The mercury is rising as bushfire season descends on the Sunshine Coast. Record-breaking spring temperatures and recent wet weather, which has sparked vegetation growth, are adding fuel to the mix.
As recently as October 27, Maroochydore recorded an unheard-of temp of 39.1°C . And with peak bushfire season earmarked from now until early summer, residents are being urged to prepare their homes for the season.
In a new move, those living in bushfire-prone areas are also being invited to help predict fire risks through mobile technology.
The NOBURN mobile app allows residents to upload photos of fuel loads, such as grass and shrubs, dryness and structures of forests to help scientists gather useful data.
Released in 2023, it was developed by a team of researchers from the University of the Sunshine Coast (UniSC) and University of Adelaide’s Australian Institute for Machine Learning, in partnership with Noosa Shire Council and funded through the Federal Department of Industry, Science and Resources.
The team is currently being led by UniSC’s Dr Sam Van Holsbeeck and research fellow, Shelby Schumacher.
Ms Schumacher says the project is the culmination of years of research and collaboration between researchers in artificial intelligence, forestry, human factors and science communication.
NOBURN works by using AI to determine bushfire risk based on the uploaded images.
‘Citizen scientists’ around Australia are encouraged to take safe, geo-tagged photos of local fuels such as surface litter, shrubs, canopy and bark.
Ms Schumacher says the app records the photos’ time and location, and the user adds a couple of simple descriptors. The labelled photos then go to the AI team.
“The AI is then trained on these images plus weather, vegetation, topography and fire-history layers to learn what combinations of fuels are most likely to result in harder to control fires,” Ms Schumacher says.
“In short, community photos make the AI smarter, and the AI turns photos into decision-ready fuel intelligence.”
Ms Schumacher’s role is to link the field/data-collection side with the machine-learning team and to collaborate with council and emergency partners to ensure the data is useful for fire management.
“We’re currently fine-tuning the AI model so it can provide instant fuel-hazard classifications and support better bushfire-readiness and education for Australians.”
The app was developed in the wake of the black summer bushfires (2019-2020), which burned across multiple states, but the worst, most prolonged fires were based in New South Wales and Victoria, with major fires also in Queensland, South Australia (including Kangaroo Island and the Adelaide Hills), Western Australia, Tasmania and the ACT. They destroyed around 2000 homes and burned more than 10 million hectares nationally.
Recent uploads of photos taken on the Coast show a build-up of fast-curing, fine fuels and denser vegetation after several wet seasons, which Ms Schumacher says is consistent with what local fire agencies have been reporting.
Ms Schumacher stresses that NOBURN itself doesn’t replace fire risk ratings. Rather, it feeds the agencies better fuel intel.
“What our uploads are showing is that with a wetter than usual start, the Coast still has pockets of elevated risk because of heavy, continuous fuels. On hot, dry, windy days, those fuels can become an elevated risk, which is why councils and Queensland agencies have been pushing preparedness messages this month.”
For Ms Schumacher, her biggest achievement in this field of research has been seeing local people use the NOBURN app, especially in places that have already burned.
“That’s the whole promise of NOBURN. Everyday photos lead to smarter prediction, which leads to safer communities,” she adds.
noburn.app/
