ANU researchers are using algorithms, drones and satellites to detect bushfires before they become natural disasters.
The flooding rains of La Niña may have lowered Australia’s bushfire risk in recent times, but the devastation caused by the 2019-20 Black Summer remains seared in the country’s collective memory.
This week marks the 20th anniversary of the 2003 Canberra firestorm, which killed four people, injured hundreds and destroyed 500 homes and 23 government and commercial buildings in a single catastrophic day. Almost 70 per cent of the ACT was burned, including The Australian National University’s (ANU) internationally-renowned Mount Stromlo Observatory.
But from the ashes of disaster innovations emerge, and ANU is now fertile ground for the growth of hi-tech solutions to stop bushfires from evolving into natural disasters.
Spotting and attacking blazes early are instrumental parts of reducing the risk of large-scale fires, Associate Professor Marta Yebra, Director of the ANU-Optus Bushfire Research Centre of Excellence, says.
The $6-million centre is creating what Yebra refers to as an integrated “layer cake” approach that spans technologies from ground level to satellites in the sky.
“One of the problems is that bushfire detection in Australia is currently very dependent on public calls to emergency services,” Yebra says.
However, Australia’s varying landscapes and swathes of bushland mean different strategies will be more effective in different ecosystems.
“We think we’ll need a combination of all these layers of technology to have the highest chances of detecting and containing even a small fire,” Yebra says.
Experts from ANU and Optus are working with the Rural Fire Service, ACT Parks and Conservation Service and private companies. While some research experiments involve testing equipment with small, contained fires, the team also attends prescribed burns and uses artificial sources, including smoke machines and outdoor gas heaters.
Smart smoke detectors
At the ground level, the centre is developing sensors to place in areas of high risk or ecological value. The sensors are designed to detect specific elements of a fire, including smoke and sounds, and alert the authorities.
Fire towers have been used around Australia since the early 1900s and are typically located in mountainous areas and forested regions. These lookouts are traditionally staffed by professional or volunteer fire spotters, but Yebra and team are bringing artificial intelligence (AI) capability to what has long been a manual and solitary job.
The ANU researchers have built an automated algorithm that can detect smoke. Cameras running the algorithm could be mounted on fire towers and programmed to send alerts to emergency services when smoke is sighted.
“Basically, you give a lot of images of fire to the algorithm and it learns to identify the smoke in those pictures,” Yebra explains.
“We want to be ahead of the game so that we can prepare better for the next bushfire season.”Dr Nicolás Younes
Fire and ecology researcher Dr Nick Wilson, from the ANU Fenner School of Environment and Society, is part of a group evaluating how quickly and accurately various technologies can detect and locate fires. He says challenges with fire tower cameras remain, as it can be tricky for them to calculate exactly where smoke is coming from.
“Although smoke can be visible pretty quickly after a fire ignites, it can be quite difficult to figure out where it’s coming from. What can seem like a minor error from a fire tower can be tens of kilometres or over an hour’s drive for responding crews.”
The fight takes flight
The majority of bushfires are the result of blazes sparked by lightning strikes in remote locations, which is why the centre is collaborating with Sydney firm Carbonix to test long-range drones that can fly over secluded landscapes, day and night.
“It’s dangerous to fly at night, so that’s why we want to develop new technologies to fill that gap in remote areas where it’s difficult to get to,” Yebra says.
The idea is to determine the locations of potential lightning strikes and use drones to verify ignition and existing fires.
“A lightning storm can have thousands of strikes, so it would take a lot of time for a drone to inspect them all. But if we can narrow that down to tens of strikes using the new AI algorithms, the process is easier.”
Satellite eyes in the sky
Monitoring fire risks from the sky is also one of the remits of OzFuel, an ANU Institute for Space satellite mission that aims to assess the flammability of the Australian landscape.
The OzFuel satellite mission uses novel infrared technology to pick up what types of vegetation are more prone to catching fire.
“We want to be able to tell firefighters and emergency services when there are patches of forests that are very dry,” Postdoctoral Research Fellow and project manager Dr Nicolás Younes says. “If there’s an ignition source, like lightning or a cigarette, those sites will burn very quickly and intensely.”
The group is currently subjecting eucalyptus trees to drought conditions and using a simulated satellite to monitor how the dry conditions affect the biochemistry of the leaves.
“Hopefully the sensors will be able to detect changes in the leaf characteristics way before our eyes can,” Younes says. “We want to be ahead of the game so that we can prepare better for the next bushfire season.”
Though OzFuel is focusing on eucalyptus trees, which cover 77 per cent of Australia’s total native forest, Younes is optimistic the same principles could be applied to evaluate other types of plants and trees, and shared with other countries to improve bushfire management globally.
While modelling indicates early bushfire detection could save the economy billions of dollars, Yebra says technology will never fully replace boots on the ground but promises to provide more time and information for an effective response.
“We want to give firefighters more time and create the highest chance of being successful,” she says.
Top image: Jamie Kidston/ANU
You may also like
A new research partnership will help build a national defence system against catastrophic bushfires and stop them in their tracks before they become deadly.
New ANU technology could help predict a person's risk of developing Alzheimer's disease up to 20 years before symptoms show.
Earth’s inner core holds the secrets to the planet’s past. But how can scientists analyse something that's impossible to see from the surface?