- The Arctic: Because winds scattered the ice of the eastern Arctic, the sea-ice extent, at 5.6 million square kilometres, was moderate.
- Powerful winds and ice migration, plus a warmer ocean, explain the differing types of sea-ice development in the eastern and western Arctic Ocean.
- The Antarctic: For the time being at least, a short-term decline in mid-winter dashed hopes that the sea-ice situation would rapidly improve. The mean sea-ice extent was the third-lowest for August since the beginning of record-keeping.
The Arctic: Headed towards “split” sea-ice cover?
On the last day of August 2026, the sea-ice cover on the Arctic Ocean seemed to be split in the middle. The ice concentration map from 31 August 2026 predominantly shows densely compressed pack ice, with concentrations from 90 to 100 percent, in the western part (Figure 1). Conversely, the various shades of blue on the map reflect moderate to very low ice concentrations in the eastern part. In other words, the satellites detected very little sea ice in these areas, and the floes they did find were widely scattered. In this regard, there was an extensive strip with extremely little ice that ran roughly along the Prime Meridian and extended throughout the central Arctic.
“This is the first time that we’ve seen such a clear split in Arctic sea-ice concentration. In early August, the French polar cruise ship ‘Le Commandant Charcot’ took advantage of the extremely loose floe clusters in the eastern part of the Arctic to proceed directly to the North Pole. At the end of the month, it would have been theoretically possible to sail from Svalbard to the Bering Strait via the North Pole without the risk of becoming trapped in thick pack ice along the way,” says AWI sea-ice physicist Lena Buth with regard to the sea-ice situation at the end of the month (Figure 2).
Figure 1: Sea-ice concentration on 31 August 2026. The remaining ice looks like a pie with two different halves: on the left, the dense pack ice of the western Arctic; on the right, the mostly scattered groups of floes of the eastern Arctic. Moreover, the strip-like area of very low sea-ice concentrations, extending from Svalbard to far beyond the North Pole, is clearly recognisable.
Figure 2: AWI sea-ice physicist Lena Buth and her colleague Thomas Krumpen investigated the causes of the pronounced “split” in the Arctic sea-ice cover. Photo: Dr. Janna Rückert, University of Bremen
Moderate sea-ice extent despite the lack of ice
The fact that the sea ice in the eastern Arctic was widely scattered explains why the monthly mean sea-ice extent, at 5.6 million square kilometres, was comparatively moderate. The August 2026 value came in at thirteenth place on the list of lowest extents for the month and was more than 500,000 square kilometres above the trend line (Figure 4).
Under these conditions, any predictions regarding the September sea-ice minimum would be pure speculation. “In August, we saw how susceptible the sea ice is to winds. We won’t know just how low the September minimum is until it’s actually reached,” says Lena Buth.
Figure 3: Time series of mean sea-ice extent in the Arctic for the month of August.
Ice drift: Unusually calm in winter, driven by storms in summer
Over the past several days, the PhD candidate and her colleague Dr Thomas Krumpen have intensively investigated the causes of this phenomenon and identified several unique factors that could explain the differing sea-ice concentrations in the central Arctic.
“For one thing, we’re seeing clear differences in the ice drift between last winter and this summer,” says Lena Buth. From January to March, in broad expanses of the central and western Arctic the sea ice drifted more slowly than the long-term mean for the years 2010 – 2025 (Figure 4).
As a result of this slow drift, surprisingly little sea ice was transported through Fram Strait to the North Atlantic: between November 2025 and April 2026, only ca. 550 cubic kilometres of sea ice drifted out of the Arctic Ocean. “That’s the lowest value in our 15-year time series and only about a third of the mean value for the past few years,” explains Thomas Krumpen, an expert on sea-ice physics and satellite remote sensing at the AWI (Figure 5).
