Ice cores are treated as sealed archives of ancient air, and most of the shallow blue ice at the Larsen blue-ice area held more carbon dioxide than standard records, with concentrations falling with depth until they levelled near 4.60 metres — the authors argue photochemical reactions may contribute to the shallow gas anomalies, with the clearest evidence confined to the top one to two metres

Blue ice looks like a shortcut to old air. At the Larsen blue-ice area in northern Victoria Land, East Antarctica, glacial ice that would otherwise sit deep in a borehole is already at the surface, stripped of snow by wind and sublimation, and it is blue because the ice itself is what you are looking at.

That exposure is also the problem. A team led by Giyoon Lee and Jinho Ahn at Seoul National University reports in Communications Earth & Environment that greenhouse gases trapped in the shallowest metres of that ice do not match the records taken as unaltered atmosphere. Carbon dioxide concentrations fall with depth and then level off near 4.60 metres. Methane rises from the surface to somewhere between 0.35 and 1.15 metres, then declines until it too stabilises at 4.60 metres. Most of the Larsen samples are enriched in both gases relative to the EPICA Dome C and WAIS Divide cores of matching age. Two cores, LS306 and LS300, also show depleted methane in the top metre.

Photochemical reactions, they write, may contribute to the anomalies. The findings, they say, suggest that photochemical processes can modify trapped greenhouse gases in shallow exposed blue ice. The clearest isotopic fingerprint they can point to, a depletion in the oxygen-isotope ratio of trapped O2, sits only in the upper one to two metres. Deeper than that, they say they cannot pin the remaining offsets on sunlight alone.

Core counts, depths and gases

Fifty-eight shallow cores, two to ten metres long, were drilled in the austral summer of 2018/19 along a one-kilometre transect, 20 to 30 metres apart. The site sits at 74.93 degrees south, 161.60 degrees east, at about 1,080 metres above sea level. Mean annual temperature there is given as minus 24.4 degrees Celsius, with a one-sigma spread of 11.7 degrees. Dust bands in the ice dip at roughly one to six degrees. The ice at the surface spans the Last Glacial Maximum through the mid-Holocene.

The results section says the greenhouse-gas analysis used eleven of those cores. The methods section names seven cores for the Seoul gas-chromatograph measurements (300, 111, 119, 122, 125, 206 and 200) and a separate set for the Japanese mass-spectrometer work. A supplementary comparison of the two laboratories is in the paper. Ages were taken from a 2022 chronology by the same first author, tied to the WAIS Divide 2014 timescale at 1.95 metres depth and supported by krypton-81 dates, then interpolated and extrapolated along the local dip of the layers.

Carbon dioxide was extracted dry, with a needle crusher, and measured against NOAA standards on the WMOX2019 scale. Methane was released by a melt-refreeze method and measured against a 740.6 parts-per-billion standard on the WMOX2004A scale. Control-group scatter was 0.6 plus or minus 0.6 parts per million for carbon dioxide and 3.3 plus or minus 1.4 parts per billion for methane.

Total air content in the samples ran 0.07 to 0.10 cubic centimetres per gram, the range typical of Antarctic ice. The authors take that as evidence that melting had not produced a considerable reduction in trapped air.

Melt, modern air, dust and microbes

Katabatic winds keep snow off blue ice, and melting at the surface would need near-freezing air and low wind. Daily average temperatures at Larsen Glacier, from the Italian Antarctic meteorological record, peaked at minus 6 degrees Celsius in 2015 to 2018. Bubble-free blue bands, the usual mark of melt and refreeze, were not seen in the samples. Surface melt at the original snowfall site, farther inland and colder, is treated as unlikely for the same reason.

Cracks could let modern air in. Modern air has a nitrogen-isotope ratio of zero; ice that closed off under a firn column typically sits between 0.2 and 0.5 per mil because heavier nitrogen settles. The difference between samples from 0.10 to 1.65 metres and unaltered ice at 1.95 metres of the same gas age is 0.02 plus or minus 0.02 per mil. From that the authors estimate that only 5.2 plus or minus 6.5 percent of the trapped air is modern. That is equivalent to 14.6 plus or minus 17.6 parts per million of extra carbon dioxide and 43.5 plus or minus 56.6 parts per billion of extra methane, one-sigma on each figure. Those amounts, they write, are too small to explain the excess at the surface.

