New insights using the Oceananigans.jl modeling framework

Reporting by Helen Hill for MITgcm

Understanding how sea ice reshapes upper‑ocean energetics is central to improving polar ocean simulations. In their Journal of Geophysical Research: Oceans study, Gupta, Thompson, and Klein (2026) use a high‑resolution, floe‑resolving coupled ocean–sea‑ice model to examine how energy is both dissipated and generated beneath sea ice. Their work challenges the prevailing view of sea ice as a simple frictional lid suppressing eddy activity, instead revealing a dynamic interplay between surface damping and subsurface baroclinic production.

A defining feature of their study is its use of Oceananigans.jl, a GPU‑accelerated, finite‑volume ocean model written in Julia. Oceananigans and MITgcm share numerical DNA, with Oceananigans solving the three‑dimensional Boussinesq equations using a numerical scheme closely based on the MITgcm finite‑volume core. In this case, the flexible, high-performance, and GPU-ready architecture of Oceananigans allows the authors to simulate a 128 km regional domain at ~500 m resolution with vertically stretched grids and WENO (weighted essentially non-oscillatory) advection, thereby resolving mesoscale and submesoscale features essential to ice–ocean coupling.

The physical picture emerging from Gupta et al’s simulations is compelling. Sea ice does indeed damp surface eddy kinetic energy (EKE) through enhanced frictional dissipation at the ice–ocean interface. But at the same time, the drifting ice modifies mesoscale circulation patterns, producing spatially variable Ekman pumping and suction. These vertical buoyancy fluxes inject available potential energy into the interior of eddies, giving rise to baroclinic energy production beneath the mixed layer. Remarkably, this subsurface production compensates between 30% and 69% of the surface frictional losses across a wide range of ice concentrations, from 30% to solid‑ice cover.  In effect, sea ice shifts the location of energetic activity rather than simply eliminating it, redistributing energy vertically in ways not accounted for in most coarse‑resolution climate models.

The Gupta study further identifies the role of heterogeneous sea‑ice melt, concentrated in the upper ~5 m of the ocean. Patchy melt generates sharp buoyancy gradients that spawn mixed‑layer eddies typical of the marginal ice zone (MIZ). These fine‑scale features amplify baroclinic production in the shallowest layers, adding an additional pathway for energetic activity beneath sea ice. Such processes: vertical buoyancy fluxes in mesoscale eddies, mixed‑layer eddy generation from melt patterns, and upper‑ocean baroclinic adjustment, all demand the kind of high‑resolution modeling made possible by Oceananigans and, increasingly, by GPU‑accelerated MITgcm builds.

Another central innovation lies in the explicit resolution of individual sea‑ice floes using a discrete‑element model (DEM). Rather than representing sea ice as a continuum – standard in most large‑scale simulations – Gupta et al treat each floe as a rigid disk subject to collisions, drag forces, and melt. This floe‑scale detail produces emergent behaviors such as anticyclonic floe rotation, clustering around eddies, and size‑dependent melt rates. These features significantly shape both frictional damping and baroclinic production, revealing how sensitive upper‑ocean energetics are to floe geometry and distribution. Because continuum rheologies cannot easily reproduce such interactions, the authors emphasize the need for new parameterizations in coarse models.

Story image: Sea Ice, NASA 

About the Researcher:

Mukund Gupta is an Assistant Professor at the Delft University of Technology (TU Delft) in the Department of Geoscience and Remote Sensing. His research lies at the intersection between sea ice, ocean, and climate physics. He has been using MITgcm since 2016.

This Month’s Featured Publication

 

Other New Publications last month

Chen, Xi et al (2026), Dispersal and isolation of the scaly-foot snail across abyssal insular habitats and through time, Current Biology, doi: 10.1016/j.cub.2026.01.033

Dunshaw, Brian D. Dushaw et al (2026), Ocean Observing System Design: Transatlantic Acoustic Propagation for Acoustic Thermometry, Journal of Atmospheric and Oceanic Technology, doi: 10.1175/JTECH-D-24-0124.1

Li, Bingtian et al (2026), The response of East China Sea internal tides to ocean warming: intensified while maintaining seasonal contrasts, Ocean Modelling, doi: 10.1016/j.ocemod.2026.102696

Li, C., Dang, H., Du, J., Li, H., Zhang, J., & Jian, Z. (2026), Orbital variability in grazing proportion: New insights from sedimentary amino acid δ15N records of the western equatorial Pacific, AGU Advances, doi: 10.1029/2025AV002024

Li, Zhiyuan (2025), Understanding the Double-ITCZ Bias in Climate Models: From Ocean-Atmosphere Interactions to Future Climate Projections, Yale University ProQuest Dissertations & Theses,  2025, 32242374

Nakajima, R., Nomaki, H., Osafune, S. et al. (2026), High surface microplastic abundance at 30°S, 90°W supports eastward extension of the South Pacific garbage patch, Micropl. &Nanopl., doi: 10.1186/s43591-026-00179-4

Nakayama, Y., Hyogo, S., Lin, Y., Park, T., Lee, J., Caillet, J., Mohan, G., Poinelli, M., Dutrieux, P., Nakata, K., Zhang, H., Loose, B., and Kowalski, L.(2026), Development of ECCO-downscaled Amundsen-Bellingshausen Sea regional simulation using MITgcm(66j), EGUsphere [preprint], doi: 10.5194/egusphere-2025-5958

Ran, Junlin et al (2026), Research Progress of Deep Learning in Sea Ice Prediction, Remote Sensing, doi: 10.3390/rs18030419

Shu, Ruiqi et al (2026), HybridOM: Hybrid Physics-Based and Data-Driven Global Ocean Modeling with Efficient Spatial Downscaling, arXiv: 2602.00598

Tessier, Jonathan et al (2026), A Zonally-Averaged Model of the Meridional Overturning Circulation, Journal of Physical Oceanography, doi: 10.1175/JPO-D-25-0197.1

Yu, Kai et al (2026), The Seasonality of the Submesoscale SST Variability Over the Kuroshio-Oyashio Extension, JGR Oceans, doi: 10.1029/2025JC023305

Zahn, M. J., Fournier, S., Fenty, I. G., Steele, M., Wood, M., & Gaube, P. (2026), Mackenzie River freshwater controls early sea ice formation in the eastern Beaufort Sea, Geophysical Research Letters, doi: 10.1029/2025GL118871

Zhang, L., Song, Y., Huang, W. et al. (2026), Interaction between atmospheric rivers and marine heatwaves in the North Pacific, npj Clim Atmos Sci, doi: 10.1038/s41612-026-01350-7

Zhao, Xiaoyu and Yanjiang Lin (2026), SWOT Observations of Bimodal Seasonal Submesoscale Processes in the Kuroshio Large Meander, Remote Sensing, doi: 10.3390/rs18030384

—
Do you have news about research using MITgcm? We are looking for contributions to these pages. If you have an interesting MITgcm project (ocean, atmosphere, sea-ice, physics, biology or otherwise) that you want to tell people about, get in touch. To make a post, contact Helen