Somewhere in the mid-Pacific, in a region oceanographers had long classified as boringly stable, satellite altimeters recently recorded waves reaching 35 metres from trough to crest. On a radar map, they appeared as streaks and blobs — slightly brighter textures on a grey sheet. Then someone overlaid the scale bar and realised those blobs were higher than almost every building in the nearest island city. The scientists in the darkened lab staring at the data, according to one oceanographer involved in the analysis, forgot to breathe. Not because they did not understand what they were seeing. Because they did. These were not rogue waves of the kind that occasionally ambush ships in the North Atlantic during winter storms. These were organised, long-period swells — broad-shouldered, strangely muscular, moving with a slower rhythm than the models predicted — appearing in a part of the ocean that was supposed to be quiet.
The satellite data is not an anomaly. It is a signal. Across multiple ocean basins, the same pattern is emerging: long-period swells with wave periods of 20 seconds or more are being detected at unexpected locations, arriving from directions that do not correspond to current wind fields, and carrying energy that the standard forecast models — which focus on wave height and wind correlation — are not capturing accurately. The ocean is not behaving within the parameters the industry built its software around. And the industry, in many cases, trusts the software more than the horizon.
I. The Signal the Models Are Missing
Standard ocean wave forecasting focuses on two primary parameters: significant wave height and wind speed and direction. The assumption underpinning most commercial routing software is that dangerous sea states are generated by local wind conditions — that a ship can be routed around a storm by tracking the storm. This assumption works well for the short-period, chaotic waves that storms generate directly. It works less well, and sometimes dangerously, for long-period swells that have travelled thousands of kilometres from their origin storm, have become highly organised and coherent in the process, and arrive at a ship's position from a direction that bears no relationship to the current local wind.
Physical oceanographers have long understood this distinction. When a storm generates waves, the resulting sea is short-crested and irregular, with periods typically between 5 and 12 seconds. As these waves travel away from the storm, a process of natural selection occurs: shorter period waves lose energy faster and dissipate, while longer period waves travel faster and persist. What arrives at a distant location is a swell, a train of smooth, well-defined, long-period waves that has the appearance of calm but carries enormous energy. A vessel that encounters this swell from an unexpected direction, in what the weather routing software classified as moderate conditions, is encountering something its software did not model.
The bigger wave is the one the map does not show. A swell that originated in a North Pacific storm three days ago, with a period of 20 seconds, travels at approximately 60 kilometres per hour. It crosses the Drake Passage. It enters the Atlantic. It arrives at a cargo ship operating in what the routing software shows as a moderate sea state — because the local wind is moderate. The swell is not local. The model did not track it. The ship did not know it was coming.
The record confirms this. In one documented case, a Quirin-class North Atlantic storm generated swells with peak periods of up to 25 seconds that were recorded north of Ireland two days after the storm itself had passed — travelling at speeds that outpaced the storm system and arriving in a region where the local sea state appeared manageable. Severe damage was inflicted in a harbour due to seiche resonance driven by these long-period waves. The harbour masters did not see a storm approaching. They saw calm weather. The damage came from energy that had been generated far away and was invisible to their local instruments.
II. Parametric Rolling and the Container Loss Crisis
The connection between long-period swells and the container shipping industry's most dangerous and least understood risk is direct and documented. Parametric rolling is a resonance phenomenon that occurs when a vessel encounters waves — particularly long-period swells — whose encounter period is close to half the ship's natural roll period. When this resonance condition is met, periodic changes in the vessel's stability as it moves through the waves can trigger sudden, extremely rapid increases in roll angle that overwhelm the lashing systems securing deck cargo.
What makes parametric rolling particularly treacherous is its invisibility in standard weather reports. It does not require a storm. It does not require large local waves. A cargo ship can be operating in conditions that any weather forecast would classify as moderate — a sea state of 3 or 4, reasonable visibility, wind speeds well within operating parameters — and suddenly begin rolling with accelerations that exceed the design loads of its container lashing systems. The rolling can develop from nothing to catastrophic within minutes. By the time the crew recognises what is happening, the moment for routine corrective action has already passed.
In one frequently cited incident, a 13,000 TEU container ship lost multiple containers overboard. A maritime authority investigation concluded that the most likely cause of lashing equipment failure was parametric rolling. The sea state at the time of the incident did not, on paper, suggest exceptional risk. The long-period swell that triggered the resonance was not visible in the standard routing forecast. The containers were in the ocean before the crew fully understood what had happened.
