Supercomputers Predict Tsunami in 10 Minutes Following Earthquake!

Predicted tsunami arrival times

Predicted probability of building destruction

Failure to accurately estimate tsunami height after large earthquake

The Great East Japan Earthquake that occurred on March 11, 2011, exceeded the scale previously envisioned for earthquakes and thus the height of the ensuing tsunami was not accurately predicted. While the estimated tsunami height was announced three minutes after the start of the earthquake, the large tsunami that swept through the coastal cities far exceeded expectations.

This underscored the difficulties of estimating the tsunami height together with the start of an earthquake in real-time. This disaster also illustrated the need for information beyond wave height, such as the scope of flooding, to minimize the damage.

That led Tohoku University and Fujitsu Laboratories to launch the development of a new tsunami simulation model. Now Tohoku University and Fujitsu Laboratories have jointly developed a high-resolution tsunami model that is able to run on a supercomputer--K computer--and based on Tohoku University's TUNAMI-N2 simulation model, which is in widespread use globally.

Tsunami model predicts scope of flooding in approximately 10 minutes following the start of an earthquake

This tsunami model quickly predicts the extent of flooding from sea surface deformation, which is the source of a tsunami and is estimated based on observation data on the shape of the tsunami out at sea and on tectonic shifts onshore at the time of the earthquake. It has been shown that using this model, if an approximate 10 km area around coastal districts is modeled at a high resolution of even 5 m, simulations of the general extent of flooding at the time when the Great East Japan Earthquake hit could be completed in two minutes. This simulation is very computation-intensive, and performing the same computations on a workstation would take several days.

Flooding from the Great East Japan Earthquake started in Sendai one hour after the earthquake hit. Using this model makes it possible to provide basic flood predictions in approximately 10 minutes including the time needed to analyze the tsunami wave source.

For example, it will become possible to widely report on TV or warn residents in advance via smartphones or other devices using specific estimated images. These images can demonstrate the impact to roads located in the area, preventing a tsunami from going through passages beneath roads, as occurred during the Great East Japan Earthquake. This will allow residents to quickly evacuate based on more detailed predictions and help the government encourage residents to take more appropriate evacuation measures.

The practical use and implementation of this model around the world could make sure that regions and countries, where damage from tsunamis is anticipated, can be better prepared.
Tohoku University and Fujitsu Laboratories have conducted joint research on tsunamis since 2014. Going forward they will test research results against a hypothetical case of a massive earthquake in the Nankai Trough, which is a large region from which earthquakes originate, and in doing so Tohoku University and Fujitsu Laboratories will contribute to develop effective disaster-response measures.