Bushfire Simulation:  PHOENIX RapidFire

Our group has a developed PHOENIX RapidFire, a model that simulates the spread of fires across the landscape.

Previously, fire predictions were performed manually – fire behaviour had to be computed and maps had to be plotted by hand.  This required the analysis of large amounts of information and could be time-consuming – a problem when land managers are looking for guidance during emergency situations.  PHOENIX can rapidly replicate this process; accounting for changes in the weather, patterns in fuel, the efforts of firefighters and the effect of varying topography.  Large fires, such as can be simulated in under four minutes on a desktop PC.

Such fire predictions can help identify the potential threat to homes and buildings and indicate the likely arrival times of fire.  Rapid predictions can help provide more timely warnings to communities, aid evacuation planning and help guide firefighting efforts.  PHOENIX is being used routinely by Victorian fire managers and is being trialled operationally in other states.

Current work is looking to improve PHOENIX predictions by looking at ways of better measuring model performance, looking at ways to understand and represent uncertainty, analysing risk at a landscape level and investigating unusual fire behaviour that occurs under extreme weather conditions.

Fire and the tall mist forests of Victoria

We’re interested in the conditions under which the iconic mountain forests of Victoria (also known as ‘tall mist forests’ become vulnerable to fire.  Most of the time these forests are very damp and lush – this means that under normal circumstances they suppress fire and limit its spread between neighbouring drier forest types.

However, in very hot dry seasons, these forests dry out and become flammable.  This can allow fires to become very large and difficult to control.  Additionally, while the mist forests can withstand periodic infrequent fire, if they are burnt multiple times in short succession by hot fires, sensitive species may be lost from a site.

We’re investigating how the flammability of these forest changes at multiple time scales; within a day, within a season and over the years as the forest ages.  A better understanding of the drivers of flammability will help managers make decisions to optimise the system to protect a range of values including human lives and property, water, carbon, biodiversity and social assets.  This is particularly import to understand in the face of variable weather and changing climatic conditions.

Vegetation fuels across the landscape

When trying to understand fire at a landscape level, it is important to understand the fuel (consisting of vegetation) that is being burnt.  Internationally, the way to do this has to been to take maps of vegetation type and then convert these to equivalent fuel values.

We are investigating a new approach to mapping fuel in the landscape – rather than use existing vegetation maps, we are using modelling to model fuel information directly.  We have records of fuel properties from across the state of Victoria, including litter depth, elevated fuel hazard and bark hazard.  In partnership with the Arthur Rylah Institute (part of Victorian the Department of Environment, Land, Water and Planning), we are using regression modelling methods to relate these fuel properties to environment conditions (such as rainfall and soil type) and measured landscape data (such as satellite imagery and historic vegetation measurements).  Ultimately we hope to improve fuel mapping by generating more accurate fuel maps for the entire state.  We are also exploring the use of other data sources, such as LIDAR, for represent fuel in the landscape.

Improved fuel maps should allow for improved fire prediction with models such as PHOENX.  Additionally they can highlight areas with high fuel hazard in the landscape to target for strategic management.  By linking fuel to environmental attributes, we also have the opportunity to get a better understanding of the ecology and fuel.  This could be important to predict changes in fuel properties in a changing climate.