Figure 4: Mean sea-ice drift in the Arctic from January to March 2026. The black arrows indicate the mean direction and speed of the sea-ice drift. An arrow represents a speed of 5 kilometres per day. The colours indicate the drift speed anomaly compared to the long-term mean for the years 2010 – 2025, with blue shades representing a lower and red shades representing a higher drift speed than the long-term mean. In the first three months of 2026, the sea-ice drift was markedly less pronounced than the mean for the years 2010 – 2025 in broad expanses of the Arctic.
Figure 5: Time series of winter sea-ice export (from November to April) through Fram Strait. With a total amount of just 550 cubic kilometres of ice, the ice export for the winter half-year 2025/2026 was far below the trend line since 2011/2012.
Due to the sluggish ice transport in winter, a comparatively high volume of ice remained in the central Arctic through the beginning of the melting season. “However, lower winter export is no guarantee that there will be exceptionally large ice masses at the end of summer: long-term studies show that the ice export through Fram Strait is only responsible for part of the fluctuations in the September sea-ice extent, and that its effects greatly depend on the season,” Thomas Krumpen explains.
That statement also proved true this year: “From June to August, the weather and sea-ice situations in the Arctic changed fundamentally. An intense low-pressure cell over the central Arctic produced a counterclockwise pattern of motion and accelerated ice drift in many parts of the Arctic Ocean. As a result, more pack ice from the Canadian sector and from the region north of Greenland was transported toward Fram Strait,” says Lena Buth, describing the current situation (Figure 6).
This drift situation also explains why the ice export from the Arctic was unusually high in the summer months May to August. The low-pressure cell over the North Pole region also correlated with comparatively cool air temperatures in the region north of 80°N (Figure 7). These temperatures slowed summer sea-ice melting in the central Arctic.
Figure 6: Mean sea-ice drift, June to August 2026. The arrows indicate the direction and speed of the sea-ice drift; an arrow represents a speed of 5 kilometres per day. The colours indicate the drift speed anomaly compared to the long-term mean for the years 2010 – 2025. Clearly recognisable: the wide-scale, counterclockwise ice drift surrounding a low-pressure cell over the central Arctic Ocean. In broad expanses of the Arctic, the sea-ice drift was much more intense than the long-term mean. This contributed to the increased export of ice masses from the coastal waters of northern Greenland and the Canadian Arctic Archipelago through Fram Strait and into the North Atlantic.
Figure 7: The Danish Meteorological Institute records the daily mean temperature in the region north of 80°N. The temperature curve for 2026 (in orange) shows that the summer of 2026 was cooler than the long-term mean. Conversely, in the first four months of the year it was substantially warmer. Graphic: Danish Meteorological Institute Reference: https://ocean.dmi.dk/arctic/meant80n.uk.php
Unusually thick sea ice in northern Canada
Off the northern coast of Canada, due to the slower ice drift last winter the young sea ice wasn’t immediately transported to the open ocean; instead, it remained in the coastal waters, where it grew to impressive thicknesses. In the course of their aircraft-supported spring sea-ice monitoring campaign (IceBird), AWI sea-ice physicist Luisa Wagner and her team surveyed the sea ice in selected regions of North America – including the pack ice north of the Mackenzie Delta.
“Our readings indicate that the seasonal ice in the region surveyed was on average roughly 2.8 metres thick. This mean value is one of the highest in the Mackenzie Delta time series since 2008 and explains, together with the fact that the multiyear ice extended exceptionally far to the south, why the sea ice in the Beaufort Sea broke up later than the long-term mean in the spring and early summer,” explains Luisa Wagner (Figure 8).
The late breakup of the sea-ice cover in the Beaufort Sea could also explain why substantially more ice was still in the region in August, compared to both the long-term mean and to the previous year (Figure 9).
Figure 8: Mean thickness of seasonal sea ice in North America’s Mackenzie Delta, measured in the course of the AWI’s IceBird aerial monitoring campaigns since 2008.
