Ocean Science
Ocean Science
OS
Articles | Volume 21, issue 1
Articles | Volume 21, issue 1
https://doi.org/10.5194/os-21-113-2025
© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.
Special issue:
https://doi.org/10.5194/os-21-113-2025
© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.
Articles | Volume 21, issue 1
Research article
 | 
24 Jan 2025
Research article |  | 24 Jan 2025

Convolutional neural networks for sea surface data assimilation in operational ocean models: test case in the Gulf of Mexico

Olmo Zavala-Romero, Alexandra Bozec, Eric P. Chassignet, and Jose R. Miranda

Related authors

The role of cyclonic eddies in the detachment and separation of Loop Current eddies
Marco Larrañaga, Julien Jouanno, Eric P. Chassignet, Giovanni Durante, Ilkyeong Ma, Julio Sheinbaum, and Lionel Renault
EGUsphere, https://doi.org/10.5194/egusphere-2025-5574,https://doi.org/10.5194/egusphere-2025-5574, 2025
This preprint is open for discussion and under review for Ocean Science (OS).
Short summary
Simulated and Observed Transport Estimates Across the Overturning in the Subpolar North Atlantic Program (OSNAP) Section
Gokhan Danabasoglu, Frederic S. Castruccio, Burcu Boza, Alice M. Barthel, Arne Biastoch, Adam Blaker, Alexandra Bozec, Diego Bruciaferri, Frank O. Bryan, Eric P. Chassignet, Yao Fu, Ian Grooms, Catherine Guiavarc'h, Hakase Hayashida, Andrew McC. Hogg, Ryan M. Holmes, Doroteaciro Iovino, Andrew E. Kiss, M. Susan Lozier, Gustavo Marques, Alex Megann, Franziska U. Schwarzkopf, Dave Storkey, Luke van Roekel, Jon Wolfe, Xiaobiao Xu, and Rong Zhang
EGUsphere, https://doi.org/10.5194/egusphere-2025-5406,https://doi.org/10.5194/egusphere-2025-5406, 2025
This preprint is open for discussion and under review for Geoscientific Model Development (GMD).
Short summary
Improving accuracy and providing uncertainty estimations: ensemble methodologies for ocean forecasting
Ibrahim Hoteit, Eric Chassignet, and Mike Bell
State Planet, 5-opsr, 21, https://doi.org/10.5194/sp-5-opsr-21-2025,https://doi.org/10.5194/sp-5-opsr-21-2025, 2025
Short summary
Ocean forecasting at the regional scale: actual status
Marina Tonani, Eric Chassignet, Mauro Cirano, Yasumasa Miyazawa, and Begoña Pérez Gómez
State Planet, 5-opsr, 3, https://doi.org/10.5194/sp-5-opsr-3-2025,https://doi.org/10.5194/sp-5-opsr-3-2025, 2025
Short summary
Core services: an introduction to global ocean forecasting
Yann Drillet, Matthew Martin, Yosuke Fujii, Eric Chassignet, and Stefania Ciliberti
State Planet, 5-opsr, 2, https://doi.org/10.5194/sp-5-opsr-2-2025,https://doi.org/10.5194/sp-5-opsr-2-2025, 2025
Short summary
More articles (3)

Cited articles

Agrawal, S., Barrington, L., Bromberg, C., Burge, J., Gazen, C., and Hickey, J.: Machine learning for precipitation nowcasting from radar images, arXiv [preprint], https://doi.org/10.48550/arXiv.1912.12132, 11 December 2019. a
Beauchamp, M., Bocquet, M., and Fablet, R.: Multimodal 4DVarNets for the reconstruction of sea surface dynamics from SST-SSH synergies, arXiv [preprint], https://doi.org/10.48550/arXiv.2203.06003, 11 March 2022. a
Bleck, R.: An oceanic general circulation model framed in hybrid isopycnic-Cartesian coordinates, Ocean Model., 4, 55–88, 2002. a, b
Boukabara, S.-A., Krasnopolsky, V., Stewart, J. Q., Maddy, E. S., Shahroudi, N., and Hoffman, R. N.: Leveraging modern artificial intelligence for remote sensing and NWP: Benefits and challenges, B. Am. Meteor. Soc., 100, ES473–ES491, 2019. a
Bozec, A., Chassignet, E. P., and Srinivasan, A.: GOMb0.04 Reanalysis for the Gulf of Mexico, https://www.hycom.org/data/gomb0pt04/gom-reanalysis (last access: 21 January 2025), 2025. a
More articles (30)
Download
Short summary
This study shows AI can speed up data assimilation in ocean models. Researchers used convolutional neural networks (CNNs) to assimilate sea surface temperature and height observations in the Gulf of Mexico, learning to replicate corrections made by traditional, computationally expensive methods. CNN design and training window size significantly impacted accuracy, but the percentage of ocean pixels did not. These findings suggest CNNs may accelerate data assimilation in realistic settings.
Read more
Share
Special issue
Special Issue for the 54th International Liège Colloquium...