Estimating the AMOC from Argo profiles with machine learning trained on ocean simulations

Wölker, Yannick; Rath, Willi; Renz, Matthias; Biastoch, Arne

doi:10.5194/os-21-3541-2025

Articles | Volume 21, issue 6

https://doi.org/10.5194/os-21-3541-2025

© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/os-21-3541-2025

© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 21, issue 6

Research article

| Highlight paper

|

18 Dec 2025

Research article | Highlight paper |

| 18 Dec 2025

Estimating the AMOC from Argo profiles with machine learning trained on ocean simulations

Yannick Wölker, Willi Rath, Matthias Renz, and Arne Biastoch

Data sets

Estimating AMOC from Argo Profiles with Machine Learning Trained on Ocean Simulations [Dataset] Yannick Wölker et al. https://doi.org/10.5281/zenodo.15643439

Estimating AMOC from Argo Profiles with Machine Learning Trained on Ocean Simulations [Experiment Results] Yannick Wölker https://doi.org/10.5281/zenodo.15644013

Model code and software

ML-based AMOC Reconstruction [python package] Yannick Wölker https://doi.org/10.5281/zenodo.15643942

Interactive computing environment

Estimating AMOC from Argo Profiles with Machine Learning Trained on Ocean Simulations [Reproduction Notebooks] Yannick Wölker https://doi.org/10.5281/zenodo.15644138

Co-editor-in-chief

This is an important contribution to the science surrounding the Atlantic Meridional Overturning Circulation (known as AMOC), in particular with regard to observations and methods. This work may have strategic relevance for planning observations in the future. It also represents an interesting effort at incorporating artificial intelligence (AI) technology into obtaining optimal results from observational data analysis.

Short summary

The Atlantic Meridional Overturning Circulation (AMOC) is a large current system that helps regulating Earth's climate. Monitoring the AMOC relies on fixed instruments anchored to the seafloor. This study explores, in a high-resolution model, whether data from Argo floats, autonomous drifters collecting hydrographic profiles, can be used to monitor the AMOC cost-effectively with the help of Machine Learning. Results suggest that Argo floats can extend AMOC monitoring beyond current fixed arrays.