AI shows monster wave risk

Under certain conditions, isolated monster waves can form in the sea, threatening ships and oil platforms. RugliG/iStock

When and where could the dreaded “Caventsmen” strike? Using extensive wave data and artificial intelligence, researchers have developed a formula that can show the respective threat risk from extreme waves at sea. The scientists say that the calculations could therefore benefit route planning in shipping.

For a long time they were considered sailor's tales: stories of gigantic waves on the high seas that seem to come out of nowhere and then crush ships. But it is now clear that the monster waves, also known as Kaventsmann, actually threaten shipping and oil platforms. It is becoming apparent that many, sometimes mysterious, accidents at sea can be traced back to this force of nature. One case was particularly precisely documented in 1995, when a 26-meter-high monster wave broke over the Norwegian oil platform Draupner. Various methods of ocean monitoring have now shown that water walls actually build up quite frequently in various ocean regions around the world.

Monster wave dynamics in sight

Researchers have already developed models to describe the factors that cause these extreme waves. In contrast to tsunamis, which are caused by geological effects, the monster waves “rock” through certain superposition patterns. However, there are still uncertainties and predicting regional monster wave risks is difficult. Scientists led by Dion Häfner from the Niels Bohr Institute at the University of Copenhagen have now taken on the challenge of modeling again. “Basically, if one of these waves hits, it’s just really bad luck. They are caused by a combination of many factors that have not yet been summarized in a single risk assessment,” says Häfner.

In order to characterize the phenomenon more precisely, he and his colleagues have now intensively used the aid of artificial intelligence for the first time. “In the study, we systematically recorded the causal variables that generate monster waves. “We then used artificial intelligence to transfer this information into a model that can calculate the probability of its creation,” says Häfner. Their results are based on extensive data on ocean movements under specific conditions as well as water depths and other oceanographic information. The scientists then combined this with sea state information from measuring buoys in 158 different ocean areas. Taken together, this information contained feature data on more than a billion waves, the researchers write.

Not only the extreme specimens, some of which were over 20 meters high, were classified as monster waves: the team defined mountains of water that were at least twice as high as the surrounding waves as “abnormal”, i.e. a wave with a monster character. “Our analysis showed that such abnormal waves occur frequently. In fact, in our dataset we registered 100,000 waves that can be defined as monster waves. “But these are not necessarily specimens of extreme size,” explains co-author Johannes Gemmrich from the Canadian University of Victoria.

Potential for route planning in shipping

The team then used various artificial intelligence or machine learning methods to process and evaluate the data. The highlight is that these methods can detect connections in the data that are difficult to recognize. In the current case, between certain constellations in the sea and the probability of the formation of extreme waves. The AI ​​approach has now led to a formula that can be used for description and prediction. In concrete terms, this means: Based on available data about ocean movements, this algorithm can predict the risk of being hit by a monster wave in an area of ​​the ocean under certain conditions.

The results also demonstrated that the dominant factor in the creation of extreme waves is a special form of superposition. Two wave systems cross each other, whereby they reinforce each other for a short time. “If two wave systems meet at sea in such a way that the chance of generating high wave crests and then deep wave troughs increases, there is a high risk of monster waves,” says Häfner. The algorithm can now show when this combination of factors exists that increases the risk of monster waves.

According to the researchers, this could benefit shipping safety: “When shipping companies plan their routes, they could use our algorithm to obtain a risk assessment about the extent to which ships are at risk of dangerous waves on the way. “On this basis, they can then select alternative routes,” says Häfner, summing up the possible application potential of the study results.

Source: University of Copenhagen, specialist article: Proceedings of the National Academy of Sciences, doi: 10.1073/pnas.2306275120

Recent Articles

Related Stories