- New study: Two low-ice winters dash hopes that sea-ice decline in the Arctic had lessened over the past two decades
- The Arctic: High sea surface temperatures and very little sea ice in the northern Barents Sea and in Fram Strait
- The Antarctic: Only a moderate winter increase in sea-ice extent. Nevertheless, pack ice in the Bellingshausen Sea bounces back from its deficit in the first half of winter
The Arctic: Two low-ice winters dash hopes that sea-ice decline had lessened
For anyone interested in the climate and living in Europe, this July hardly offered the hoped-for break from bad news. Instead, it was characterised by more than 12,000 heat-related deaths in Germany; devastating wildfires in Spain, France and Greece; and by a global ocean in which, outside the polar seas, surface temperatures easily exceeded past record highs. (Figure 1)
Figure 1: Daily sea surface temperature (°C), mean value for the global ocean beyond the polar seas (60° S – 60° N) for the years 2023 (yellow), 2024 (orange), 2025 (red) and 2026 (dark red). All other years in the period 1979 – 2022 are represented by grey lines. The daily mean for the reference period 1991 – 2020 is represented by a dashed grey line. Data source: ERA5. Image credit: C3S/ECMWF
In the second half of July, these headlines were joined by news from the field of sea-ice research, which was hardly more encouraging. The extremely low Arctic sea-ice extents in the past two winters have impacted long-term sea-ice statistics to such a degree that researchers now speak of a clear, statistically significant downward trend. In the 21 previous years – that is, from 2004 to 2024 – the values for the annual winter maximum extent had varied considerably, but had generally remained at the same level and had not indicated a decline. Some experts felt this development was grounds for hoping that sea-ice decline in the Arctic might be slowing.
The low winter maximum extents from February/March 2025 and 2026 dashed those hopes. In a new study, British researchers have shown that the long-term trend becomes clearly negative when the winter maximum extents from 2025 and 2026 are included (Figure 2). As they report, the sea-ice decline from the 2024 winter maximum to the 2025 winter maximum was 6 percent, the largest year-on-year loss to date.
“The latest observations restore a statistically significant downward trend, showing how a few unusual years can quickly change the overall picture,” explain the authors Duo Chan and Alessandro Silvano in an article for the online portal “The Conversation”.
Figure 2: Winter maximum sea-ice extent in the Arctic from 1980 to 2035. The values for the years 2027 to 2035 are predictions based on climate models. The period with the slowest decline is shown in grey. The 2025 winter maximum represents the lowest value to date in the time series for the past 45 years. Figure: Chan & Silvano (2026), The Conversation, https://doi.org/10.64628/AB.p3etjjf7c
The Arctic: Very little sea ice in the northern Barents Sea and in Fram Strait
Arctic sea-ice development held few surprises in July 2026. With a mean sea-ice extent of 7.72 million square kilometres, satellites recorded the seventh-lowest July sea-ice extent since the beginning of satellite observation. In addition, this value was clearly above the trend line and nearly matched the mean July values from 2024 and 2025 (Figure 3).
“The constant level of the mean sea-ice extent shouldn’t distract us from the fact that there are clear differences between the sea-ice distribution in July 2024, 2025 and 2026. Especially in the eastern Arctic and in the North Atlantic, this summer we’re seeing areas in which the ice margin has retreated far to the north, whereas melting in the Beaufort Sea and the northern part of the Bering Strait has slowed,” says Dr Renate Treffeisen, an atmospheric researcher at the Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research and co-initiator of the Sea Ice Portal (Figure 4).
In July 2026, there was significantly less sea ice compared to the long-term mean and to previous years particularly in Fram Strait and the northern Barents Sea (Figure 5). “In the waters north of Svalbard and of the Russian Arctic island Franz Josef Land, large areas were nearly ice-free. As such, we’re seeing the continuation of a trend, one that we reported on the first signs of this spring. We’re currently combining and analysing all atmospheric, oceanic and sea-ice data for the year to date. We’ll present the results of our analyses in the August sea-ice update,” she explains.
Figure 3: Time series of mean sea-ice extent in the Arctic for the month of July.
Figure 4: Sea-ice concentration in the Arctic – to the left, the monthly mean for July 2026; to the right, on 31 July 2026. The turquoise line represents the ice margin for the mean sea-ice extent in July 1981 – 2010. A comparison of the two ice-concentration maps shows that, by the end of the month, nearly ice-free areas had formed in the Atlantic influence zone of the Arctic Ocean. Large regions with low sea-ice concentrations in the eastern part of the Central Arctic can also be clearly recognised.
Figure 5: Comparison of the sea-ice extent in July 2026 to the long-term mean (left) and to the monthly mean July values from the years 2025 and 2012. In all three figures, the sea-ice decline in the northern Barents Sea and in Fram Strait is evident.
