March 2026
Happy spring in the Northern Hemisphere! There has been a lot to talk about lately climate-wise, from new sea-ice records to global ocean sea surface temperatures surging again to record highs, to summer-like heat across the southwestern United States, and of course, increasingly confident forecasts for a sizable El Niño later in 2026. If you are not up to date on some of these topics, I recommend scrolling back through my social media feeds… it has been a very alarming period. Anyways, this month’s ‘Climate Viz of the Month’ is a bit shorter as I catch up on some rest from last month. Here, I just wanted to quickly highlight a new visualization showing a different way the Arctic has warmed in recent years relative to a historical baseline, along with revealing the types of uncertainties present in station-based surface temperature observational datasets.
Using NOAA’s NOAAGlobalTempv6.1.0 and NASA’s GISTEMPv4 datasets, I have separately plotted the decade of the warmest annual mean temperature for every grid point on the polar stereographic map. Note that these datasets are provided on relatively coarse spatial grids, which is why the “boxes” are quite visible in the plot rather than showing smoother contours with finer spatial features. To produce this map, I considered all available years in each dataset (beginning in either 1850 or 1880), calculated annual means, and then ranked these values independently at each latitude and longitude point. Rather than go into all of the technical differences between these two observational datasets (see the Climate Data Guide for summaries: NASA vs. NOAA), it is important to note that they use different approaches for spatial interpolation and infilling, which becomes especially relevant in the polar regions where in situ observations from weather stations and buoys are sparse. I should also point out that NOAA’s product was recently updated from v6 to v6.1 this year, and not that long ago it did not even provide temperature coverage across large portions of the polar regions.
Comparing these two visualizations reveals some striking (though largely unsurprising) similarities, particularly in the broad-scale patterns of where the warmest decades have occurred. Across much of the Arctic Ocean, the warmest conditions are most often found in the 2010s, while over parts of Siberia they more frequently occur in the 2020s. There are some small differences between the datasets at individual grid points, especially near the edge of the Arctic Circle (white dashed line), reflecting differences in data input coverage and interpolation. However, the overall climate signal is clear: for nearly all locations, the warmest years on record have occurred within the past two decades, even when considering the full historical record back to the mid/late 19th century. Another key takeaway is that uncertainties in Arctic temperature estimates can be substantial, particularly in data-sparse regions, which reinforces the importance of comparing multiple datasets rather than relying on any single product (including for using atmospheric reanalysis). Continued research and sustained support are critical for maintaining and improving these observational records, especially as the Arctic continues to change rapidly. Now onto my summary of the region from last month…
March was a historic month for temperatures across the Northern Hemisphere, especially over the United States. Farther north, the pattern was also striking, with unusually sharp contrasts: record-breaking warmth over the contiguous U.S., anomalous cold over Canada and Alaska, and more anomalous warmth across the northernmost Arctic. While March 2026 certainly did not rank among the warmest Arctic-wide monthly means, there were notable regional extremes. In fact, temperatures were statistically tied for the warmest on record near the North Pole, as well as across the Barents Sea region and near Svalbard. In these areas, anomalies exceeded 5°C above the 1981-2010 average, extending from the Greenland Sea across the Arctic and into eastern Siberia. Meanwhile, temperatures were more than 5°C below average over the Kara Sea and western Siberia, along with northern land areas of North America, where monthly mean near-surface air temperatures in some locations remained below -30°C.
Arctic sea-ice extent remained unusually low and was statistically tied for the lowest on record for March. This follows the new record low for the annual maximum set earlier this year, breaking the previous record from 2025, which I discussed in last month’s blog.
