On the screen, a neon blue curve labeled “Tohoku”. Here is the simplest geophysical summary of the 2011 earthquake in Japan responsible for the Fukushima disaster. Worldwide, earthquakes have killed nearly one million people in the last 30 years, mainly in Indonesia and Haiti. In Sophia Antipolis, west of Nice, Quentin Blatari and his team try to predict and prevent these disasters using artificial intelligence.
“Earthquake prediction is the holy grail in seismology”, Says Geophysicist at the Research Institute for Development (IRD). The research fellow received a European grant from the ERC program (European Research Council), which amounted to 1.5 million euros over five years. “Early Seismic Signal Detection for Artificial Intelligence”. The work was published in the journal in mid-May Nature, Based on a recent discovery: gravitational waves associated with earthquakes, in this case Fukushima. Gravitational waves coincide with the “fold” of space-time due to the mass movement, much like a bowling ball digging into an extended web.
Previously detected during infinite collisions in space, as two black holes engulf each other, gravitational waves are also generated during large earthquakes on Earth, where large mass motions occur. The idea came during a seminar at Caltech in the United States, where a man in charge of a space gravitational wave detector was trying to figure out how to control seismic noise so as not to distort the device properly.I remember Jean-Paul Ampuro, IRD’s research director and co-inventor of the signal.
Too early but difficult to detect
The gravitational signal of an earthquake, known as pegs (“prompt elastogravity signal”), is particularly discreet, but can be detected by a good quality seismometer. “All earthquakes produce pegs, but these signals are only observable above magnitude 7.5 or 8.As Kevin Juhel describes, geophysicist. In total, there are less than ten events for which we have been able to record them. “The scientific community was skeptical because the signal was so weak compared to the sound of a very strong earthquake.” Jean-Paul could recognize Ampuero.
These gravitational waves travel at the speed of light as the fault ruptures, where the P wave, the first seismic wave to appear, spreads “just” at a speed of about six kilometers per second. During a large-scale earthquake, a seismometer will therefore first register a peg before receiving the P wave. “Using gravitational waves, the alert system survives for about ten secondsGabriella Arias is a doctoral student at Quentin Blatter’s team. It doesn’t seem like much but, combined with the automated system, it can shut down a power station or an airport. A Or just to stop a surgical procedure at the hospital that could cause the shock to slip.
Seismometer which is perfect for large earthquakes
The project is more suitable for tsunami. In Japan, in 2011, authorities issued a warning three minutes after the quake. The American-led Pacific Warning Center took nine minutes to respond. This new approach could ultimately accelerate siren triggering, even in developing countries without infrastructure. Because, in addition to their instantaneous, gravitational waves are felt all over the world. Countries equipped with seismometers can thus act as sentinels for others.
In general, the formation of a tsunami after an earthquake depends on two factors: the magnitude and the focal mechanism, i.e. the manner in which the rupture occurred. At sea, the process of subduction, where one plate moves beneath another, leads to a tsunami, as the sea level rises sharply.
But “Most systems are full and are very weak in determining the actual magnitude of an earthquake and therefore the magnitude of a related tsunami.”, Details Gabriella Arias. In Japan, the predicted wave was much smaller than the wave that eventually broke. “Systems expected tsunami with 8 magnitude earthquake, not 9 magnitude earthquake”, Quentin Bletery explains. On the contrary, “The peg is not full”, Gabriella Arias continued. The team therefore developed an artificial intelligence feeding algorithm that uses these initial signals to instantly determine the size and location of cracks.
At first but not a precaution
“Fortunately, there are not many major earthquakes. So we simulated more than half a million of them to have enough data to train a neural network.Says Quentin Bletery. AI is then able to give the magnitude of any earthquake by observing the gravitational waves emitted. A“We have already tested the algorithm against the earthquake, it worksWelcome to Andrea Licciardi, who created the computer program within the team. Now, we need to test it in real time and in real situations. This is how I imagine science: useful to the population. A
As much as the future is not enough to read. Although instantaneous, gravitational waves only occur when one explodes, not before. In prediction, the second part of his research grant, Quentin Bletery is so cautious. “We do not know if there is anything at present, a signal or otherwise, which could indicate a future earthquake.She admits. Artificial intelligence can help if you know what you are looking for. A
The latest major earthquake
August 2021. A magnitude 7.2 earthquake shook Haiti and killed at least 2,000 people. The catastrophe is reminiscent of 2010, when more than 200,000 people died.
• March 2011. The Tohoku earthquake, measuring 9.1 on the Richter scale, caused a massive tsunami and destroyed the coast of Japan, affecting the Fukushima nuclear power plant.
December 2004. An earthquake measuring 9.1 on the Richter scale and the ensuing tsunami near Sumatra have killed at least 200,000 people. The devastating wave spread to Africa and Australia.