Antarctic ice-shelf melt is a major source of uncertainty in future sea-level rise projections. Current climate models often fail to capture the interplay between melt and ocean circulation.

There´s a new study uses a circumpolar Antarctic ocean–sea-ice model with interactive ice shelves to examine how externally forced changes (e.g., from atmospheric warming) and melt-driven feedback influence the continental-shelf response.

The melt feedback is of comparable importance to the direct forced response — and in some regions, it has the opposite sign. The positive melt feedback accounts for about two-thirds of the increased melt rate across all ice shelves in the simulations.

The research was led by University of Maryland’s Madeleine Youngs, with collaborators including Andrew L. Stewart and others. It used the MITgcm (Massachusetts Institute of Technology General Circulation Model) configured with interactive ice shelves.

_____________________________________________________________________________________

Antarctic ice-shelf basal melt shaped by competing feedbacks

The paper “Antarctic ice-shelf basal melt shaped by competing feedbacks” by Madeleine K. Youngs and colleagues was published in Nature Geoscience on May 15, 2026.

This study uses a circumpolar Antarctic ocean–sea-ice model (MITgcm configuration with interactive ice shelves) to separate the direct forced response to future climate change (e.g., atmospheric/ocean warming from CMIP6-style boundary conditions) from the melt-driven feedbacks caused by the freshwater released from increased basal melting itself.

Previous global climate models and many projections (including those feeding into IPCC assessments) often treat ice-shelf basal melt as a somewhat passive or parameterized response to external forcing, without fully capturing the two-way coupling where meltwater alters ocean density, stratification, circulation, and thus further melting.

The melt feedback is of comparable magnitude to the direct forced response — and in some regions has the opposite sign. Across all ice shelves in the simulations, the positive melt feedback accounts for about two-thirds of the total increase in melt rate.

MITgcm (Massachusetts Institute of Technology General Circulation Model) supports interactive ice shelves primarily through the pkg/shelfice (SHELFICE) package, with optional extensions like remeshing and coupling to ice dynamics.

Key Modeling Techniques and Features

Grid Configurations:

• Global: Latitude-longitude (e.g., 4° × 4° tutorial with 15 levels) or cube-sphere for even resolution without polar singularities.
• Regional: Nesting or curvilinear grids; open boundary conditions (pkg/obcs) for limited domains.
• High-resolution: Eddy-permitting to eddy-resolving (1–4 km common near shelves/coasts); partial cells (“shaved cells”) for realistic topography.

Subgrid Parameterizations:

• Vertical mixing: KPP (K-Profile Parameterization), convective adjustment, or GM/Redi for mesoscale eddies.
• Horizontal dissipation: Laplacian or biharmonic viscosity/diffusion.
• Boundary layer processes: Critical for ice-ocean interfaces (e.g., in pkg/shelfice).

Forcing and Boundary Conditions:

• Surface: Wind stress, heat/freshwater fluxes, bulk formulae, or prescribed SST/SSS relaxation.
• Open boundaries: Radiation, Orlanski, or prescribed values (time-varying possible).
• Tides, rivers, and atmospheric coupling via ESMF or other frameworks.

Tracers and Biogeochemistry:

• Passive and active tracers (temperature, salinity, age, biogeochemical species via pkg/dic, etc.).
• Offline tracer mode for efficient transport studies. mitgcm.readthedocs.io

Ice-Ocean Interactions (pkg/shelfice):

• Thermodynamic basal melt/freezing with three-equation or ISOMIP parameterizations.
• Interactive/remeshing for evolving ice draft.
• Freshwater, heat, and salt fluxes into the ocean. Used in the Youngs et al. (2026) circumpolar Antarctic study.

Advanced Capabilities:

• Adjoint modeling: Automatic differentiation for sensitivity analysis, state estimation, and optimization (e.g., with DART for ensemble Kalman filtering). youtube.com
• Coupling: Sea ice (pkg/seaice), dynamic ice sheets, atmosphere, and biogeochemistry.
• Non-hydrostatic: Essential for convection, overflows, and small-scale turbulence. mitgcm.readthedocs.io
• Parallelization: MPI, OpenMP; highly scalable on HPC systems.

