Njord: A Probabilistic Graph Neural Network for Ensemble Ocean Forecasting
Daniel Holmberg, Joel Oskarsson, Erik Wikingsson · 2026
Read Njord: A Probabilistic Graph Neural Network for Ensemble Ocean Forecasting by Daniel Holmberg, Joel Oskarsson, Erik Wikingsson (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-15470.txt.
Description
Daniel Holmberg, Joel Oskarsson, Erik Wikingsson, Fredrik Lindsten, and Teemu Roos present Njord, a sea forecast model that samples full sets of forecasts. Old machine-learning sea models give one fixed forecast. Njord pairs a latent-variable frame with a graph net. It uses cluster meshes that fit jagged coasts. Run worldwide and for the Baltic Sea, it beat fixed machine-learning rivals. It gave the lowest mean errors on upper-sea variables in OceanBench tests. Its biggest gains were in surface temperature.
Catalog record
| CREATOR | Daniel Holmberg, Joel Oskarsson, Erik Wikingsson |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 104647 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.15470