Science

AI predicts when and where "monster" rogue waves will swell up

AI predicts when and where "monster" rogue waves will swell up
Scientists have developed an AI system that can predict giant rogue waves in advance
Scientists have developed an AI system that can predict giant rogue waves in advance
View 1 Image
Scientists have developed an AI system that can predict giant rogue waves in advance
1/1
Scientists have developed an AI system that can predict giant rogue waves in advance

Gigantic waves can seemingly come out of nowhere, threatening ships and oil rigs. Now, an AI system trained on centuries of data has been able to predict when these rogue waves will occur, revealing new insights into how they form.

Rogue waves don’t necessarily have to be the biggest waves in the ocean – by definition they just need to stand at least twice as high as surrounding waves. They form due to interactions between things like currents and winds, which separates them from larger, more destructive waves like tsunamis, which are caused by seismic activity.

However, that’s not to say rogue waves can’t be dangerous themselves. They can potentially damage ships and oil rigs, and to make matters worse they seem to appear at random. Scientists have previously investigated ways to predict them with a few minutes’ warning, but a comprehensive new system could provide a broader heads-up.

Researchers at the University of Copenhagen started by feeding a huge amount of wave data into an AI system. This data was gathered by buoys in 158 locations around the world, all working 24 hours a day, constituting over 700 years’ worth of data and more than a billion waves.

“Our analysis demonstrates that abnormal waves occur all the time,” said Johannes Gemmrich, second author of the study. “In fact, we registered 100,000 waves in our dataset that can be defined as rogue waves. This is equivalent to around one monster wave occurring every day at any random location in the ocean.”

From their analysis, the team found that the most common cause of rogue waves isn’t what scientists thought. It occurs through a phenomenon known as linear superposition, where two wave systems cross over and briefly reinforce each other.

“If two wave systems meet at sea in a way that increases the chance to generate high crests followed by deep troughs, the risk of extremely large waves arises,” said Dion Häfner, first author of the study.

The team says their new algorithm could be applied to incoming data from buoys to raise the alarm when the exact combination of risk factors for rogue waves arises in a region of ocean. Shipping companies could then use this information to plan alternate routes if necessary.

The research was published in the journal PNAS.

Source: University of Copenhagen

2 comments
2 comments
Karmudjun
Good write-up Michael. I was not surprised that the "700 years worth of data" gave researchers a clue that rogue waves (twice the size of surrounding waves) could occur in the hundreds of thousands times during their data collection period. Hint - it wasn't 700 actual years of data. One Monster wave (that would be on the scale of ~20 meter wave (or taller) - That is a 60 foot wave for the US readers - EVERY DAY! Somewhere on the oceans.....but forecasting would be great! Good job, U. of Copenhagen.
Brian M
'It occurs through a phenomenon known as linear superposition, where two wave systems cross over and briefly reinforce each other.'

Think most scientists new that, well at least by instructor did at sailing school!

Great use of an AI. With AI weather predictions competing as well as large weather data models its changing the way we will look at things.


Great use of AI althiugh I do think most scentyiset new it ws due to


terra firma