Science

PARIS: Better tsunami prediction thanks to gravitational waves and artificial intelligence – Presse Agence

French and American scientists were able to instantly estimate the magnitude of large earthquakes by analyzing gravitational waves using artificial intelligence (AI).

This progress should make it possible to improve tsunami warning systems and avoid tragedies like the Fukushima tragedy.

Dazzling and invisible gravitational waves are a valuable warning signal that accompanies the occurrence of large earthquakes. Generated by the movement of a huge mass of rocks, they propagate at the speed of light, that is, 100,000 times faster than destructive seismic waves. But their signal is so weak that it has not been possible to use it until now. “We managed to overcome this hurdle by developing a deep learning algorithm that learned how to interpret this data in real time to estimate the magnitude and location of powerful earthquakes,” explains Andrea Licciardi, geophysicist, GEOAZUR postdoctoral fellow and first author of the study. study of the subject. And it could change the face of tsunami detection.

Seismic Wave Alert

Currently, earthquake warning systems are based on the detection of seismic waves, the fastest of which travels at a speed of 7 km per second. Depending on the location of the epicenter, the position of the nearest sensor, and the population that needs to be protected, an alert can give a few precious seconds to take cover, for example, under a table. In Japan, where earthquakes are very frequent, residents receive an alarm SMS this way, and this method works well.

In this case, the use of gravitational waves will not fundamentally change the quality of detection: 100,000 times faster than a few seconds, which does not give much time to find safer shelters … And most importantly, the gravitational wave signal is so weak that it can only be used for earthquakes force of more than 8 points, which is already potentially very destructive. But this weakness could be a decisive advantage for tsunami detection…

Imprecise systems

The occurrence of an earthquake under water can cause a tsunami in the next minutes or hours, depending on the distance from the epicenter to the coast. The warning systems then seek to estimate the magnitude of the event and therefore the size of the wave associated with it.

The challenge is to shelter the public and turn off sensitive equipment if flooding exceeds coastal protection dams. We are talking, for example, about railways and nuclear power plants. “But it is very difficult for warning systems to quickly assess the magnitude of very large earthquakes, which is nevertheless important information for tsunami control,” explains Quentin Blethery, geophysicist at GEOAZUR, who led the study. For both instrumental and fundamental reasons, it is not possible to tell the difference between a magnitude 8 event and a magnitude 9 event from the first recorded seismic waves that then tend to saturate the instruments. »

Avoid Fatal Mistakes

And the consequences of such a mistake can be catastrophic: during the 2011 Fukushima earthquake, all systems declared a magnitude of 8, which did not require special precautions, since the country has 13 meters high dams, largely sufficient for the expected three-meter waves. But in fact, the magnitude was 9 points and waves over 15 m high washed everything away …

“Our algorithm is very good at distinguishing between a magnitude 8 earthquake and a magnitude 9 earthquake, simply because the gravitational signal from the former is too weak to detect,” the scientist says. And it is capable of estimating the magnitude of very strong earthquakes in real time and more reliably, without therefore saturating, as is the case with existing tsunami detection systems. »

The study proved its effectiveness on data recorded during past earthquakes, events for which the limits of seismic wave systems could be measured. “The algorithm tested in Japan has proven to be able to estimate the magnitude of the Fukushima earthquake faster and more reliably than any existing system,” says Andrea Licciardi.

“The next step will be to integrate gravitational wave analysis into operational warning systems, which will better protect the population and infrastructure,” concludes Quentin Blethery.

SOURCE: IRD Press Office Newsletter

Back to top button