Skip to content
DraconiticResearch
Search Sources API
201,818 studies31 sourcesUpdated 51 min ago
Peer-reviewedCC BYGeophysical Research Letters

FloeNet: A Mass‐Conserving Global Sea Ice Emulator That Generalizes Across Climates

  • William King Gregory1
  • Mitchell Bushuk2
  • James Duncan3
  • Elynn Wu3
  • Adam Subel4
  • Spencer Koncius Clark3,2
  • William J. Hurlin2
  • Oliver Watt‐Meyer3
  • Alistair Adcroft1
  • Christopher S. Bretherton3
  • Laure Zanna4
4 affiliations
  1. 1Atmospheric and Oceanic Sciences Program Princeton University Princeton NJ USA
  2. 2Geophysical Fluid Dynamics Laboratory NOAA Princeton NJ USA
  3. 3Allen Institute for Artificial Intelligence (Ai2) Seattle WA USA
  4. 4Courant Institute School of Mathematics, Computing, and Data Science New York University New York NY USA
Open at sourceDOICite

Abstract

We introduce FloeNet, a graph neural network trained to emulate the Geophysical Fluid Dynamics Laboratory global sea ice model, SIS2. FloeNet is a mass‐conserving model, emulating 6‐hr mass and area budget tendencies related to sea ice and snow‐on‐sea‐ice growth, melt, and advection. We train FloeNet using SIS2 data from a reanalysis‐forced ice‐ocean simulation and test its ability to generalize to pre‐industrial and 1% CO 2 forcing conditions. FloeNet outperforms a non‐conservative model at reproducing sea ice and snow‐on‐sea‐ice mean state, trends, and inter‐annual variability, with volume anomaly correlations above 0.96 in the Antarctic and 0.76 in the Arctic, across all forcings. FloeNet also produces the correct thermodynamic‐dynamic response to forcing, enabling physical interpretability of emulator output. Finally, we show that FloeNet outputs high‐fidelity coupling‐related variables, including ice‐surface temperature, ice‐to‐ocean salt flux, and melting energy fluxes. We hypothesize that FloeNet will improve polar climate processes within existing atmosphere and ocean emulators.

Related

  1. Extreme events and statistics in an AI emulator of a global storm-resolving model
    Anna Kwa, W. Andre Perkins, Troy Arcomano +7
    arXivPreprintCC BY 4.0Full textAtmospheric science & meteorologyarXiv
  2. HClimRep-Ocean: A Global Ocean Emulator on an Unstructured Mesh
    Kacper Nowak, Aleksei Koldunov, Nikolay Koldunov +7
    arXivPreprintarXiv licenceFull textAtmospheric science & meteorologyarXiv
  3. Matching Wind-Solar Portfolios with Electricity Consumption Across Climates: A Physical Framework to Minimize Residual Demand
    Justine Duval, Fabien Perdu, Arthur Clerjon +1
    HALPreprintPublisher termsFull textEnvironmental science & pollutionHAL
  4. Ocean-Aware Sea-ice Emulator for Future Hybrid Coupled Prediction Systems
    Charlotte Durand, Pierre Rampal, Laurent Brodeau
    HALPreprintCC BY 4.0Full textOceanography & cryosphereHAL
  5. The diversity of planetary climates in the Solar System as a basis for reflection and work on the study of exoplanet climates
    Olivier Planchon
    HALPreprintCC BY 4.0Full textPlanetary science & exoplanetsGeneral earth scienceHAL

Cite

BibTeXRIS
BibTeX
@article{gregory2026floenet,
  title = {{FloeNet: A Mass‐Conserving Global Sea Ice Emulator That Generalizes Across Climates}},
  author = {William King Gregory and Mitchell Bushuk and James Duncan and Elynn Wu and Adam Subel and Spencer Koncius Clark and William J. Hurlin and Oliver Watt‐Meyer and Alistair Adcroft and Christopher S. Bretherton and Laure Zanna},
  journal = {Geophysical Research Letters},
  year = {2026},
  month = jul,
  doi = {10.1029/2026gl122981},
  url = {https://doi.org/10.1029/2026gl122981},
  note = {CC BY},
}
RIS
TY  - JOUR
TI  - FloeNet: A Mass‐Conserving Global Sea Ice Emulator That Generalizes Across Climates
AU  - William King Gregory
AU  - Mitchell Bushuk
AU  - James Duncan
AU  - Elynn Wu
AU  - Adam Subel
AU  - Spencer Koncius Clark
AU  - William J. Hurlin
AU  - Oliver Watt‐Meyer
AU  - Alistair Adcroft
AU  - Christopher S. Bretherton
AU  - Laure Zanna
PY  - 2026
DA  - 2026/07/12
JO  - Geophysical Research Letters
DO  - 10.1029/2026gl122981
UR  - https://doi.org/10.1029/2026gl122981
AB  - We introduce FloeNet, a graph neural network trained to emulate the Geophysical Fluid Dynamics Laboratory global sea ice model, SIS2. FloeNet is a mass‐conserving model, emulating 6‐hr mass and area budget tendencies related to sea ice and snow‐on‐sea‐ice growth, melt, and advection. We train FloeNet using SIS2 data from a reanalysis‐forced ice‐ocean simulation and test its ability to generalize to pre‐industrial and 1% CO 2 forcing conditions. FloeNet outperforms a non‐conservative model at reproducing sea ice and snow‐on‐sea‐ice mean state, trends, and inter‐annual variability, with volume anomaly correlations above 0.96 in the Antarctic and 0.76 in the Arctic, across all forcings. FloeNet also produces the correct thermodynamic‐dynamic response to forcing, enabling physical interpretability of emulator output. Finally, we show that FloeNet outputs high‐fidelity coupling‐related variables, including ice‐surface temperature, ice‐to‐ocean salt flux, and melting energy fluxes. We hypothesize that FloeNet will improve polar climate processes within existing atmosphere and ocean emulators.
SN  - 0094-8276
SN  - 1944-8007
ER  -