Articles | Volume 21, issue 3
https://doi.org/10.5194/os-21-1065-2025
https://doi.org/10.5194/os-21-1065-2025
Research article
 | Highlight paper
 | 
20 Jun 2025
Research article | Highlight paper |  | 20 Jun 2025

Long-term prediction of the Gulf Stream meander using OceanNet: a principled neural-operator-based digital twin

Michael Gray, Ashesh Chattopadhyay, Tianning Wu, Anna Lowe, and Ruoying He

Model code and software

OceanNet Training and Prediction Michael A. Gray and Ashesh Chattopadhyay https://doi.org/10.5281/zenodo.15675792

Download
Co-editor-in-chief
The use of neural networks to emulate the dynamics of the Earth system (ocean, atmosphere) is of critical importance since it enables efficient and operational forecast, data assimilation, and affordable uncertainty quantification. This field has been very productive in numerical weather forecasts but relatively fewer results were demonstrated in the oceans. This paper could be a reference for mesoscale ocean dynamics emulation.
Short summary
The Gulf Stream is a prominent oceanic feature in the northwestern Atlantic Ocean that influences weather patterns in the Northern Hemisphere and is notoriously difficult to predict. We present a machine learning model, OceanNet, to predict the position of the Gulf Stream months in advance. OceanNet is able to perform a 120 d prediction 4 000 000 times faster than traditional methods of ocean modeling with great accuracy.
Share