Coupled fire-atmosphere simulations of five Black Summer fires using the ACCESS-Fire model | Natural Hazards Research Australia

Coupled fire-atmosphere simulations of five Black Summer fires using the ACCESS-Fire model

Black Summer final report

Research theme

Learning from disasters

Publication type


Published date


Author Mika Peace , Barry Hanstrum , Jesse Greenslade , Dragana Zovko-Rajak , Abhik Santra , Jeff Kepert , Paul Fox-Hughes , Harvey Ye , Tasfia Shermin , Jeffrey Jones


This research builds on previous work completed during the Coupled fire-atmosphere modelling ACCESS-Fire project, by conducting case studies of five major bushfires that occurred across Australia during the 2019-20 summer. The previous work provided valuable insights into the drivers of periods of extreme fire behaviour associated with the Waroona fire in WA, and the Sir Ivan fire in NSW.

The main objectives of this project are: to investigate the meteorological drivers of extreme fire behaviour at five further fire events; to further test the reliability and stability of the model in different environments across Australia; and to support and share learnings from the events of the 2019-20 summer with partner agencies.

The fire events nominated by state jurisdictions for investigation were: Badja Forest (New South Wales), Stanthorpe (Queensland), Green Valley Talmalmo/Corryong (NSW/Victoria), Kangaroo Island (South Australia); and Yanchep (Western Australia).

The case studies show that the local context for each fire is important. Each of the fires posed significant challenges in different ways. In the heavily forested areas of NSW and Victoria, the sheer size, intensity, and duration of the fires were distinctive. At other fires, the proximity of local communities and infrastructure posed serious challenges. All fires were active in the overnight period when conventional fire danger indices typically suggest a decrease in fire activity.

The marine boundary layer, local coastal effects and topography combined to generate complex wind flows that influenced the progression of the Kangaroo Island fire. We also studied the way in which the sea breeze circulation interacted with the Yanchep fire, and examined the impact of a prescribed burn on the rate and direction of spread of the Stanthorpe fire under extreme fire weather conditions.

This project has explored gaps in knowledge and understanding of the processes at play during these unusual events by using the Bureau’s ACCESS-Fire model to examine the local three-dimensional interactions between the fires and the atmosphere. The learnings from this project may be applied to shape future fire weather products and services.


Based on discussions with our partners in fire agencies, together with documents (including photographs and video) and information on fire spread, fuel types and fuel moisture and atypical fire behaviour, we have run a series of simulations and prepared case study chapters for each of the five fires.

The ACCESS-Fire model was initialised using information provided by fire agencies on ignition location or fire line perimeter and fuel type and availability. Several simulations were conducted for each of the fires, with a focus on different time periods of interest and using updated fire polygon boundaries based on available line scan imagery.

The model configuration does not permit the simulations to account for suppression activities by fire agencies, changes in fuel types on the urban interface, or the impact of roads or other fire breaks in reducing the rate of spread of the fire. Another limitation is that the simulations are a single realisation, when in reality, consideration of forecast uncertainty (for example through running an ensemble scheme) would capture a range of potential outcomes. 

Learnings from the case studies reinforce and refine some previous learnings from historical fire events and uncover some new insights into the processes driving the periods of unusual fire behaviour that were observed during the 2019/20 season.

Key findings

The case study analyses have provided valuable insights into meteorological aspects of observed extreme fire behaviour during the 2019/20 summer. Key findings for each of the five case studies are included in the individual chapters for each fire and a synthesis is provided here. 

The ACCESS-Fire simulations run at 300 m spatial resolution show that some of the most dangerous effects of the fires, in particular damaging and destructive winds, are fire driven and therefore very localised at a much finer scale than current fire weather products.

The depth, elevation and structure of low-level jets over topography was a critical driver of overnight fire activity.  Given that these jets are driven by larger scale processes (typically synoptic scale), there is an opportunity to develop new tools based on numerical weather prediction output that highlight the location and intensity of these features. Such tools would provide additional information for meteorologists and FBANs on the strength and height of representative winds for input to two-dimensional fire spread models and could be used to alert fire managers to the risk of atypical fire spread in the overnight period.

A key theme that emerged from the case studies and discussions with stakeholders was the consistent occurrence of heatwaves and their influence on boundary layer structure overnight and on inhibiting fuel moisture recovery during the overnight period. Further research quantifying the effects of overnight heatwaves is required to appropriately plan for the increased risk to communities and fire crews in a changing climate, as the frequency of elevated overnight temperatures has increased over the last few decades and is projected to increase further (along with other factors influencing fire activity).  

Deep moist pyroconvection producing pyrocumulus (pyroCu) or pyrocumulonimbus (pyroCb) clouds was a feature of the 2019-20 fire season and a record number of sustained outbreaks of pyroCbs occurred.  However, the five fires examined here were not all associated with pyroCb, highlighting that it is not the sole weather phenomenon associated with extreme fire behaviour. In particular, no lightning was detected within the Green Valley Talmalmo fire (that subsequently became the Corryong fire when it burned into Victoria) as it burned through Green Valley, which shows that dangerous tornado-strength fire-generated winds can occur without pyroCb.

The simulations produced fire-generated extreme winds associated with both rotation and straight-line flow, which present a serious hazard. ACCESS-Fire simulated rotating winds in the updraft of the Green Valley Talmalmo/Corryong fire. More work is required to understand the mechanisms for these, they are likely related to the high intensity of the fire, coupled with topographic influences.

Spot fires were an important component in the observed rate of spread, particularly in the forest fires overnight. The extremely dry, drought-affected fuel beds meant that the ignition efficiency of spot fires was enhanced compared to an average season. Inclusion of a spotting parameterisation in ACCESS-Fire would accelerate the forwards fire spread and improve validation of the simulated fire perimeters. 

Coastal processes were evident in the simulations of the Kangaroo Island and Yanchep fires. Localised wind effects in the simulations showed interactions between the fire and maritime airmass including sea breeze circulations can modify the updrafts associated with the fire as well as produce rapid transitions in fire intensity around a fire perimeter. Cooler moister air associated with a maritime airmass may not necessarily reduce fire intensity, depending on its depth and penetration inland.

Utilisation - where to from here?

The high level of engagement in this work from a range of stakeholders has demonstrated the value of this research and the appetite for operational application. 

The learnings from this work should be incorporated into training packages tailored for fire meteorologists and FBANs that describe characteristics of the local meteorological environment favourable for extreme fire behaviour. Such training will complement other current research and training on fire behaviour aspects such as spotting and potential for pyroCb.

This work, which has been conducted in close collaboration with partners in fire agencies, has demonstrated that a comprehensive understanding of the mechanisms driving fire behaviour requires a multidisciplinary approach. Similarly, the successful application of fire behaviour and meteorology knowledge in operations requires locally connected specialised expertise, combined with tailored modelling tools to inform objective, evidence-based decisions.

The insights gathered from these five case studies have further demonstrated the value of this high resolution coupled modelling approach. In order to progress ACCESS-Fire to an effective operational tool, more robust testing of the model is required, in addition to the more subjective assessments that have been made through the case study work. This would include technical testing, additional sensitivity testing and routine verification.

This project has demonstrated that the Bureau’s technology and simulation capability, scientific expertise and established relationships with fire agencies can meet the Australian community’s need to understand the drivers of fire behaviour during the 2019-20 fire season. In doing so, we have further developed and validated our modelling systems and made progress towards their future operational use.

Year of Publication
Date Published
Bushfire and Natural Hazards CRC
Report Number
Locators Google Scholar

Related projects

Modelling fire weather interactions using the ACCESS-Fire model