2023-08-09 17:55:00
DISPATCH — A small town in the hinterland of Nice very exposed to the risk of fire, Tourrettes-sur-Loup is one of the first in the Alpes-Maritimes, in the south-east of France, to experiment with sensors based on the artificial intelligence to detect forest fire outbreaks as quickly as possible.
Installed since June on the tower of this small town of 4,000 inhabitants, but also on a natural belvedere in the surrounding countryside, the five detectors of the town are in permanent operation, each taking a photograph of its area every five seconds.
As soon as suspicious smoke is detected, the alert is given on the mayor’s telephone and to the department’s “risks, disasters, environment” operational force (Force 06), which relays the information to the firefighters. In the event of an alert, the mayor or the agents of this force hear a ring tone on their telephone and receive the photo of the place where the smoke was detected as well as its GPS point.
“I have to protect the inhabitants, the heritage, but also the environment, that’s part of my duties as mayor”, explains Frédéric Poma, also a volunteer firefighter.
With 30 km², its municipality is 80% in the red zone of risk of forest fires.
“We have already had a certain number of fires in the past, the installation of these devices was something important”, affirms the elected official. Installed free of charge, these five detectors, made in Poland, will cost him 6,000 euros for rental per year, if the municipality continues the experiment in a year.
Less than five minutes
“The interest of these sensors is to be able to detect smoke from a forest fire in less than five minutes, 24 hours a day, 365 days a year”, explains Pierre Borghini, a professional firefighter from Nice who has founded the company distributing these detectors.
“Artificial intelligence brings this little extra which is the comparison between them of the photos taken at regular intervals, a bit like a game of seven errors, which makes it possible to analyze the appearance of smoke following a fire outbreak,” he said.
“The cameras installed by the department on its lookout posts are doubt-removal cameras, they check for the presence of smoke when there has already been an alert, while these detectors produce the alert”, adds Mr. Borghini, who also emphasizes the ability of artificial intelligence to constantly improve.
At first, the system triggered the alert for a cloud of dust on a track, which it rectified.
Head of operations of the Departmental Fire and Rescue Service (Sdis 06), Commander Xavier Wiik believes that these technologies are “a real plus”.
“The French doctrine is a massive and rapid attack of any start of forest fire”, he notes. “The faster the fire is detected, the more reliable the removal of doubt, and the faster we can send the appropriate means, men on the ground with trucks or water bomber helicopters”, adds Mr. Wiik.
However, these sensors mainly find their interest in areas far from the coast. On the Mediterranean coast, very urbanized and frequented by tourists, detection is made by calls to 18 or 112 quite quickly.
“In more remote areas, there is a real interest that technology can bring us this look, as well as at night when there is less traffic”, specifies the fire chief.
Since the start of the summer season, nearly a hundred hectares of vegetation have gone up in smoke in the Alpes-Maritimes, compared to 36 last summer which had benefited from tropical conditions and humid air not conducive to fires .
According to official data, on the French Mediterranean rim, we went from nearly 32,500 hectares burned each year between 1973 and 1982 on average, to 12,700 over the period 1993-2002, then to 8,780 between 2013 and 2022, thanks to the strategy alert and rapid reaction.
#hinterland #Nice #artificial #intelligence #alert #forest #fires