Two technical notes for this month’s Arctic recap, both tied to the decommissioning of the NCEP/NCAR R1 reanalysis. One of my earliest visualizations I created (I think from back when I was in grad school) showed a ranking plot using 925 hPa temperatures from NCEP/NCAR R1 to track how warm or cold each Arctic month has been since 1979. I chose 925 hPa in part because R1 has known limitations in representing Arctic boundary layer temperatures, and because surface values can be influenced by heat flux exchanges, especially near sea-ice anomalies and around the marginal ice zone. With R1 now retired, I have transitioned this visualization to ECMWF’s ERA5 and will continue updating it in near-real time on my Arctic Temperature page. The original R1 version remains available on my website for archival purposes though. You may notice some differences in exact rankings between ERA5 and R1, which is expected given improvements in data assimilation and reanalysis-model physics, along with persistent observational gaps in the Arctic, such as the limited radiosonde coverage. Despite this, the agreement is better than I expected, and the long-term warming patterns are very similar.
Finally, we are still missing an update from PIOMAS this month for Arctic sea-ice thickness and volume (and GIOMAS for the Antarctic). I have not yet identified a workaround while we await further updates to the model (see last month’s blog for more details). In addition, the merged satellite-derived product from CryoSat-2/Sentinel-3/SMOS has issued its final thickness update for the season and will resume in the fall. Even so, all available evidence suggests that mean Arctic sea-ice thickness remains near historic lows.
February 2026
It’s time for my next “climate viz of the month” blog, but first some unfortunate news. The widely used Pan-Arctic Ice-Ocean Modeling and Assimilation System (PIOMAS; Zhang and Rothrock, 2003) will be pausing in the short term. The reason is that NOAA NWS/NCEP has terminated production of the NCEP/NCAR Reanalysis 1 (R1) as of 18 March 2026 (see notice), which is a key input for PIOMAS.
So why is this important? PIOMAS is an ice-ocean model, meaning it does not include an atmosphere. Instead, it reads in atmospheric information such as surface winds, air temperature, and cloud cover fraction, along with sea surface temperatures, from R1. While it is true that R1 is quite old and will be replaced by newer reanalysis products like CORe, PIOMAS was specifically tuned and calibrated using R1. That matters, as changing the atmospheric forcing is not a simple swap and can influence simulated sea-ice thickness and volume trends (see this study). This means there will be a significant amount of work required to re-calibrate PIOMAS for a different reanalysis product.
And why does this matter? PIOMAS is the only dataset that provides a temporally and spatially complete record of Arctic sea-ice thickness and volume since January 1979 (to my knowledge). There are other modeled products, including those from the Danish Meteorological Institute, as well as satellite observations, but they do not provide a long enough or consistent record to fully capture climate change trends (check out this comparison together we put together). Satellite altimetry products are also more limited in summer because melt ponds interfere with thickness retrievals (though this might be changing). PIOMAS has been widely evaluated, including in my own research, and is one of the go-to datasets in polar climate science. Even a temporary pause is a big loss.
This all creates a gap in one of the most physically meaningful indicators of Arctic climate change: sea-ice volume. It also comes at a time when Arctic sea-ice volume has been near record lows, with especially thin ice in parts of the central Arctic and near the North Pole. I also expect that GIOMAS may be affected given its reliance on similar atmospheric forcing, though I have not yet seen any official notice (I sent an email to follow-up). This will also affect my Antarctic/Global sea-ice volume graphics. It is a frustrating situation. At a time when we should be expanding and refining long-term climate data records, we are instead facing more challenges, especially due to limited funding and institutional support. Hopefully there is a clear path forward soon. I will also do my best to explore alternative sources and develop new graphics for continued real-time monitoring of sea-ice thickness and volume on my website, though this will take some time to fully assess their strengths and limitations. Stay tuned.
Now that I have shared that we are losing a key dataset for Arctic sea ice, I also want to highlight where things currently stand. The annual maximum Arctic sea-ice extent reached another record low this year, effectively tied with the previous record set just last year. The maximum occurred on 15 March 2026 at 14.29 million square kilometers (5.52 million square miles) (NSIDC Sea Ice Index v4 data). 2025 was at 14.31 million square kilometers (5.53 million square miles). This is about 1.3 million square kilometers (500,000 square miles) below the 1981 to 2010 climatological average. The timing was close to normal. The average date of the maximum over 1981 to 2010 is March 12.
