A research team led by the University of Canterbury has developed an AI-based system that can predict wildfire risks significantly faster and more accurately than existing methods. The system uses machine learning to analyse weather data and identify patterns that typically occur before fires ignite.
In contrast to many current warning systems that update only once per day, the new model provides updated risk assessments every 30 minutes, enabling near real-time forecasting. Especially in the context of increasing extreme weather conditions driven by climate change, this could be crucial for responding more quickly to rapidly changing risk situations.
Tests conducted across multiple regions in Australia show that the AI model outperforms existing systems by 10–30% and detects significantly more fires in advance. The analysis is based on more than 60 years of historical weather and fire data.
In addition to improved forecasting performance, the system also demonstrates clear economic benefits: more precise warnings can reduce both false alarms and missed fires, leading to substantial cost savings for emergency services. As the model relies solely on existing weather station data, it can also be deployed without the need for additional infrastructure.
More information here. The complete study can be found here.
(Image source: AI-generated)