } @article {bnh-5194, title = {How do weather and terrain contribute to firefighter entrapments in Australia?}, journal = {International Journal of Wildland Fire}, volume = {27}, year = {2018}, month = {02/2018}, pages = {85-98}, chapter = {85}, abstract = {

Adverse weather conditions and topographic influences are suspected to be responsible for most entrapments of firefighters in Australia. A lack of temporally and spatially coherent set of data however, hinders a clear understanding of the contribution of each weather type or terrain driver on these events. We investigate coronial inquiries and internal fire agencies reports across several Australian states from 1980 to 2017 and retrieve 45 entrapments. A first analysis reveals that most entrapments happen during large fires and that the number of deaths has decreased over the last few decades. Comparing the meteorological and topographical conditions of the entrapments with the conditions of a reference set of fires without entrapment, we build a linear regression model that identifies the main contributors to firefighter entrapment. A change in wind direction, which was associated with 42\% of the incidents examined, is the main factor contributing to entrapments. Interaction between strong winds and steep slopes also influences the likelihood of entrapment and suggests that dynamic fire behaviours may also play important roles. As further details of this relationship between dynamic fire propagation and firefighter entrapment is now required, the understanding of weather and terrain contribution is a first step to produce comprehensive safety guidance.

}, doi = {https://doi.org/10.1071/WF17114}, url = {https://www.publish.csiro.au/WF/WF17114}, author = {Lahaye, S and Jason J. Sharples and Stuart Matthews and Heemstra, S and Owen Price and Badlan, R} } @conference {bnh-5024, title = {Pyroconvective interactions and dynamic fire propagation}, booktitle = {AFAC18}, year = {2018}, month = {09/2018}, address = {Perth}, abstract = {

Modelling the dynamic propagation of wildfires remains a significant challenge. Pyroconvective interactions between the fire and the atmosphere, or between different parts of the fire itself, can produce distinctly non-steady modes of fire propagation that cannot be accounted for using current operational models.


While sophisticated three-dimensional models (e.g. computational fluid dynamics (CFD) models or coupled fire-atmosphere models) have been successfully applied to wildfires, their computational requirements render them impractical for operational usage.


Here we discuss a computationally efficient two-dimensional propagation model, which can accurately replicate dynamic features of fire spread that cannot be simulated using existing two-dimensional models. These features include the development of a wind-driven fire line into a parabolic shape, attraction between nearby fires and the observed closing behaviour of junction fires. The model is compared to experimental results with good agreement.
The model incorporates a simple sub-model to account for the inflow of air generated by a fire, which allows the model to run orders of magnitude faster than full physical models, while still capturing many of the essential features of dynamic fire propagation. We argue that such a model could lead to significant improvements in operational wildfire prediction.
In addition, we will highlight some recent insights in to how the geometry of a fire line and the flaming zone influences development of the pyroconvective plume above a fire. In particular, we present evidence that the geometry of the burning region can affect plume development in a way that is comparable to the effect of total energy release.

}, author = {James Hilton and Badlan, R and Sullivan, Andrew and Swedosh, W and Thomas, C. M. and Jason J. Sharples} }