A devastating tsunami hit northeastern Japan in 2011, killing an estimated 18,500 people. Since then, the Japanese government has made great efforts to prevent such incidents from occurring. Using machine learning technology, RIKEN scientific laboratory researchers have developed a system that can predict tsunami effects with high accuracy in less than one second.
The chief researcher of the new RIKEN laboratory project says that the main advantage of this method is the speed of prediction; Because speed is very important in informing and using early warning systems: “Conventional tsunami modeling provides results after 30 minutes, which is too late. “Our model can make predictions in just a few seconds.”
to report Interesting EngineeringIn order to achieve a high-speed tsunami prediction system, Japan has gone to the use of a large network of sensors that continuously monitor the movements of the ocean floor. About 150 offshore stations in cooperation with each other form the said network and lead to the production of a system for tsunami early warning.
The data recorded by the sensors are converted into Tsunami through special systems. In these systems, the height and extent of the tsunami on the coastline is predicted. This process normally requires solving several complex nonlinear equations, which takes 30 minutes on standard computers. The long calculation process makes people not have enough time to evacuate. That is why the RIKEN artificial intelligence model is very important to save people’s lives. Thanks to this system, people have at least half an hour more time to protect themselves.
RIKEN Science Lab researchers say they trained their machine learning system using more than 3,000 computer-generated tsunamis. They also tested the system in 480 different scenarios using data from three real tsunamis.
Researchers say that the RIKEN artificial intelligence system provides comparable results with only one percent of the computing power required by conventional approaches. They claim that RIKEN is applicable to all natural disasters where time is of the essence. Researchers are working on a storm surge forecasting system.
In 2021, RIKEN Laboratory collaborated with Fujitsu Corporation to design an artificial intelligence system capable of real-time prediction of floods caused by tsunamis. At that time, Fugaku, the world’s fastest supercomputer, was used to build a flood forecasting system.
The flood forecasting model was trained by the staggering processing power of the Fugaku supercomputer, but the result was the work of a system used on ordinary computers. The system in question predicted the flood in a few seconds through the power of standard computers.
In late 2021, researchers developed a new system that detects tsunamis through magnetic fields. A tsunami creates a magnetic field as it moves through the ocean water. These magnetic fields can be detected minutes before reaching the sea surface.
The two systems in question are very admirable in their own right, but they cannot be considered on par with the new RIKEN system. RIKEN’s artificial intelligence system is currently used for tsunamis with a height of more than 1.5 meters. Researchers are trying to optimize this system for smaller tsunamis as well.