Developed by an international team led by Dr Alberto Ardid, a lecturer in Civil and Environmental Engineering at UC, the system uses machine learning to analyse weather station data and detect patterns that often occur before fires ignite.

Wildfire risk is rising globally as climate change drives hotter, drier conditions. Rapid shifts in weather can cause fire danger to spike within hours, yet many official warning systems currently update only once a day.

The UC-led research aims to address this gap with an AI-based forecasting system that updates every 30 minutes, providing near real-time insight into changing fire risk.

“Wildfires don’t wait for the daily forecast. If fire weather conditions can change within hours, our warning systems need to move just as quickly,” Dr Ardid says.

The latest study builds on earlier research published in 2025 that first demonstrated the concept using data from a single region in Queensland, Australia. The new research expands the analysis to multiple regions with different climates, including the Sunshine Coast, Brisbane and Hobart.

“We originally developed the model as a proof of concept in one region. However, in this new study we tested it across multiple regions to see whether the approach works under different fire weather conditions,” he says.

Across all three locations, the AI model consistently outperformed the standard Fire Behaviour Index, part of Australia’s official fire danger rating system. The machine-learning system improved forecasting performance by 10–30 percent, detecting significantly more fire events in advance.

Researchers evaluated the model using more than 60 years of historical weather and fire data, simulating how it would perform under real-world conditions.

The study also measured the economic value of improved forecasts, which is a new element rarely included in wildfire prediction research.

Wildfires in Australia can be devastating; in the 2019-2020 season, across the country wildfires burnt almost 17 million hectares, destroyed 1.5 billion wildlife animals, caused $1.9b in insurance claims and 33 people lost their lives.

Using a cost–loss framework, the team found the AI system could double the economic savings when compared with existing forecasting tools by reducing missed fires and unnecessary false alarms. Additionally, as the model relies on standard weather station data, it could be deployed widely without requiring new infrastructure.

Dr Ardid says the approach could also be applied in New Zealand, where similar meteorological monitoring networks already exist.

“The models rely on weather station data, which already exists through monitoring networks in New Zealand. That means the system could potentially be implemented without new infrastructure, either at a regional level for fire management agencies, or also at a more local or sector level, for example within the forestry industry.”

Faster, data-driven forecasting could help fire agencies respond earlier, allocate resources more effectively and reduce the environmental and economic impacts of large fires as wildfire risk continues to increase.