Science · 6 min read
What Antarctic Ice Cores Show About Carbon Dioxide and Temperature
What ancient air bubbles and ice isotopes establish about climate coupling, leads and lags, and the limits of paleoclimate inference.

What Antarctic Ice Cores Show About Carbon Dioxide and Temperature
Antarctic ice contains two related archives. The ice itself records aspects of past snowfall and temperature through its isotopic composition. Bubbles sealed within it preserve samples of ancient air, including carbon dioxide and methane. Together, these measurements show repeated glacial cycles in which Antarctic temperature and atmospheric carbon dioxide changed in close association. They also show that the modern carbon dioxide concentration lies well above the range directly observed in ice spanning hundreds of thousands of years.
Those findings are strong. A simpler claim—that an ice core supplies a perfectly dated global thermometer, or that carbon dioxide initiated every ancient warming—is not. Ice and enclosed gas have different ages at the same depth. Isotopes require calibration and model interpretation. The timing varies by event and location. Researchers use these records to test mechanisms, not to read a single unambiguous causal sequence from a graph.
Snow becomes two clocks in one core
On the Antarctic plateau, snowfall accumulates and compresses. Snow crystals become granular firn and then solid ice. Air continues to diffuse through pores in the firn until those pores close, trapping bubbles. The enclosed air is therefore younger than the ice around it: the snow fell before the air was sealed. The size of this gas-age/ice-age difference depends on accumulation rate, temperature, firn thickness, and the width of the sealing interval.
Scientists date cores by combining methods. Annual chemical or isotopic layers can sometimes be counted in younger, higher-accumulation ice. Volcanic ash and sulfate peaks can be matched among cores or to independently dated eruptions. Models of ice flow and snow accumulation extend chronology where layers thin at depth. Gas records can be aligned using globally mixed methane or oxygen-isotope signals. Each method contributes uncertainty, and published chronologies report it.
Temperature is inferred rather than measured by a fossil thermometer. Water molecules containing heavier isotopes—deuterium or oxygen-18—behave differently during evaporation and condensation. In present Antarctic precipitation, isotope ratios correlate with temperature, but the ancient relationship can also reflect moisture source, transport path, elevation, seasonality, and changes in the ice sheet. Borehole temperatures and spatial comparisons help constrain the calibration.
The European Project for Ice Coring in Antarctica, or EPICA, Dome C record extended direct greenhouse-gas measurements through roughly 800,000 years. The underlying air samples provide atmospheric concentrations, while isotope measurements provide a regional temperature proxy. Calling both lines “ice-core temperature data” obscures their different measurement chains.
The long record shows coupling, not a one-way switch
Across glacial–interglacial cycles, Antarctic temperature proxies, carbon dioxide, and methane broadly rise and fall together. The EPICA collaboration reported carbon dioxide values roughly between 172 and 300 parts per million over the 800,000-year record available at the time. Direct atmospheric monitoring at Mauna Loa has since measured annual means above 420 parts per million. Ice cores and modern instruments have different resolution and sampling characteristics, but their overlap can be checked; the modern rise is not an artifact of splicing unrelated scales.
Orbital variations alter the seasonal and geographic distribution of sunlight, pacing glacial cycles. They do not by themselves explain the full global temperature change. Ice sheets, vegetation, dust, oceans, sea ice, and greenhouse gases respond and feed back. When warming oceans and changing circulation release carbon dioxide, the additional gas reduces outgoing infrared radiation and amplifies warming. During cooling, uptake and circulation changes can work in the other direction.
The Intergovernmental Panel on Climate Change’s assessment of paleoclimate archives treats the temperature–carbon relationship as a feedback within a coupled system. That is compatible with Antarctic warming beginning before a detectable local carbon dioxide rise in some terminations. A feedback need not be the first perturbation in order to influence the eventual magnitude and global pattern of change.
Leads and lags depend on the question asked
Early studies of the Vostok core estimated that Antarctic temperature began rising centuries before carbon dioxide during some deglaciations. These results were sometimes presented as if they ruled out a carbon dioxide effect. The inference exceeded the evidence. Gas-age/ice-age uncertainty was comparable to some proposed lags, and Antarctic temperature is not global mean temperature.