Figure 9: Difference in the mean position of the ice margin in August 2026, compared to the long-term August mean and the means for the year 2025 and the record-low year 2012. As the comparison shows, the sea-ice decline in the Beaufort Sea in summer 2026 was less pronounced than in the reference periods.
Unusually thin ice north of Svalbard
While the seasonal ice in the coastal region of the Beaufort Sea was exceptionally thick this spring, the sea ice north of Svalbard was unusually thin, as readings from the sea-ice satellite CryoSat-2 confirm (Figure 10). “We assume that these low ice thicknesses contributed to the early sea-ice cover breakup in the northern Barents Sea,” says Thomas Krumpen.
The thin ice can most likely be attributed to warm Atlantic water, which penetrated far into the Arctic Ocean in winter. “It will take further analyses to confirm that assumption. But our sea-ice data shows that the ice in the region was unusually thin throughout the winter – in other words, that the anomaly was stationary and didn’t move along with the ice drift, which indicates that the ocean was the source of the warmth,” he relates.
Figure 10: Mean ice-thickness anomaly in the Arctic Ocean in April 2026, compared to the long-term mean for the years 2011 – 2025. The dark red areas north of Svalbard and off the northern coast of Greenland indicate sea ice that was up to one metre thinner than the long-term mean in April 2026. In the Beaufort Sea, in contrast, the sea ice in the coastal region was thicker than the long-term mean (blue shades).
Highly dynamic ice motion
When the sea-ice cover breaks up early, collapsing into individual floes, the latter are easy pickings for winds, which drive the floes faster and, depending on the wind direction, often drive them together or apart faster. Experts refer to this as highly dynamic sea ice. These conditions can produce abrupt changes in the sea-ice extent and compactness, as winds can rapidly scatter individual floes or drive them together. And there was precisely this type of abrupt change in ice density in mid-August (Figure 11).
“Ice compactness is a mathematical quantity and doesn’t belong to the standard parameters used in sea-ice research. We can calculate it by dividing the actual sea-ice area by the sea-ice extent – that is, including the areas with little ice cover. In August, the ice compactness declined significantly in the span of two weeks, which indicates that winds scattered part of the Arctic sea ice in that time. It’s especially unusual that this happened so fast,” says Lena Buth.
The AWI experts used ice-drift data to reconstruct the events. “In the first half of August, the sea ice in the eastern part of the Arctic Ocean was primarily driven poleward. Its compactness rose. But from 14 August, the drift changed. In several regions, winds scattered the ice, causing the compactness to rapidly decline,” Lena Buth explains (Figures 11, 12 & 13).
Figure 11: This map shows the development of sea-ice compactness in the Arctic. The red line represents the development in the first eight months of 2026. The grey lines represent the curves from the years 2012 – 2025. A major spike can be seen in August 2026: a rapid increase followed by a rapid decline. Both phases are attributable to the ice drift.
Figure 12: Map of the sea-ice drift on 9 August 2026. Winds and ocean currents drove the sea ice north of the Barents and Kara Seas together and poleward.
Figure 13: Map of the sea-ice drift on 24 August 2026. In the eastern part of the Arctic Ocean, and in some areas of the western part, the ice motion reversed and winds scattered the ice.
The Arctic: Baleen whales now feeding in the ice-free waters off eastern Greenland
Over the past two decades, rising ocean temperatures, dwindling sea-ice cover, and the progressive Atlantification of the Arctic Ocean have produced far-reaching changes in marine life along the eastern coast of Greenland. As experts report on in a new study, fin whales, humpback whales and minke whales now regularly come by the thousands to the newly ice-free waters off the eastern coast of Greenland, where they linger for at least four months (from July to October) to gorge themselves on fish and krill. Before the 21st century, these baleen whale species were rarely seen in waters north of 70°N, as the pack ice kept them from reaching the coastal Arctic waters.
The new study is based on data from systematic surveys conducted in August 2015 and 2024. In addition, the researchers had tagged individual whales with trackers and analysed the reports of local whalers. According to the reports, at least the minke whales hunted in the waters off eastern Greenland especially feed on capelin (Mallotus villosus) and Arctic krill (Thysanoessa inermis). Taken together, the three whale species consume an estimated one million metric tons of fish and krill per season.