The TopTier project, launched by the shipping industry in May 2021 following a series of catastrophic container losses, investigated parametric rolling systematically and reached a conclusion that the industry found uncomfortable: parametric rolling in following seas is especially hazardous, and it is a phenomenon that is not well known among many vessel crews and routing officers. The IMO's second-generation intact stability criteria are designed to address it. Not all vessels have been assessed against them. Not all routing software integrates wave period data with the same weight it gives to wave height. The gap between what the science understands and what the software models remains significant.
III. The Climate Connection and the Shifting Baseline
The appearance of 35-metre wave events in ocean zones previously considered stable is not occurring in isolation from the broader climate signal. Stronger winds over larger stretches of open water push more energy into swells. Shifting storm tracks create new fetch geometries — the distances over which wind blows continuously to generate waves — that the historical baseline of ocean behaviour did not include. Marine heatwaves are restructuring the thermal architecture of ocean basins in ways that alter circulation patterns and, consequently, the pathways through which swell energy travels.
A 2026 study published in Nature Communications demonstrated that ocean dynamics significantly promote marine heatwave intensity and duration in mid-to-high latitude oceans. The Atlantic overturning circulation, in particular, was shown to make North Atlantic marine heatwaves potentially predictable over several years — but also to modulate them in ways that the current generation of operational forecast models does not fully capture. The ocean is not a passive recipient of atmospheric forcing. It is an active participant in the energy system, with its own internal dynamics that operate on timescales from hours to decades.
What the satellite data is revealing, and what the buoy records are confirming, is that the statistical baseline for what constitutes a normal sea state is shifting. Wave events that were previously classified as once-in-decades occurrences are appearing with greater frequency in some regions. The software that routing officers use was calibrated to a historical distribution of wave heights and periods. If that distribution is changing, the software is providing guidance based on a world that no longer exists at the extremes that matter most for safety.
The satellite does not tame the ocean. It lets us see its mood swings from a safer distance. The mistake is thinking that better maps mean total control. The ocean still writes the final line. From space, you see a wave field that looks almost mathematical. Then you remember each spike is a moving wall of water that could flip a ship.
IV. What the Shipping Industry Must Do Differently
The responses available to the shipping industry are known, documented, and in some cases already being implemented. The question is the pace and comprehensiveness of adoption relative to the speed at which the physical environment is changing.
Wave period must be integrated into routing decisions with the same weight as wave height. The current industry practice of optimising routes primarily around wind and wave height forecasts is insufficient for managing parametric rolling risk. ClassNK, ABS, and Britannia P&I Club have all published detailed guidance on calculating vessel-specific parametric rolling risk based on wave encounter period, vessel speed, heading, and the ship's natural roll period. This guidance exists. It is not universally applied in real-time routing decisions.
Wave radar technology, which can directly measure the direction, period, height, and wavelength of approaching swells, is not yet commonly installed as a standard preventive measure against parametric rolling on most commercial vessels. ABS is actively developing PROLL, an advanced software tool aimed at predicting the onset of parametric rolling in real time. The IMO, beginning in 2026, is making vessel masters responsible for mandatory reporting of container losses and containers sighted floating in the water. The regulatory framework is tightening. The physical environment is not waiting for it.
The deeper problem is epistemological. The shipping industry has optimised its decision-making around the outputs of models that were built on a historical distribution of ocean conditions. When that distribution shifts at the extremes — when the waves that occur once in a generation start occurring once in a decade, when swell energy from distant storms arrives at unexpected angles in regions that forecast software classified as safe — the industry's confidence in its models becomes a risk factor rather than a risk management tool. The crews that recognised the warning signs of parametric rolling in time to take corrective action were, in many documented cases, the ones who trusted the horizon more than the forecast. That is not a coincidence.
The ocean is not malfunctioning. It is operating according to physics that have always been true but that the historical record of the past century did not capture at full scale. The waves that satellite altimeters are now measuring at 35 metres in previously stable zones have always been possible. What has changed is the combination of conditions that makes them more likely, more frequent, and more geographically distributed than the models assumed. The shipping industry carries approximately 80 percent of global trade by volume. Its routing decisions are made on software calibrated to a world that the ocean is in the process of revising. The predictability gap between what our models assume and what the water is doing is not a scientific curiosity. It is a supply chain risk, an insurance liability, a maritime safety failure in waiting, and a signal that the Earth's thermal balance is moving in ways that our digital infrastructure has not yet fully learned to read. The sea has always been the final authority. It remains so. The question is whether we are listening carefully enough, and with the right instruments.