Where to Next with Fire Modelling?
Excessive amounts of heat and moisture generated by large bushfires can affect the surrounding atmosphere, changing wind speed and creating pyroconvective (thunderstorm) clouds. These in turn can profoundly change the way a bushfire behaves. Researchers are starting to incorporate these vital elements into operational fire behaviour models.
“Our biggest fires influence the atmosphere and weather around them,” says Dr Simon Heemstra, Acting Group Manager Community Resilience at the New South Wales Rural Fire Service, and Lead End User Representative for the Bushfire and Natural Hazards Cooperative Research Centre (CRC).
“These are our worst fires, and yet they are the ones that are really hard to forecast what is going to happen, because we do not know how the fire and the weather interact.”
Fire modelling is a vital part of the firefighting toolkit, with significant advancements made in the past decade. The current knowledge for fire behaviour analysts has been recently published as a practical guide by AFAC and CSIRO, based on a Bushfire CRC-commissioned synthesis report. The guidebook presents the here and now; but coupled fire-atmosphere models, incorporating how bushfires influence the atmosphere, and therefore the weather, are a key pathway to the future.
This new technology and know-how is still years away from operational use, but the research is underway now through the Bushfire and Natural Hazards CRC, ensuring fire agencies across Australia will continue to use the latest knowledge to fight fires. Dr Jeff Kepert, Head of High Impact Weather at the Bureau of Meteorology (BoM), is leading the project with BoM colleague and recent Bushfire CRC PhD scholarship graduate Dr Mika Peace part of the team, along with Dr Heemstra.
“By having a coupled fire-atmosphere model, we are going to get a much more accurate forecast leading into the really bad days,” Dr Heemstra said.
The same applies after bad fire days too, if there have been big fires.
“The State Mine Fire in the Blue Mountains in October 2013 ran 35 km in one day and left a massive amount of black ground behind,” Dr Heemstra said.
That affects the weather models, as they are based on a certain amount of reflectance from the earth’s surface. Changing from green vegetation to burnt black makes a big difference.
“We found that the winds and the temperatures in the days following were not as forecast. A coupled fire-atmosphere model will give a better indication of this,” he said.
The project is using the Weather Research and Forecasting fire model, known as WRF-Fire, which was developed in the USA. Dr Peace completed her PhD through the Bushfire CRC in 2014 using WRF-Fire to investigate interactions between fire and the atmosphere leading to extreme, or unexpected, fire behaviour.
“We have learnt all sorts of things about dynamical fire-atmosphere interactions using the model – things we could not have found out about through taking observational measurements at the fireground,” Dr Peace said.
“That is the big advantage from coupled fire-atmosphere models from a research point of view.”
One of the big challenges of combining a fire model with an atmosphere model is computer power, and it has only been in recent years that the computing speeds required to merge the two have become realistic.
Naturally WRF-Fire is being configured for Australian conditions, with the first year and half of the project devoted to recreating the fire model within the BoM’s operational atmospheric model. The BoM upgraded to a new modelling system in 2009, which has brought a large improvement in weather forecast performance.
Dr Peace believes it is important to define the outcomes into something fire behaviour practitioners can use. As fire danger ratings are number based, the researchers hope to integrate particular fire phenomena into a formal, quantitative system.
“Subjective ideas from research are quite hard to integrate into the formal, numerical rating structure that we have in the current fire forecasting system,” Dr Peace said.
This includes factors like interactions with topography, potential for pyroconvection, potential for three-dimensional interactions, potential for winds to change substantially around a fire, water vapour dry slots, plume development and spotting processes.
“We understand them in a subjective sense, but how do we turn them into some sort of metric which can be incorporated into a risk measure when there actually is a fire?
“I think coupled fire-atmosphere modelling offers an avenue for doing this, because if a series of simulations are run with various thresholds, you can actually find out what the thresholds are for certain phenomena,” Dr Peace said.
“When you talk to the [fire agency] operational people, they are very receptive to these sorts of ideas, because it matches what they are seeing on the fireground. They just do not have a way of measuring it.
“But in order to fit the new knowledge into the systems we have now, we need to tie it down to a more quantitative measure. The challenge is how to do it in a sensible way.”
Dr Peace said timing is important. Currently, computer speeds do not allow coupled models to run fast enough to be useful for routine operational purposes.
“When you look at computer power and how far it has come in the last five or ten years, we are probably not that far away before the computing power catches up and we can run coupled simulations in real-time,” she said.
Dr Heemstra believes there are a number of other significant circumstances where coupled fire-atmosphere models will make a difference.
“It comes down to three key factors,” he said.
“Understanding the effects of the [smoke] plume on the local winds, how the plume might affect a wind change and the potential of the plume to generate lightning through pyroconvection.”
Dr Heemstra uses the example of how a smoke plume can affect local winds and therefore spot fire behaviour. Fire behaviour analysts will benefit greatly from a much better understanding of when a spot fire will continue with the prevailing winds fanning the main fire front, or when it is more likely to be drawn back into the main fire front, operating under different wind conditions from the prevailing winds.
“There are some really complex interactions that occur in that kind of fire behaviour and hopefully we will get a better insight,” he said.
“Spot fires can land a long way in front of the main front. At what point do they start getting drawn back in? Is it 500 metres? Is it a kilometre? Is it 4 km? Just understanding how the plume is affecting local winds is really critical.”
How the topography surrounding a fire influences wind speed and direction is also vital.
“The coupled model will also show how the winds around the fire change with the topography. This is a pretty big advantage that we do not have at the moment,” Dr Peace said.
“This will give us additional information about spotting processes and how a fire will move along ridgelines and through valleys.”
Current atmospheric models are run operationally at around 5 km resolution, but in research mode, the resolution is 200 or 400 metres. Fire models are run operationally at 10’s of metres. Models that take an hour to run in current operational use can take half a day at the increased resolution.
“There can be an order of magnitude difference with wind speeds through gullies and along ridges,” Dr Peace said.
“This difference has a big impact on how the fire propagates.”
The increase in knowledge that will come from this research will inform weather forecasts, which in turn will benefit warning messages about any fire threat.
Dr Peace and Dr Kepert are aiming to conduct case studies to assess the results of the coupled fire-atmosphere work.
“We will do a comparison with what we have done in the past, using three fires from my PhD studies, so there is a benchmark to work with,” Dr Peace said.
These case studies will cover two bushfires on South Australia’s Kangaroo Island (the D’Estrees Bay Fire and the Rocky River Fire) in December 2007, and the West Australian Layman fuel reduction burn in October 2010.
Dr Peace said the end user support for the work is very encouraging.
“It makes sense with what they see in the real world. Their response has been overwhelmingly positive.”
For more information, go to www.bnhcrc.com.au