OCEAN ICE Publication: Freshwater Sources in the Global Ocean Through Salinity-δ18O Relationships: A Machine Learning Solution to a Water Mass Problem

Understanding where freshwater in the ocean comes from — and how it moves — is crucial for predicting changes in ocean circulation and climate.
In our new paper, Xabier, Elaine et al. used a machine learning approach to trace freshwater sources from precipitation, ice sheets, rivers, and melting sea ice across the global ocean.

🔑 Key findings:
In the Southern Ocean, reduced sea ice formation is making deep waters fresher.
In the Arctic, surface waters are becoming fresher due to increased sea ice melt.
The Northern and Southern Hemispheres have very different freshwater budgets, mainly because sea ice plays contrasting roles in forming deep waters.
These results help us better understand how the ocean’s freshwater system is shifting in response to climate change.

The paper shows that some of the deepest water masses in Antarctica are getting fresher, not because ice sheet melt, but rather because there is less sea ice being formed. Similar but different, in Arctic, waters are getting fresher because there is more sea ice melt.

How do we do this? We follow a machine learning approach to find relationships between salinity and stable oxygen isotopes that allows us to separate sea ice meltwater from precipitation, riverine input and ice sheet melt.

🙏 Huge thanks to Xabier Davila,

📖 Read the full paper here: Freshwater Sources in the Global Ocean Through Salinity‐δ18O Relationships: A Machine Learning Solution to a Water Mass Problem - Davila - 2025 - Journal of Geophysical Research: Oceans - Wiley Online Library

The author of the article - Michael Meredith (British Antarctic Survey) and Xabier Davila (NORCE)