While devastating forest fires are more and more present each year, an Australian university claims to have developed technology that makes it possible to anticipate their spread from space using artificial intelligence.
Anticipating the spread of forest fires from space
In recent years, forest fires have regularly made the headlines. Although few regions are really spared, this phenomenon favored by global warming is indeed a growing concern, especially in North America, the Amazon and Australia. In Australia, researchers are at the origin of an innovation in the fight against these firesas indicated in a press release published on June 5, 2024. In recent months, the University of South Australia (UniSA) has led a project funded by the SmartSat CRC, a consortium of industry and research organizations which aimed to objective of relaunching the Australian space industry by developing revolutionary technologies.
The project in question involves the launch of CubeSat-type satellites which carry artificial intelligence. The goal is to better anticipate the spread of fires from space. This should allow ground teams to react more quickly to these fires and thus limit the loss of human, animal and plant lives, as well as material losses.
Up to 500 times faster detection capacity
Australian researchers used cutting-edge AI technology in the development of a fire smoke early detection system. This system also has the advantage of being energy efficient. A first CubeSat was launched as part of the Kanyini mission. Its objective will be to monitor the quality of inland and coastal waters, but also land management. The device integrates a sensor with a hyperspectral imager which allows light to be captured in different wavelengths. Detailed surface maps can therefore be developed by the teams. According to UniSA, this device could help detect fires up to 500 times faster. Let us remember, however, that Earth observation satellites do not carry any means of real-time analysis of the images obtained.
“Smoke is usually the first thing you can see from space before the fire gets hot enough and big enough for sensors to identify it. Early detection is therefore crucial”said Stefan Peters, the project’s chief scientist.
During the AI training phase, researchers used satellite images of recent bushfires in Australia. Machine learning then allowed the AI to detect smoke and, above all, to differentiate it from clouds. Their simulated case study was then satisfactory, because the device took only fourteen minutes to detect the problem and send the information to the ground. Finally, the team hopes that its device will be fully operational in orbit by 2025.
