New Tsunami Prediction Method

Scientists have developed a new way to predict tsunamis faster and more accurately than before—especially those waves that arise not from earthquakes, but, for example, from underwater landslides or other oceanic processes.

Usually, early warning systems focus primarily on seismic signals. But such methods do not always recognize other types of threats well. The new approach uses data from coastal pressure and sea-level sensors, as well as neural networks that help quickly reconstruct the picture of what is happening in the ocean.

At the core of the method lies a special model—the so-called Green's functions. They describe how a wave propagates from the source to the shore. Previously, such calculations required significant computational resources and time. Now, neural networks allow them to be performed much faster.

It is particularly important that the system continues to function even if some sensors fail or transmit incomplete data.

The method has already been successfully tested on historical scenarios and models of possible future tsunamis off the coast of Japan.

This reduces forecast uncertainty and makes early warning systems more reliable—especially for rare and complex events that were previously difficult to assess in advance.

What does this event add to the sound of the oceans?

Sometimes safety begins not with the waves, but from the moment we learn to hear the ocean before it reaches the shore.

New computational methods help read water movements more accurately—and turn ocean observation into a dialogue where the warning arrives before the danger.