Ancient Arctic Sediments Suggest Winter Sea Ice Could Shrink Faster Than Climate Models Predict

Arctic winter sea ice may vanish faster than scientists have expected, according to new research that used clues from 6,000 years ago to check how well today's climate models perform. The findings, published in Nature Geoscience, suggest future winter ice coverage could be roughly one million square kilometers smaller than earlier predictions indicated.

Much of the public focus on Arctic ice loss has been on the retreat of summer sea ice. Winter ice has received less scrutiny, even though, according to the researchers, climate models disagree most about how much ice will cover the Arctic in the colder months. That matters because winter sea ice helps control how much heat and moisture escapes from the ocean into the atmosphere. Declining ice cover has been associated with colder winters and increased snowfall in the United Kingdom, northern Europe and North America.

Recent observations underline the concern. In March 2025, Arctic winter sea ice reached the lowest annual maximum extent recorded since modern monitoring began, and the Arctic matched that record again in 2026.

To improve projections, scientists at the University of Cambridge and Mexico's Centro de Investigación en Matemáticas (CIMAT) turned to evidence preserved beneath the Arctic Ocean. The team assembled what they describe as the largest database of its kind of a chemical marker produced by algae that live within sea ice. The samples came from sediment cores deposited during the mid-Holocene, about 6,000 years ago, when small changes in Earth's orbit created a warmer Arctic with less extensive ice.

The marker, known as IP25, is produced by sea-ice-associated algae. The researchers combined IP25 measurements with other chemical markers left by organisms living in open water, allowing them to reconstruct ice conditions thousands of years in the past.

They then compared those reconstructions with simulations of the same period from 27 climate models. Using a statistical method called Bayesian weighting, they gave more influence to models that more closely matched the geological evidence and less to those that did not. Because the physical processes governing sea ice remain consistent across different climates, the team could apply the same weights to simulations of both the recent past and the future.

The comparison exposed a problem: collectively, the models predicted more winter sea ice than the ancient evidence supported. "When we apply the model weights to historical simulations, we completely eliminated the positive bias," said lead author Crystal Fu, from Cambridge's Scott Polar Research Institute. "The future projections similarly end up being lower than previous estimates, by about a million square kilometers."

The approach also tightened the forecasts. According to the researchers, it reduced the range of possible outcomes in some projections by as much as 85%. Fu noted that linking paleoclimate records to future projections has been difficult because of uncertainty at both ends. Co-author Marco Aquino-López, of CIMAT, said the method yields more reliable maps of where Arctic winter sea ice is likely to disappear, while showing how much uncertainty remains.

The mid-Holocene warming was driven by orbital changes rather than human activity, which is responsible for today's warming. Fu said the sea-ice response follows the same basic physics regardless. "It doesn't matter if the warming is natural variation or caused by human activity: if it warms, sea ice is going to melt," she said.

The researchers also applied their weighting to precipitation projections. They found indications that current models may underestimate future snowfall in southeastern Greenland and rainfall across the Mediterranean region. More snowfall over Greenland could affect the stability of the Greenland ice sheet, while heavier Mediterranean rainfall could influence flood risks in southern Europe.

The source material describes these precipitation results as indications rather than firm conclusions, and it does not detail specific limitations of the study or planned next steps. The revised simulations, the researchers suggest, provide a clearer foundation for investigating how a changing Arctic could affect weather elsewhere.

The research received partial funding from the US National Science Foundation and the Cambridge Trust.