Extreme sea surface temperatures in the ice-free areas of the Arctic Ocean
The rapid and extensive sea-ice melting in the Atlantic-influenced part of the Arctic Ocean was most likely primarily due to the unusually high ocean temperatures. Our map of mean sea-surface temperature anomalies shows large areas in which the surface water was 2.5 to 4 degrees Celsius warmer than in the reference period 1971 – 2000 (Figure 6). This virtually inconceivable amount of thermal energy has now catapulted the Arctic Ocean into a new dimension of temperatures. It remains to be seen what the consequences will be for the sea-ice development in the remainder of the year. But it would hardly come as a surprise if the formation of new sea ice this autumn were delayed more extensively and longer than we’ve seen in the past.
In July 2026, air temperatures in the Arctic were chiefly shaped by a large low-pressure cell over the central Arctic Ocean, which produced unusually low temperatures off the northern coasts of Greenland and Canada (Figure 7). Air masses over parts of Alaska, over the Bering Sea and over the eastern part of Siberia were also cooler than in the reference period 1971 – 2000. In contrast, air masses over northern Canada, over Greenland and over the rest of Siberia were 0.5 to 5 degrees Celsius over the July mean in the reference period. As such, it was unusually warm over the Russian marginal seas, over Greenland, and over the Canadian and Russian Arctic.
To understand the year-to-date sea-ice development, it’s worth taking a closer look at Arctic air temperatures in the past 12 months, as our colleagues from Berkeley Earth did and recently reported on in their July update (Figure 8). The mean air-temperature anomalies map shows e.g. temperature anomalies of up to 6 or 7 degrees Celsius over Svalbard and Franz Josef Land, and therefore over precisely the same region that ostensibly experienced the most substantial sea-ice decline this summer. The temperature anomalies over Fram Strait were up to 3 degrees Celsius. However, we wish to point out that Berkeley Earth compares its current temperature data with the July mean from the years 1951 – 1980. In contrast, with regard to temperature anomalies the Sea Ice Portal normally uses the reference period 1971 – 2000. Accordingly, the individual temperature anomalies cannot be directly compared with one another.
Figure 6: Mean Arctic sea-surface temperature anomalies in July 2026. The map shows very few regions in the North Atlantic in which the surface water was not substantially warmer than in the reference period July 1971 – 2000.
Figure 7: Mean Arctic temperature anomalies in July 2026, compared to the reference period July 1971–2000. Whereas the air masses over the marginal zones of the Arctic were in part much warmer than in the reference period 1971 – 2000, those over the central Arctic Ocean and over the North Pacific were colder. Source: Copernicus ERA5 data
Figure 8: Mean temperature anomalies for the period August 2025 – July 2026, based on the July mean for the years 1951–1980. Figure: berkeleyearth.org
The Antarctic: Pack ice in the Bellingshausen Sea bounces back
The formation of new sea ice in the Antarctic was slightly below expectations in July 2026. Especially in the second half of the month, ice growth slowed, causing the daily sea-ice extent curve to drop below the span of minima and maxima for the years 1981 – 2010.
In the course of the month, the sea-ice extent climbed from 14.1 million square kilometres to 16.16 million square kilometres – a gain of roughly 2 million square kilometres. The monthly mean value was 15.14 million square kilometres, making it the fifth-lowest July sea-ice extent since the beginning of satellite observation in the Antarctic (Figure 9).
“The development of the sea-ice situation in the Antarctic over the past three winters is promising, as the winter sea-ice extent has gradually increased since the record-low year 2023, bringing it closer to the long-term mean. That being said, it’s unclear how long this trend can continue,” says Dr Renate Treffeisen.
With two exceptions, there was less sea ice than in the reference period 2003 – 2014 in all sectors of the Arctic Ocean in July 2026. Satellites only detected more pack ice in the Amundsen Sea and off the coast of Adélie Land (Figure 10).
The sea-ice situation in the Bellingshausen Sea has now improved. At the beginning of winter, the formation of new sea ice there was extremely sluggish. By the end of July, however, the region’s pack ice had nearly completely recovered. Though the rate of new sea-ice formation was still below the long-term mean at the end of the month, the mean sea-ice extent in the Bellingshausen Sea in July 2026 was significantly higher than in July 2025 (Figure 11).
Figure 9: Time series of mean Antarctic sea-ice extent for the month of July. Whether the monthly means from July 2024, 2025 and 2026 actually constitute a trend is something we won’t be able to say until next July at the earliest.
Figure 10: Difference in the mean position of the ice margin in July 2026, compared to the long-term mean for the years 2003 – 2014. Regions marked in blue had more Antarctic sea ice in July 2026 than in the reference period; those marked in red – this month, all sectors of the Southern Ocean except the Amundsen Sea and the waters off the coast of Adélie Land – had less.
Figure 11: Difference in the mean position of the ice margin in July 2026, compared to July 2025: the area marked in blue in the Bellingshausen Sea (waters west of the Antarctic Peninsula) reflects the sea-ice growth compared to the previous year.
Graphics