National

Japanese team developing AI-based system to forecast chance of tsunami and scale of damage

by Magdalena Osumi

Staff Writer

Drawing lessons from one of the worst disasters in the nation’s history, a team of Japanese researchers is developing an artificial intelligence-based tsunami-forecasting system set for release in fiscal 2020 that could help limit loss of life and property in future calamities.

In March 2011, massive tsunami 30 meters high triggered by a 9.1 magnitude earthquake destroyed a large swath of the Tohoku coastline, taking not only residents but also entire communities and businesses by surprise. The researchers hope the new system will help municipalities and companies nationwide better prepare for any future calamities and prevent related disasters, such as the triple core meltdown at the Fukushima No. 1 power plant that resulted from the tsunami.

The team, made up of researchers from risk management consultancy Tokio Marine & Nichido Risk Consulting Co. and the National Research Institute for Earth Science and Disaster Resilience are working on the nation’s first system for predicting the likelihood of tsunami based on location, as well as the scope of damage in areas expected to be hit.

“The existing forecasting system only estimates the maximum height of a tsunami but not its likelihood … and sometimes there are no available measures to prepare for the worst-case scenario,” a spokesman for Tokio Marine & Nichido Risk Consulting said by phone.

He explained that the new system will provide a detailed forecast for each postal address that will enable companies or other facility operators to draft effective disaster-preparedness plans.

The researchers plan to make the system available to municipal governments and companies nationwide so they can improve disaster-preparedness and revise contingency procedures.

For instance, by inserting one’s address, a business owner can learn the probability of 3-meter or higher tsunami arriving if a quake strikes along Japan’s coastline, and the depth of any floodwater expected in the area where it is situated.

“With such knowledge it will be easier to make a decision to relocate if the current area is at high risk,” the spokesman said.

The spokesman also said he believes the system could help prevent disasters like the one at the Fukushima No. 1 plant.

The company said the system will use AI to analyze data compiled in the government’s earthquake-prediction database, topographic data and the height of seawalls built along coastal areas. The system will calculate the projected heights of waves and expected immersion depth.

Initially, the forecasts will be based on earthquake predictions for areas along the Nankai Trough, Japan Trench, Kuril-Kamchatka Trench and Sagami Trough, which are all located along the Pacific coast. But the researchers also plan to add information for areas along the Sea of Japan coast at a later date.