In the 10 years since the Great East Japan Earthquake and tsunami of March 11, 2011, scientists have sought answers to a variety of questions relating to the deadly tsunami that began tearing through coastal communities just 15 minutes after the quake.

Researchers have probed how a tsunami gathers height as it nears a shoreline and how this affects the damage it can cause. They’ve also begun to assess technologies for the early detection of tsunamis and improving tsunami observing systems across Japan.

“The speed of a tsunami offshore is the same as a jet airplane and its speed inland similar to Usain Bolt,” says Nobuhito Mori, a professor in the Coastal Disaster Research section of the Disaster Prevention Research Institute at Kyoto University.

While tsunamis are fast, they are not as fast-moving as the earthquakes themselves. They slow down and produce higher waves in shallower water.

Further, the behavior of tsunami waves is less understood once they begin to travel through streets and buildings, making it difficult to make accurate predictions.

“That’s why early evacuation is so important,” Mori says.

The Meteorological Agency (JMA) issued a local tsunami warning three minutes after the earthquake struck, and the Pacific Tsunami Warning Center, run by the U.S. National Oceanic Atmospheric Administration (NOAA), also issued its regional warning six minutes later. The response times were indicative of the complex calculations and heavy computation required for tsunami warnings. NOAA had to first determine whether the earthquake had happened in the ocean basin, the likely state of deformation of the ocean floor, and the kind of motion it created.

The JMA’s warning was criticized for underestimating the size of the tsunami. The agency acknowledged that the underestimated forecast had led to a slow evacuation, giving some people just 15 minutes to evacuate in the hardest hit areas. In March 2013, the agency updated its warning system and introduced new analysis procedures based on an earthquake’s maximum possible magnitude.

A tsunami is a series of extremely long waves caused by a large and sudden displacement of the ocean, usually the result of an earthquake below or near the ocean floor. This force creates outward waves in all directions away from their source, sometimes crossing entire ocean basins. Whereas wind-driven waves only travel through the highest layer of the ocean, tsunamis move through the entire water column, from the ocean floor to the surface.

The Global Historical Tsunami Database reports that since 1900, more than 80% of likely tsunamis were generated by earthquakes at destructive plate boundaries. However, tsunamis can also be caused by landslides and volcanic activity. Once a tsunami forms, its speed depends on the depth of the ocean and its wavelength, the distance from crest to crest, may be hundreds of kilometers.

Mori says tsunamis are trickier to detect early than earthquakes and that alerts take longer to generate.

“Current tsunami forecasting by the JMA is based on selecting the closest tsunami scenario from a large number of pre-computed tsunami scenario databases focused on the epicenter and fault information. This makes tsunami forecasting time-consuming because the source information is required first.”

Scientists are investigating ways to speed up computer alerts without sacrificing accuracy.

One recent project, announced in February, looks at how a new artificial intelligence model can harness the power of the world’s fastest supercomputer, Fugaku, to predict flooding in coastal areas before the tsunami reaches land.

The new technology was developed by a collaborative team of researchers from the International Research Institute of Disaster Science (IREDeS) at Tohoku University, the Earthquake Research Institute at the University of Tokyo and Fujitsu Laboratories.

Scientists used Fugaku to generate training data for 20,000 possible tsunami scenarios based on high-resolution simulations. An AI model then uses offshore waveform data generated by the tsunami scenario to predict flooding before landfall.

Whereas conventional prediction technologies require the use of supercomputers, once trained on Fugaku, the AI model can be run on ordinary computers. The research team applied the model to a simulation of tsunami flooding in Tokyo Bay on a standard PC. The predictions proved highly accurate within seconds, matching the flood modeling data released by the Cabinet Office.

The research could make it possible to more accurately and rapidly obtain flooding forecasts in specific areas, as well as predict the potential impact of localized waves on buildings and roads in coastal urban areas. Researchers will introduce new scenarios and continue working with the system this year with the goal of being able to use AI to predict tsunami flooding over wider areas. It is hoped the system will also help to make evacuation measures more efficient.

The possibilities of AI show how far technology has advanced since the earliest form of tsunami disaster prevention, where Japanese coastal communities carved warnings into stones. Hundreds of these tsunami stones, some more than six centuries old, dot the country’s coastline, standing in silent testimony to past destruction. More recently, the worst affected areas have signs marking how high the water level reached in 2011.

The nation’s tsunami warning system has been bolstered in recent years through various additional systems, which allow the JMA to estimate tsunami heights and the arrival times of waves more accurately. The warning system gathers data from ocean-bottom tsunami sensors further offshore, GPS buoys, coastal-area tsunami meters, broadband strong-motion seismometers and DART buoys.

The installation of a large-scale seafloor observatory network for tsunami and earthquake detection, called S-net, in 2017, has so far proved effective, says Mori.

The S-net system consists of 150 seafloor observatories, which are connected with submarine optical cables, measuring roughly 5,700 km in total length along the Kuril and Japan trenches. Each observatory has multiple seismometers and highly sensitive water-depth sensors for tsunami detection.

Mori believes the GPS buoys and ocean-bottom tsunami sensors are currently the most accurate ways to monitor tsunamis. However, he thinks there’s room for development and sees the assessment of technologies, such as AI models, as having a potentially bigger long-term impact.

“Although the early warning system is important and most reliable for evacuation, long-term assessment combining hardware and software countermeasures will save many lives,” he concludes.

In a time of both misinformation and too much information, quality journalism is more crucial than ever.
By subscribing, you can help us get the story right.