Improved chronologies have refined the picture without producing one universal number. A 2013 analysis led by Frédéric Parrenin used nitrogen isotopes and other measurements to reduce dating uncertainty at EPICA Dome C. It found that Antarctic temperature and carbon dioxide were closely coupled during the last deglaciation, with no significant multicentury lag at the start within reported uncertainty. Other work finds regional and phase-dependent differences.
A separate 2013 reconstruction led by Jeremy Shakun compared global temperature proxies with carbon dioxide during the last deglaciation. It found that global mean warming lagged the initial Antarctic warming and was more closely aligned with rising carbon dioxide. The proposed sequence begins with orbital forcing and Northern Hemisphere ice-sheet changes, proceeds through ocean circulation and Southern Hemisphere warming, releases carbon, and then includes greenhouse amplification. This is a model supported by multiple observations, not a directly filmed chain of events.
The distinction among Antarctic local temperature, hemispheric temperature, and global mean is load-bearing. A gas mixed through the atmosphere can influence global radiation while a regional proxy responds early to ocean-circulation changes. Asking “which came first?” without specifying place, variable, event, and chronological uncertainty produces an answer too crude for the data.
Ancient change and the modern rise are related but not identical
Ice cores establish that carbon dioxide participated in past climate changes. They are not the only evidence that current emissions cause the modern increase. Carbon accounting shows human sources from fossil fuels and land use. Atmospheric oxygen declines as fossil carbon burns. The carbon-isotope composition of atmospheric carbon dioxide has shifted in the direction expected from adding old plant-derived carbon. Oceans have gained carbon and become more acidic. Together these independent observations attribute the rise primarily to human activity.
The rate and initiating cause distinguish the modern event from a glacial termination. Orbital forcing did not trigger the industrial release of fossil carbon. Humans are injecting carbon first, and temperature is responding alongside other forced changes. Ancient records help estimate sensitivity and identify feedbacks, but they are not exact analogues: continental ice configuration, aerosols, vegetation, and the speed of forcing differ.
Nor does an 800,000-year record define all possible natural states. Older carbon dioxide levels are reconstructed through other proxies with larger and different uncertainties. The Beyond EPICA consortium recovered a 2.8-kilometer core expected to span at least 1.2 million years. At a scientific meeting in May 2026, the team presented preliminary measurements extending the continuous climate and atmospheric record through that interval. A contemporary report in Nature noted that the findings had not yet completed peer review and that much of the core remained to be analyzed. The age, gas chronology, and integrity of its oldest sections must be established through those measurements, not assumed from depth alone.
The boundary of the evidence
Antarctic cores directly preserve ancient air and physically measured ice chemistry. They establish that greenhouse gases and Antarctic climate varied together over repeated cycles, that feedbacks were important, and that recent carbon dioxide exceeds the long ice-core range. Multiple cores, laboratories, dating tools, and modern overlap strengthen those conclusions.
They do not yield an uncertainty-free calendar, a single global temperature series, or one lag applicable to every transition. The precise carbon release pathways during each deglaciation, the behavior of deep-ocean reservoirs, and the size of some slow feedbacks remain active research questions. Better chronologies can narrow these questions but cannot turn a coupled climate system into a one-cause sequence.
The defensible conclusion stops there: ice cores are unusually direct archives of past atmosphere and powerful, calibrated proxies for regional climate. Their strongest lesson is not that one line invariably moves first. It is that carbon, ice, ocean circulation, and temperature interact—and that claims about direction and timing must match the resolution of the record.
Sources
5 sources, every one with a link you can follow.
- EPICA Community Members, 800,000-year Antarctic greenhouse-gas record.
- Frédéric Parrenin and colleagues, synchronous change during the last deglaciation.
- Jeremy Shakun and colleagues, global temperature and carbon dioxide during deglaciation.
- IPCC, assessment of paleoclimate archives and uncertainties.
- Nature, report on Beyond EPICA’s preliminary 1.2-million-year record.
