Antarctic Snowfall’s Hidden History: Winds and Meltwater Explain Why Models Overpredict a Warming-Driven Gain

For decades, climate scientists have expected a simple thing to happen on Antarctica as the planet warms: more snow. Warmer air holds more moisture, and that moisture should fall on the vast ice sheet as snow, locking away water that would otherwise raise global sea level. Ice core reconstructions confirm this happened during the twentieth century, when increased snow accumulation offset sea level rise by roughly 11 millimeters. Yet in the most recent four decades of intense observation and rapid warming, the continent-wide accumulation rate has barely budged. A new study published in Earth System Dynamics by David P. Schneider of Phare Manchot LLC and the University of Colorado Boulder and colleagues untangles this apparent contradiction, and its answer carries a warning for every projection of future sea level.

The team’s central finding is that greenhouse gases really are driving an increase in Antarctic snowfall, just as thermodynamics predicts. But the observed increase is only about half of what rising greenhouse gas concentrations alone would have produced. According to the ice core reconstruction used in the study, snow accumulation on the grounded Antarctic Ice Sheet gained a cumulative mass equivalent to 10.5 millimeters of sea level mitigation between 1901 and 2000. In the Community Earth System Model version 2, or CESM2, running with all historical forcings, the ensemble-mean gain was 16.8 millimeters. When the model ran with greenhouse gases as the only evolving forcing, the gain ballooned to 21 millimeters, twice the reconstructed value. Something in the real climate system has been holding the warming-driven snowfall surge in check.

To identify that something, the researchers deployed an unusually broad suite of experiments. They used the 100-member CESM2 Large Ensemble, a single-forcing large ensemble in which greenhouse gases, industrial aerosols, biomass burning aerosols, and other forcings evolve separately, and a tropical Pacific pacemaker experiment called TPACE that nudges the model’s tropical sea surface temperatures toward observations. They added a wind-nudging experiment constrained by the ERA5 reanalysis across the middle and high southern latitudes, prescribed sea surface temperature experiments in the AMIP style, and a CESM1 meltwater experiment that pours anomalous freshwater from thinning ice shelves into the Southern Ocean. Each experiment isolates a different piece of the puzzle, allowing the team to separate forced signals from internal variability with a rigor that multi-model comparisons rarely achieve.

The single-forcing experiments revealed two opposing giants. Greenhouse gases dominate the accumulation increase, while industrial aerosols, which cool the planet, offset it by roughly 9 millimeters of sea level equivalent. The sum of the two, about 12 millimeters, sits within the reconstruction’s error range. But the spatial patterns told a deeper story. The reconstruction shows a characteristic dipole across West Antarctica, with more accumulation on the Antarctic Peninsula side and less on the Ross Ice Shelf side, alongside a deepening of the Amundsen Sea Low and a strengthening and poleward shift of the Southern Hemisphere westerly winds. These wind patterns redistribute snow in a predictable orographic fashion, piling it up on windward coasts such as Queen Maud Land and the Peninsula while starving the lee sides, including Wilkes Land in East Antarctica.

Crucially, winds alone cannot explain the suppressed continent-wide accumulation. When the researchers removed the wind-congruent portion of the reconstructed trends, a large residual warming-driven accumulation increase emerged, closer to what the model’s forced response predicts. The missing ingredient was temperature, specifically the temperature of the ocean surface. In the reconstructions, the Pacific sector of the Southern Ocean, including the Amundsen Sea, has cooled or warmed only weakly, and this region supplies much of the moisture that falls as snow on West Antarctica. The free-running model, by contrast, warms the Southern Ocean too much, and its winds are only weakly linked to sea surface temperature anomalies. This decoupling between winds and ocean temperatures is precisely where the model and the real world part ways.

The study points to two culprits for that decoupling. The first is meltwater. Ice shelf thinning and glacier retreat in West Antarctica began in the mid-twentieth century, decades before the satellite era, and the resulting freshwater flux into the Southern Ocean has been sustained by northerly wind anomalies in the Amundsen Sea Embayment. Meltwater is not included as a forcing in standard CMIP6 historical experiments, yet a growing body of work argues it should be. In the CESM1 meltwater experiment, prescribing ice shelf basal melt cools the Southern Ocean surface and suppresses the accumulation rate, producing a pattern that matches the observed one remarkably well. When the researchers added the meltwater-induced sea surface temperature pattern to the greenhouse-gas-only response, the combined pattern correlated at 0.91 with the wind-corrected reconstruction over the southern mid-latitudes.