MITgcm includes a rich set of verification and tutorial experiments:

• Barotropic/Baroclinic Ocean Gyre: Basic wind-driven circulation.
• Global Ocean (lat-lon or pressure coords): Realistic bathymetry, accelerated time-stepping to equilibrium.
• Southern Ocean Reentrant Channel: Idealized ACC-like flows.
• Convection and Gravity Plumes: Over topography.
• Ice Shelf and Biogeochemistry setups.

For the Youngs et al. Antarctic work, configurations are available on Zenodo, typically using high-resolution circumpolar domains with interactive shelves and CMIP6-derived boundaries.

MITgcm excels in process-oriented studies (like meltwater feedbacks), global state estimation, and high-resolution regional modeling. Its open nature and adjoint make it powerful for uncertainty quantification and prediction.

Thinking Critically

Strengths: High-resolution enough for circumpolar coverage with interactive ice shelves; explicit decomposition of forced vs. feedback components; data/code availability for reproducibility (Zenodo links for configurations and analysis scripts). It builds on established MITgcm shelfice capabilities and prior process studies.

Limitations (as the authors note it’s “a first investigation”):

• Resolution still insufficient for smallest-scale cavity processes, grounding-zone dynamics, or full eddy/turbulence details in all areas.
• Relies on CMIP6-derived boundary conditions — inherits uncertainties from those models.
• Does not yet fully couple to dynamic ice-sheet geometry changes (e.g., grounding-line retreat, ice-shelf thinning altering cavity shape), which could introduce additional feedbacks (positive or negative).
• Short simulation timescales relative to centennial projections; tipping-point behavior may emerge later.
• Melt parameterizations (three-equation type) carry inherent uncertainties, as seen in broader ice-shelf–ocean intercomparison projects (ISOMIP+, etc.).

This aligns with other recent work showing meltwater-induced changes in the Antarctic Slope Current, undercurrents, and cross-slope heat transport. It does not suggest runaway collapse on its own, but it indicates that omitting these feedbacks likely leads to underestimation of melt in some key sectors and overall uncertainty in sea-level projections.

Implications

Sea-level rise projections: Current IPCC-style estimates may be conservative in regions dominated by the positive feedback. The study emphasizes the need for better-coupled high-resolution models.

Policy/urgency: Reinforces that Antarctic contributions to sea-level rise are highly sensitive to ice–ocean interactions, not just atmospheric warming.

Next steps (per the team): Higher-resolution simulations to identify which ice shelves are closest to tipping thresholds.

This paper is a solid advance in process understanding rather than a dramatic new prediction of doom.

It highlights model blind spots without overturning the fundamental physics: ocean-driven basal melting remains the dominant threat to West Antarctic ice shelves, and feedbacks can materially alter the rate and pattern of loss.

Published:  Nature Geoscience

DOI: 10.1038/s41561-026-01975-6

Provided: University of Maryland

Authors: Madeleine K. Youngs, 
Andrew L. Stewart, 
Yidongfang Si, 
Andrew F. Thompson & 
Michael P. Schodlok 

Abstract

Antarctic ice-shelf melt constitutes a major source of epistemic uncertainty in future sea level rise, yet the interplay between melt and ocean circulation is not captured in climate model projections. Consequently, the relative importance of feedbacks from increased ice-shelf melting versus directly forced ocean change is poorly understood. Here we examine how externally forced changes and melt-driven feedbacks each influence the continental-shelf response to future climate change, using a circumpolar Antarctic ocean–sea-ice model that incorporates interactive ice shelves. These simulations show that the melt feedback is of comparable importance and in some regions opposite in sign to the forced response. In dense shelf regions, warming creates a feedback loop: as the high-salinity shelf water becomes lighter, it lets warmer ocean water flow underneath the ice shelf, which then increases ice-shelf melting. Westward transport of meltwater from these dense shelf regions freshens the continental shelf further downstream and obstructs warm water intrusions, establishing a negative feedback. The positive melt feedback accounts for two-thirds of the increased melt rate over all ice shelves. This work highlights the importance of representing Antarctic ice-shelf melt feedbacks to predict future climate.