Dust from modern aerosols could, in principle, drive chemistry. Non-sea-salt calcium and sodium, used as proxies for terrestrial dust and sea salt, stay relatively flat with depth inside each core. Lead-isotope ratios that line up with Taylor Dome and Dome C records were recovered from some inner-core samples shallower than two metres, including ice that still carries excess greenhouse gases. Other lead measurements failed a decontamination plateau test and are treated only as limits. Either way, the authors do not attribute the gas excess to modern dust.

Microbes are the remaining popular alternative. Maximum excess carbon dioxide in the Larsen ice is about 180 parts per million. If that carbon dioxide came from respiration, the corresponding oxygen consumption would be about 0.08 percent of the trapped oxygen, shifting the oxygen-isotope ratio of leftover O2 by only 0.01 to 0.02 per mil, which would be hard to see. What they actually see in the top 0.10 to 1.65 metres is a depletion of that ratio relative to WAIS Divide ice of the same age, the opposite direction from respiration. Oxygen-to-nitrogen ratios sit in the ordinary range for bubbly Antarctic ice. Methane carbon isotopes in the excess gas, minus 33.8 and minus 35.0 per mil in two samples from core 111, are far too heavy for typical microbial methanogenesis, which runs about minus 70 to minus 50 per mil.

Photochemistry in the uppermost ice

Ultraviolet light as short as 290 to 295 nanometres reaches the Antarctic surface during ozone-hole season. Model simulations suggest that, under dust-free conditions, significant ultraviolet flux may persist to depths of several metres. Field measurements in Greenland’s ablation zone have found transmitted sunlight at 350 to 900 nanometres in the upper 12 to 124 centimetres of bare ice. No photolysis rate was measured at Larsen. The physical case is that light may get in.

The authors’ positive evidence is the depleted oxygen-isotope ratio of trapped O2 in the upper one to two metres. They sketch a pathway they themselves call speculative: nitrate photolysis and photochemistry of dissolved organic matter can make reactive oxygen species; those species could couple oxygen exchange among O2, carbon dioxide and the surrounding ice. Carbon dioxide in polar ice is already known to exchange oxygen with ice. Photochemically mediated exchange between O2 and carbon dioxide in the atmosphere can lower the oxygen-isotope ratio of leftover O2. The Larsen depletion is read as consistent with that class of process, not as a demonstration of any one reaction.

Snow cover at the site was not uniform. Some of the drilling area was bare blue ice; some carried two to three centimetres of snow, occasionally five to ten, and up to 20 centimetres on one day. Photochemistry, they write, may have occurred preferentially on bare or thinly covered ice and was likely strongest nearer the surface.

The extra carbon dioxide looks mixed. Average carbon-isotope values of the excess are minus 11.5 plus or minus 2.6 per mil in glacial-period samples and minus 14.7 plus or minus 3.4 per mil in early Holocene and termination samples. Those sit between typical organic carbon near minus 25 and inorganic carbon near zero. Glacial samples sit closer to the inorganic end, which matches higher dust, and therefore more carbonate, in glacial ice. The extra methane in core 111 looks organic.

Ablation rates for Antarctic blue-ice areas near 1,000 metres elevation run 10 to 23 centimetres of water equivalent per year. At that rate the surface modification could have accumulated over the past 20 to 50 years.

The oxygen-isotope depletion that the authors treat as the photochemical fingerprint stops at about one to two metres. Methane anomalies continue to 4.60 metres. Carbon dioxide levels off at 4.60 metres and remains higher than the Dome C and WAIS Divide records. The paper’s own limit is explicit: the clearest evidence for photochemical influence is confined to the uppermost interval, and deeper anomalies "should therefore not be attributed to photochemistry alone."

The methane decline toward the surface, seen in all the Larsen cores from 0.35 to 1.15 metres upward, has a published analogue at Taylor Glacier that was read as methanotrophs eating methane. The Larsen authors say photochemical sinks should also be considered. A Rayleigh calculation on two shallow samples from core 300 treats 20 to 30 centimetres as the starting point and 0 to 10 centimetres as the end. It gives a fractionation factor of 0.965, which sits somewhere among published factors for oxidation by OH, atomic oxygen and chlorine. That calculation stands or falls on the starting-point assumption.

Blue-ice areas are 1.67 percent of the Antarctic Ice Sheet, and they are being prospected for very old ice precisely because old ice can sit at the surface. The same first author has identified 320,000-year-old blue ice at Elephant Moraine. The authors say shallow post-depositional modification may be relevant when those surface archives are read as atmosphere. The practical note at the end of the paper is smaller: keep unnecessary light off ice that is going to be analysed for gases.

Blue ice at the surface is old air that has also been sitting in the sun.