The second culprit involves a reverse teleconnection, a signal traveling from the poles toward the tropics rather than the other way around. Cooling that originates in the Amundsen Sea can propagate toward the equatorial Pacific through cold-air advection, wind-evaporation feedbacks, and subtropical cloud feedbacks, reinforcing a La Niña-like pattern of sea surface temperatures. That pattern, in turn, sends Rossby waves poleward that deepen the Amundsen Sea Low, closing a self-reinforcing loop that also promotes further ice shelf thinning. The tropical pacemaker experiment, which nudges the model toward observed tropical Pacific temperatures, dampened the accumulation rate and brought the cumulative mass gain to 11.5 millimeters, within the reconstruction’s error range, even though it worsened the fit to the spatial circulation patterns. This suggests high-latitude cooling signals have genuinely reached the tropics, something the free-running model fails to reproduce.

The implications for the recent past are striking. When the researchers prescribed observed sea surface temperatures and sea ice in an atmosphere-only configuration, the model showed no trend in Antarctic-wide accumulation since 1979, matching reanalyses and regional climate models. The free-running coupled model, in contrast, simulated a steady accumulation increase throughout the period, physically consistent with its too-strong ocean warming and sea ice loss. A wind-nudged CESM2 experiment for 2001 through 2022 captured the observed pressure and accumulation patterns, including drying over Wilkes Land and increased accumulation on the Peninsula and at the South Pole, confirming that circulation shapes where snow falls while ocean temperature controls how much falls overall.

Looking forward, the study offers a sobering recalibration. Using a conservative lower bound that accounts for the omitted meltwater forcing, the team estimates that increased snowfall could mitigate sea level rise at an average rate of about 0.5 millimeters per year through 2050, less than standard large ensemble projections suggest and less than the ice sheet’s current dynamic contribution to sea level rise. If meltwater continues to dampen surface warming, snow accumulation will struggle to keep pace with ocean-driven ice loss, and continued warming will bring more surface melt, rainfall, and atmospheric river events that deliver heat rather than lasting mass gain. The authors caution that their conclusions rest on a single model at relatively coarse resolution, and they call for coordinated multi-model efforts, higher-resolution simulations, and more ice core observations, particularly on the East Antarctic plateau.

What makes this study resonate beyond Antarctica is its broader message about climate model evaluation. The same Southern Ocean cooling and eastern Pacific warming-pattern discrepancies that have puzzled scientists since 1979 appear here as the very mechanisms that suppress Antarctic snowfall. Snow accumulation, the authors argue, is not merely a term in the sea level budget; it is a gauge of whether a model can reproduce the magnitude and spatial pattern of climate change itself. Any projection that ignores the interplay of winds, meltwater, and ocean temperature patterns, they conclude, is likely overconfident, and the Antarctic snow archive now offers a century-long benchmark against which the next generation of Earth system models can be tested.

Subject of Research: Attribution of twentieth-century Antarctic snow accumulation trends to greenhouse gas forcing, westerly winds, and ice shelf meltwater

Article Title: Decoded Antarctic snow accumulation history reconciles observed and modeled trends in accumulation and large-scale warming patterns

Article References: Schneider, D. P., Yin, Z., Blanchard-Wrigglesworth, E., Datta, R. T., & Espinosa, Z. I. (2026). Decoded Antarctic snow accumulation history reconciles observed and modeled trends in accumulation and large-scale warming patterns. Earth System Dynamics, 17(5), 1395-1433. https://doi.org/10.5194/esd-17-1395-2026

Image Credits: AI Generated

DOI: 10.5194/esd-17-1395-2026

Keywords: Antarctic Ice Sheet, snow accumulation, sea level rise, greenhouse gases, Southern Ocean, westerly winds, Amundsen Sea Low, ice shelf meltwater, CESM2, paleoclimate data assimilation, climate models, surface mass balance