Hidden Ocean Currents Revealed in Stunning Detail by AI

Researchers built GOFLOW, a deep-learning system that converts frequent weather-satellite thermal images into detailed maps of ocean surface currents.

Trained and tested on ocean circulation simulations and GOES-East observations, it can resolve fine-scale dynamics across large areas, including the Gulf Stream.

The approach needs no new hardware and could greatly improve ocean monitoring for climate science, nutrient transport, carbon storage, and emergency response.

The work was led by scientists at UC San Diego and UCLA and published in Nature Geoscience.