The paradox at the heart of the study is a matter of feedbacks. Earlier modelling efforts typically ran ice sheet models offline, forcing them with climate fields but never letting the growing ice sheet talk back to the climate system. When the team ran their ice sheet model in this traditional one-way mode, with carbon dioxide set at six times the preindustrial level, the simulated ice remained stubbornly confined to polar latitudes above 60 degrees south, clashing badly with the geological record of glacial deposits. Lowering carbon dioxide far enough to grow a realistic ice sheet would have pushed global mean surface temperature down to around 8 degrees Celsius, far below the roughly 15 degrees suggested by reconstructions. Something crucial was missing from the one-way framework.

That missing ingredient was two-way coupling. The researchers linked the Community Earth System Model, a fully coupled atmosphere-ocean-land-sea ice model, with the Ice-sheet and Sea-level System Model developed by NASA’s Jet Propulsion Laboratory and the University of California, Irvine. Because the two models operate on wildly different timescales, they used an asynchronous coupling strategy: the climate model ran for ten simulated years, then handed monthly temperature and precipitation fields to the ice sheet model, which integrated forward for 2,500 years before returning updated ice geometry, elevation, and surface type. Eighty such cycles spanned 200,000 years of ice sheet evolution, long enough for the system to reach equilibrium, defined by stabilizing top-of-atmosphere radiation, global temperature, and ice extent.

The coupled simulation transformed the outcome. Instead of a modest polar cap, the ice sheet grew outward from a mountain range at Gondwana’s southern tip, expanding steadily toward roughly 40 degrees south and nearly blanketing the entire southern portion of the supercontinent. The total ice volume, about 9.0 times ten to the sixteenth cubic meters, corresponds to a sea level drop of approximately 210 meters, comparable to that during the Last Glacial Maximum and consistent with geological reconstructions. The simulated ice margin broadly matches the distribution of ancient glacial tillites, supporting the idea that a single, massive ice sheet, rather than disconnected smaller ones, covered the mid- to high-latitude supercontinent during the Hirnantian glaciation.

What drove this dramatic expansion? The team identified two intertwined positive feedbacks. The first is the familiar ice-albedo feedback: as ice spread, land surface albedo rose from 0.43 to 0.55, reflecting more sunlight, cooling the surface, and promoting further ice accumulation. Global mean surface temperature fell sharply from about 18.5 to 17.1 degrees Celsius within the first 30,000 years. But a control experiment in which surface type changes were disabled revealed something unexpected: substantial ice growth continued even without the albedo feedback, pointing to a second, less obvious mechanism rooted in ice sheet topography.

That mechanism involves katabatic winds, the gravity-driven torrents of cold, dense air that drain downslope from elevated ice surfaces, a phenomenon familiar today from Antarctica and Greenland, where such winds can exceed 50 meters per second. As the Ordovician ice sheet thickened, its towering dome chilled and densified the air above it, sending frigid winds cascading toward the ice margins. In the ice-free control climate, warm air had flowed poleward over parts of the continent during summer, suppressing ice growth. Once the ice sheet formed, the circulation reorganized fundamentally: cold air blew toward the continental interior in both winter and summer, expanding the sub-zero region, boosting winter and spring snow accumulation, and slashing summer ablation at the ice edges. The researchers distilled this into a self-reinforcing loop: ice sheet growth generates katabatic winds, which cool the margins, which enable further growth.

The feedback could not run forever. As the ice front pushed toward 42 degrees south, the katabatic winds remained potent in austral winter but faded in summer, when solar heating warmed the mid-latitude ice surface and eroded the temperature inversion that drives downslope drainage. Intense mid-latitude summer melt ultimately stabilized the ice margin, a negative feedback that halted expansion. Precipitation, by contrast, was never the limiting factor; moisture transport from the mid-latitude oceans remained robust throughout, in sharp contrast to situations such as the Tibetan Plateau during the Last Glacial Maximum, where moisture starvation curtailed glacier growth.

Then came the surprise. When the team examined the equilibrium climate with the giant ice sheet in place, they found that surface temperatures rose in many regions, including tropical lands, the global ocean, and even a small area near the South Pole, with local warming exceeding 4 degrees Celsius in places. The culprit was a stationary atmospheric wave. The elevated ice sheet acted as a mechanical barrier to the westerly winds, strengthening a wavenumber-1 stationary wave between 30 and 60 degrees south and shifting it westward by about 30 degrees of longitude. This redirected warm, low-latitude air poleward over the ocean, transporting air with temperature anomalies of 3 to 4 degrees Celsius across 65 degrees south. Reduced cloud cover over the Southern Ocean amplified the effect by letting in more solar radiation. In the Northern Hemisphere, a weakened meridional overturning circulation, triggered by the Southern Ocean warming, shifted heat transport northward across the equator, melting sea ice and warming the polar north.

The global mean surface temperature drop attributable to ice sheet onset was about 1.4 degrees Celsius, with strong cooling over mid- to high-latitude continents but warming nearly everywhere else, and only a muted response in the tropical oceans. This finding carries real consequences for how scientists read the geological record. Tropical sea surface temperature reconstructions, such as those derived from conodont oxygen isotopes, may simply not register major temperature shifts happening at high latitudes. The team’s isotope calculations further suggest that large positive oxygen-isotope excursions in low-latitude fossils may reflect the global ice-volume effect on seawater composition rather than equivalent tropical cooling. The simulations even hint that continental ice sheets may have existed as early as the Middle Ordovician, some 467 million years ago, without leaving a strong tropical temperature fingerprint, potentially resolving tensions between short-lived and long-lived glaciation scenarios.

The study’s broader message is methodological as much as scientific. Offline ice sheet models, however convenient, cannot capture the cryosphere’s feedback on atmospheric circulation, and may badly misjudge how ice ages begin under warm, high-carbon-dioxide climates. Coupled simulations, the authors argue, are essential for a quantitative understanding of ice sheet-climate interactions, whether applied to the Ordovician or to the future of Greenland and Antarctica. Limitations remain, including simplified orbital forcing, the omission of dust, a simple melt parameterization, and no accounting for sea level drop or isostatic adjustment. But the core lesson stands: a growing ice sheet does not simply chill the planet uniformly. It reshapes winds, waves, and ocean currents in ways that can warm the very seas surrounding the ice, a counterintuitive reminder that Earth’s climate system is a web of competing feedbacks rather than a simple thermostat.

Subject of Research: Ice sheet-climate feedbacks during the Late Ordovician glaciation and their counterintuitive ocean warming effect

Article Title: Ocean warming caused by Late Ordovician glacial onset in a coupled climate-ice sheet simulation

Article References: Sun, Y., Liu, Y., Wu, J., Man, K., Yu, H., Yuan, S., Liu, Y., Liu, Y., Cui, Q., Wei, Q., & Hu, Y. (2026). Ocean warming caused by Late Ordovician glacial onset in a coupled climate-ice sheet simulation. Climate of the Past, 22(10), 1781-1802. https://doi.org/10.5194/cp-22-1781-2026

Image Credits: AI Generated

DOI: 10.5194/cp-22-1781-2026

Keywords: Late Ordovician, glaciation, ice sheet, climate modeling, katabatic winds, Gondwana, ocean warming, ice-albedo feedback, stationary waves, mass extinction, paleoclimate, coupled simulation