A new artificial intelligence system could help detect wildfires more effectively, offering a powerful tool for early warning and environmental monitoring.
The system was developed by researchers at the Universidade Federal do Amazonas, in Brazil, and uses a convolutional neural network (CNN) – a type of AI model designed to process visual data in a way that mimics the human brain. By analysing satellite imagery, the CNN can identify wildfires with remarkable accuracy..
A New Approach to Wildfire Detection
The researchers trained the program using images from Landsat 8 and Landsat 9 satellites, which provide global coverage every 16 days. These satellites capture near-infrared and shortwave infrared imagery, which are essential for detecting vegetation changes and surface temperature shifts — key indicators of wildfires.
The model was trained on 400 satellite images—200 containing wildfires and 200 without.
It achieved a 93% accuracy rate in distinguishing between fire-affected and unaffected areas. When tested on 40 new images outside its training set, the CNN correctly identified 23 of 24 wildfire images and all 16 non-wildfire images. With further data exposure, the AI is expected to refine its accuracy even more over time.
These findings, published in the International Journal of Remote Sensing, suggest that CNN could supplement existing AI-driven wildfire monitoring systems and improve response strategies.
“The CNN model serves as a complement to well-established large-scale monitoring systems,” says co-author Professor Carlos Mendes, who has a PhD in physics.
He says this includes satellite-based Moderate Resolution Imaging Spectroradiometer (MODIS) and Visible Infrared Imaging Radiometer Suite (VIIRS) which are widely used for continuous wildfire detection.
“By combining the wide temporal coverage of the current sensors with the spatial precision of our model, we can significantly enhance wildfire monitoring in critical environmental preservation zones.”
Protecting the Amazon
The Amazon is one of the most ecologically vital yet critically endangered regions on the planet. In 2024, 44.2 million acres of the Amazon rainforest burned, with the number of fire outbreaks surging by 42% compared to 2023.
Despite near real-time satellite monitoring, existing systems struggle with low-resolution images that make it difficult to detect smaller, remote or concealed fires—such as low-temperature understory fires hidden beneath dense forest canopies.
“The ability to detect and respond to wildfires is crucial for preserving the delicate ecological balance of these vital ecosystems, and the future of this Amazon region depends on decisive rapid action,” explains lead author Professor Cíntia Eleutério.
To further enhance accuracy, the research team recommends increasing the number of training images used to refine the CNN model. They also suggest expanding its applications beyond wildfire detection, including deforestation detection.
The Future of Fire Fighting
This research contributes to a growing body of work on AI-powered fire detection, monitoring, and prediction.
Australia’s Bushfire Research Centre of Excellence has recently highlighted AI’s role in improving fire response strategies.
“AI is becoming an important tool in this field, particularly in automating fire detection from satellite, drone, and ground-based camera or sensor network data,” says Professor Marta Yebra, Director of the Bushfire Research Centre of Excellence at the Australian National University.
“As climate change fuels rising fire risks, AI-driven wildfire detection could be a game-changer, offering faster alerts, smarter prevention, and a crucial edge in the fight against escalating blazes.”
Other ways to detect wildfires
Are you interested in the energy industry and the technology and scientific developments that power it? Then our free email newsletter Energise is for you. Click here to become a subscriber.
