Tsunamis are notoriously difficult to spot on the open ocean as they race towards shore. But in the summer of 2025, scientists watched one unfold as it happened.

It was the most powerful earthquake anyone had seen for nearly 15 years. It struck off the far eastern coast of Russia's Kamchatka Peninsula in July 2025 – an 8.8 magnitude quake that also triggered a tsunami with waves racing outwards at more than 400mph (644km/h). Within minutes, alarms sounded in communities around the Pacific Ocean.

Millions people were ordered to evacuate in the tense hours that followed, including at least two million in Japan alone. But as the wave propagated out across the ocean, it triggered something other than just fear – it created ripples in Earth's atmosphere.

The ocean, moving up and down across such a vast area, was disturbing the atmosphere above it and messed with global satellite navigation signals. But this disturbance also allowed scientists to spot the tsunami almost in real-time.

Purely by chance on the previous day, US space agency Nasa had added an artificial intelligence component to a disaster alert system called Guardian, enabling it to flag major events to scientists automatically. Roughly 20 minutes after the Kamchatka earthquake happened, tsunami-watchers knew that waves were heading for Hawaii, 30 to 40 minutes before they arrived.

Fortunately fears of widespread damage from the tsunami were not realised on this occasion. The waves that hit Hawaii were up to 5ft (1.7m) high, causing only minor flooding and no severe damage. Most of the tsunami's energy dissipated in the open ocean while the largest waves hit unpopulated areas. But had it been worse, those extra minutes of warning could have been crucial.

The episode proved that Nasa had a system that can, given the right conditions, detect a tsunami well in advance of its arrival on a coastline – just by listening to radio signals used by orbiting global navigation satellites as they communicate with ground stations on Earth. The same approach can even detect volcanic eruptions, rocket launches and underground nuclear weapons tests.

"They were able to say in pretty much real time, 'there is a tsunami'," says Jeffrey Anderson, a data scientist at the US National Center for Atmospheric Research who helped develop the Guardian system. Anderson admits that, years ago when he first heard about proposals for the technology, which he later helped to develop, he thought it sounded "kind of crazy".

The idea of using radio signals beaming between ground-based receivers and satellites for near real-time tsunami detection has been around for a decades. A handful of academic papers in the 1970s discussed such a system in principle but only in the 2020s did it become a reality with the arrival of Guardian. In 2022, Anderson and authors from Nasa's Jet Propulsion Laboratory in California, US, published

© BBC