This AI can predict ship-sinking ‘freak’ waves minutes in advance

Rogue waves are freakishly large ridges of water known for rising out of the blue to ambush hapless ships and beachgoers. But a new artificial intelligence model can predict most of these surprising swells up to five minutes in advance, mechanical engineers Thomas Breunung and Balakumar Balachandran of the University of Maryland in College Park report July 18 in Scientific Reports.

Cresting more than twice as high as surrounding waves, rogue waves may form where converging swells raise a single, amplified wave or where ocean currents compress swells into powerful billows. Though researchers have noticed that certain patterns may precede these sudden surges, no effective forecasting algorithm had yet been developed (SN: 6/8/15). Such a tool could be lifesaving — from 2011 to 2018, collected eyewitness accounts indicate that rogue waves killed 386 people and sank 24 ships.

Using roughly 16 million data points collected at half-hour intervals by a network of 172 ocean buoys, Breunung and Balachandran trained an AI program to distinguish wave patterns that preceded rogue waves. The program predicted 3 in 4 rogue wave arrivals at buoys in the network one minute in advance. When the lead time was extended to five minutes, around 7 in 10 waves were predicted.

Notably, the program anticipated rogue waves roughly as well at locations where it had received no training data. “If you want to predict rogue waves at a new location, all you need to do is put your buoy [there] and you can use this [program] without training,” Breunung says.

Harnessing more powerful AI architectures along with more data may yield even higher prediction rates, he adds. “Maybe we can go to 4 out of 4 .… We will see.”