Weather satellites carry a hidden record of the ocean's smaller currents

In 2023, oceanographer Luc Lenain was looking at thermal images of the North Atlantic from a weather satellite. According to Scripps' account, the evolving warm and cool patterns made the Gulf Stream's motion visible to his eye. The research question was how to convert that impression into a measurement. The resulting GOFLOW method appeared in Nature Geoscience on 13 April 2026.[2]

The instrument was already observing the right kind of change. NOAA's Advanced Baseline Imager includes an infrared channel centred at 11.2 micrometres, with a best spatial resolution of two kilometres. This is a thermal view, available beyond daylight, rather than a photograph of objects floating on the sea. Temperature patterns provide features whose movement can be examined.[4]

Learn how a temperature pattern deforms

GOFLOW uses three snapshots one hour apart to estimate velocity at the middle time. A U-Net learns from a high-resolution ocean simulation that supplies both thermal patterns and known currents. Feeding it the logarithm of temperature-gradient strength makes weak boundaries more visible, giving it a richer pattern to follow than a few sharply defined fronts.[1]

The researchers compared estimates with independent drifting-buoy and shipboard measurements across several seasons. The reported root-mean-square error was 0.27 metres per second. Explained variance was 0.69 for east–west velocity and 0.73 for north–south velocity. These are checks against observations in the study region, rather than an accuracy guarantee for every ocean.[1]

The buoys supply an instructive complement. NOAA calculates currents from their changing positions. A submerged sea anchor, centred about 15 metres below the surface, helps each buoy travel with the water; losing it makes wind and waves more influential. Such details matter because an instrument's trajectory only represents the intended current while its physical coupling to the water remains appropriate.[3]

A satellite sequence can cover a broad area repeatedly. A drifter follows a particular path through it. Joining those perspectives creates a check on whether an inferred field of motion agrees with something transported by the water itself.[3]

Resolution has more than one meaning

The two-kilometre imagery does not imply independent two-kilometre current detail. The paper finds a sharp loss of variance near ten-kilometre scales, probably its effective resolution and a limit inherited from training. It also reproduces asymmetric patterns of rotation and convergence that coarser current estimates miss. Training used one year from a limited Atlantic region; worldwide application needs further work.[1]

Scripps' explanation connects these motions to vertical mixing: bringing nutrients upwards and carrying heat and carbon between shallower and deeper water. Being able to examine their signatures with observations gives oceanographers a way to test relationships previously explored mainly in simulations. That is a scientific use already distinct from a promise to improve tomorrow's forecast.[2]

Clouds remain a physical interruption. NOAA's standard sea-surface temperature products require cloud-free pixels because the instrument must see the water. Repeated images offer fresh viewing opportunities, but an opaque cloud still blocks the surface beneath it. A more detailed computational map earns its value through the measurements that constrain it.[5]

Report an error or suggest a correction