World’s Fastest Supercomputer Powers Real-Time AI Tsunami Predictions

Due to the high seismic activity in the region, tsunamis are a real threat to the people of Japan — a fact that was terrifyingly driven home by the 2011 tsunami that triggered the Fukushima nuclear disaster.

Now, using the world's fastest supercomputer, Riken and Fujitsu developed a powerful predictive AI tool that enables real-time predictions of the flooding caused by a tsunami.

The hardware used for the development of the new tsunami prediction tool is Fugaku, the world's fastest supercomputer, which was powered up last year and was developed by Riken and Fujitsu.

An AI tool trained on 20,000 tsunami scenarios

For the new tsunami prediction tool, a team from Fujitsu, Tohoku University, and the University of Tokyo's Earthquake Research Institute created 20,000 possible tsunami scenarios for use as training data for the AI model, New Atlas reports.

A deep learning algorithm studied and compared the waveform data of previous tsunamis with the resulting flooding conditions on dry land. Over time, the AI learned to predict tsunami flooding in near real-time, with a high spatial resolution.

Though the model required the immense computational power of Fugaku for training, it can be loaded onto regular PCs where it can carry out predictions in seconds based on waveform data once trained.

The algorithm's accuracy was demonstrated in simulations of a large earthquake and tsunami impacting Tokyo Bay. The near real-time prediction matched with flood modeling from the Cabinet Office of Japan.

As the tool can be loaded onto ordinary PCs, the researchers hope it will be deployable by disaster response teams. Such quick, accurate predictions of tsunami flooding might even allow teams to plan out their missions to reach the most affected areas ahead of time.