“The relatively high contemporary abundances of all three species almost certainly reflect a recent redistribution and expansion of North Atlantic baleen whale populations, potentially supported by an influx of animals from West Greenland, where fin and minke whale abundances are declining,” the authors write.
The whales’ migrations also indicate that the cold-loving capelin stocks from the Denmark Strait no longer spend the late summer and autumn in the waters north of Iceland but instead migrate toward eastern Greenland. This behaviour matches the data, which shows that ocean temperatures have climbed both in the waters surrounding Iceland and in the waters of the East Greenland Shelf (Figure 14).
Figure 14: North Atlantic baleen whales are now also frequently seen in northern waters by those participating in Polarstern expeditions to the Arctic. This photo shows a fin whale off the coast of Svalbard. This summer, the researchers also observed blue whales, the largest mammals on the planet, in the ice-free waters of Fram Strait. Photo: Felix Tulatz/Alfred Wegener Institute
The Antarctic: Short-term ice decline dashes hopes of sea-ice recovery
In our sea-ice update for July, we had expressed the hope that the sea-ice situation in the Antarctic might gradually be improving. But that hope proved to be in vain: sea-ice growth on the Southern Ocean this August was well below the expected level. The sea-ice extent grew by a mere 1.16 million square kilometres. The monthly mean value was 16.57 million square kilometres, the third-lowest mean August extent on record. At the same time, the monthly mean was nearly one million square kilometres below the trend line (Figure 15) and lower than the monthly mean from August 2025. As such, the positive trend observed over the past two years has at the very least been interrupted.
One explanation for the limited ice growth could be an unexpected ice decline in the first half of the month. From 7 to 11 August 2026, the Antarctic sea-ice extent shrank by 400,000 square kilometres – an area the size of Germany, and in the middle of winter (Figure 16).
Figure 15: Time series of mean sea-ice extent in the Antarctic for the month of August.
Figure 16: Antarctic sea-ice extent in comparison: the blue curve for the year 2026 clearly shows a dip in the first half of August.
The sea ice retreated most clearly in the Amundsen Sea, the marginal zone of the eastern Weddell Sea, and the Indian sector of the Southern Ocean (Video 1 & Figure 17). Although the sea-ice cover began rebounding again in these regions just a few days later, the losses couldn’t be completely recouped by the end of the month, as a comparison of the mean position of the ice margin to the long-term mean shows (Figure 18).
Figure 17: Comparison of the mean position of the ice margin on 7 August 2026 and on 11 August 2026. Regions marked in red had less sea ice on 11 August 2026 than on 7 August 2026; those marked in blue had more.
Video 1: Development of sea-ice concentration in the Antarctic in August 2026.
The causes of the abrupt sea-ice decline cannot be fully explained on the basis of our data alone. However, the mean air-temperature anomaly may offer a starting point (Figure 19). Air temperatures over West Antarctica were on average up to 6 degrees Celsius warmer than the long-term mean in August 2026. This heat could be part of the reason for the sluggish formation of new sea ice in the Amundsen Sea.
The prevailing winds may have been a further factor. Our map of the pressure anomaly at sea level in August 2026 shows offshore winds in the Amundsen Sea (Figure 20). These winds may have compressed the pack ice.
Figure 19: Mean Antarctic temperature anomaly in August 2026, compared to the reference period August 1971 – 2000. While the air masses over the Weddell Sea were significantly colder than in the reference period, those over West Antarctica were up to 6 degrees warmer. Source: Copernicus ERA5 data
Figure 20: Mean Antarctic pressure anomaly at sea level in August 2026. Offshore winds over the Amundsen Sea and over the Weddell Sea can be recognised. The latter could help to explain the higher sea-ice extent in the Weddell Sea.
Graphics
