Beneath hundreds of meters of Antarctic ice, in a place no human has ever visited, warm ocean water is quietly eating away at one of the largest frozen reservoirs on Earth. New high-resolution simulations published in Nature Communications in early 2026 reveal that narrow channels carved into the underside of East Antarctic ice shelves are trapping warm deep water and focusing it like a blowtorch against the ice. The result: basal melt rates 42 to 50 percent higher than previous estimates, according to companion research in Nature Climate Change. Standard climate models, which lack the resolution to see these channels, have been systematically underestimating how fast the ice is disappearing from below.
East Antarctica’s ice sheet is the largest on the planet, holding enough frozen water to raise global sea levels by roughly 52 meters if it all melted. Nobody expects that to happen anytime soon. But even the marine-based portions of the ice sheet, the sectors most exposed to ocean warming, contain enough ice to push seas up by several meters over centuries. The discovery that models have been missing a key mechanism of ice loss in this region changes the math for coastal cities from Miami to Mumbai.
What the new research actually found
The breakthrough centers on Fimbulisen, a cold-cavity ice shelf in East Antarctica’s Dronning Maud Land. Using ice-ocean models run at unusually fine resolution, researchers showed that the shelf’s underside is not smooth. It is scored with channels, some narrow enough that coarser models treat them as flat terrain. When warm Circumpolar Deep Water, a dense, salty current that circles Antarctica at depth, intrudes beneath the shelf, these channels act as conduits. They trap the warm water, concentrate its heat, and accelerate melting along narrow bands rather than spreading it evenly across the shelf’s base.
The physics is intuitive once you picture it. A smooth, flat ice base would allow warm water to spread thin and lose heat gradually. But a channelized base funnels the water, keeping it in contact with the ice longer and at higher temperatures. As the channel deepens through melting, it alters local water flow in ways that draw in still more warm water, creating a feedback loop. The channel’s width, depth, and curvature all determine how efficiently heat is delivered to the ice and how quickly the channel itself grows.
This pattern is not unique to Fimbulisen. At the Totten Ice Shelf, one of East Antarctica’s largest and most vulnerable, oceanographers have documented warm modified Circumpolar Deep Water entering the cavity through a deep submarine trough. Bathymetric gateways, essentially valleys and depressions carved into the continental shelf during past ice ages, act as highways for warm water. The water follows these pathways at depth, reaches the grounding zone where ice meets bedrock, and erodes the shelf from underneath.
The forces pushing warm water onto the continental shelf are themselves well understood. Shifts in subpolar westerly winds and reductions in sea-ice cover have repositioned oceanographic fronts, allowing warmer deep water to access ice-shelf cavities that were previously insulated. This means atmospheric changes occurring hundreds of kilometers offshore can dictate how much heat reaches the base of an ice shelf. Wind patterns, sea-ice extent, and deep-water intrusion form a chain that connects the open Southern Ocean to the hidden underside of the ice.
Taken together, these studies establish several points with high confidence. Warm Circumpolar Deep Water can and does reach East Antarctic ice-shelf cavities through specific seafloor pathways. Once inside, it is steered and concentrated by basal channels that focus melt in narrow bands. And high-resolution models that explicitly resolve those channels consistently produce higher local melt rates than the coarser models used in global climate projections, which means those projections are likely missing important physics at the ice-ocean interface.
Where the picture gets murkier
Not every corner of East Antarctica tells the same story. Float-based observations near the Denman and Shackleton ice shelves, reported in Science Advances, suggest that warm water at depth is reaching cavities beneath both shelves and driving melt along the ice-shelf system. If confirmed at scale, that finding would extend the channelized-melt pattern across a much wider stretch of coastline, implying that the aggregate impact on East Antarctic ice loss could be substantially larger than current assessments assume.
But a separate modeling study in Communications Earth & Environment complicates that conclusion. That analysis found that unresolved seafloor roughness in the Denman region can actually suppress the impact of warm water on melting. Small-scale bumps and ridges on the ocean floor scatter and mix the intruding water, reducing the direct heat flux to the ice base. In other words, coarse models may overestimate melt in some locations even as they underestimate it in others. The contrast between float-based evidence of strong melt and model-based suggestions of a muted response shows just how sensitive results are to the shape of the ocean floor.
That sensitivity points to a central problem: the seafloor beneath most East Antarctic ice shelves has never been mapped at high resolution. Without accurate bathymetry, models cannot reliably simulate how warm water moves through submarine channels and across rough terrain. The result is a patchwork. Detailed understanding exists at a few well-studied sites like Totten and Fimbulisen, surrounded by vast stretches of coastline where the geometry of the ocean floor is essentially unknown. In those unmapped regions, even the direction of future change, whether basal melt will accelerate or hold steady, is difficult to predict.
Another gap sits between the new science and the models that inform policy. As of June 2026, no published comparison exists between the newly measured channel melt rates and the specific climate-model ensembles used in the most recent IPCC assessment of East Antarctic ice loss. If those models systematically miss channel-scale melting, their sea-level forecasts for the coming decades could be too conservative. Researchers have not yet announced timelines for integrating the new channel-melt findings into the next generation of coupled ice-ocean simulations, leaving open the question of how quickly the science will feed into updated projections.
Observational gaps compound the uncertainty. Long-term, in-place measurements of ocean temperature and water velocity inside the newly modeled basal channels at Fimbulisen do not yet exist. The modeling results are consistent with satellite observations and short-duration field campaigns, but sustained monitoring would be needed to confirm whether the simulated melt rates hold up over years and across seasonal cycles. Deploying instruments beneath hundreds of meters of ice in one of the most remote regions on Earth remains a formidable logistical and financial challenge.
What this means for sea-level planning
The strongest evidence in this body of research comes from peer-reviewed modeling and observational studies published in Nature Communications, Nature Climate Change, Nature Geoscience, Science Advances, and Communications Earth & Environment. These papers converge on several robust conclusions: warm deep water is reaching East Antarctic ice shelves through seafloor gateways, basal channels focus that heat and amplify local melting, and the coarse climate models that underpin global sea-level projections generally fail to resolve these processes. Where independent lines of evidence, including satellite altimetry, autonomous floats, and high-resolution simulations, point in the same direction, confidence in the underlying mechanisms is strong.
Important caveats remain. The lack of comprehensive seafloor mapping and long-term under-ice observations means that regional melt rates are still subject to revision. Conflicting interpretations around the Denman region highlight how sensitive conclusions can be to assumptions about unresolved topography. Current estimates of East Antarctica’s contribution to sea-level rise should be treated as provisional, especially where they rely on models that do not yet incorporate channel-scale melting.
For coastal planners and policymakers, the practical signal is not that East Antarctic collapse is around the corner. It is that the risk envelope has shifted. The range of plausible sea-level outcomes now skews higher than older models suggested, and as process studies refine our understanding of how warm water interacts with basal channels, projections are likely to move further in that direction. Planning that can absorb a wider range of outcomes, rather than banking on best-case scenarios drawn from models that cannot yet see what is happening in those hidden channels beneath the ice, is no longer a precaution. It is a baseline requirement.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