It is also important to remember that these records begin in 1979, with the start of the temporally- and spatially-consistent passive microwave satellite observations. Within that record, the past two winters now stand out for the lowest annual maximum extents.
My special visualization for this month shows changes in daily Arctic sea-ice concentration from 1 January 2026 through late March, using the high-resolution (about 3 km) AMSR2 satellite product. You may notice some artifacts along coastlines, where the satellite can falsely detect sea ice. These are common and can mostly be ignored. The fast pace of my animation is intentional. It highlights the day-to-day growth and variability that define an Arctic winter. I have also included a second animation showing the long-term changes in the annual maximum extent from 1979 to 2026. The downward trend is obvious.
Overall, this winter has been particularly concerning across the Arctic. Sea ice remained unusually low for much of the season, and in many periods tracked at or near daily record lows for the time of year.
While the final maximum is now in, the broader winter context still stands out. Conditions were especially poor across several regions, including the Sea of Okhotsk, Baffin Bay, Labrador Sea, Barents Sea, and Kara Sea, where sea-ice extent remained persistently below average. The Sea of Okhotsk in particular saw one of the smallest maximum extents on record for that region and stands out as a major outlier compared to any other year in the satellite record as of late March. One of the few exceptions was the eastern Bering Sea near Alaska, which saw higher ice coverage compared to average, particularly over the past month when ice extent reached its farthest southward position in several years.
The thickness and volume story has been just as concerning. Based on PIOMAS, sea ice north of Greenland has been more than 2 meters thinner than average in places. Ice near the North Pole has also been at or near record-low thickness now for several months. In February, total Arctic sea-ice volume ranked as the second lowest on record.
Taken together, the Arctic is entering late winter in one of its weakest states in the satellite era. As I say every year at this point, it is still too early to say what this will mean for the summer melt season, since regional weather patterns can shift quickly. But winter 2025 to 2026 is another clear signal of how rapidly the Arctic is changing. Human-caused climate change is continuing to reshape the polar environment, with growing impacts near and far. While it seems like natural climate variability has temporarily offset some of the long-term trend for September Arctic sea ice (around the annual minimum), recent winter conditions show a more steady decline. Given that the Arctic Ocean is ice-covered in winter (and will continue to be well into the future), the declines are happening around the outer margins of the region.
This winter also featured notable temperature variability and extremes. Greenland experienced record warmth in January, followed by a sharp shift to much colder than average conditions in February across the area and parts of northern Canada. Across the Arctic, lower tropospheric temperatures (925 hPa) ranked as the 2nd warmest on record in January but dropped to around the 30th warmest in February. Near-surface air temperatures told a similar story, with February ranking 31st warmest. So while February seemed relatively cold compared to recent years, it does not offset the broader warmth. Looking more closely at cumulative cold, total freezing degree day anomalies across the high Arctic (north of 80°N) show one of the largest negative deviations on record within my tracking period (which began in July 2025 and uses a July to June year). This highlights the overall lack of sustained cold recently across the northernmost parts of the planet, even with short-term variability like in February.
That’s all for now – thanks for reading! You can find archives of my past blogs listed below. If you would like to support this website, you can also find a Buy Me a Coffee page: https://buymeacoffee.com/zacklabe.
January 2026
Hi all! Sorry for the delay in posting my first blog of the year. I’ve been spending a lot of time updating my visualizations with data through 2025 (and in some cases 2026). This means downloading and processing each dataset again and then manually updating the code and fine-tuning the figures. Maybe one of these years I’ll find a quicker way to roll everything into a new year… At this point, almost everything on my website should be up-to-date, aside from one or two remaining datasets needed for my reanalysis comparisons. If you notice any issues or typos, please let me know. I’m also in the process of adding alt text to my long-term visualizations (those updated monthly or yearly), which should help improve accessibility on these pages – thanks for your patience while I work through them.
With that said, here is a quick entry for January. As a reminder, these posts are always a bit lagged since I also summarize recent monthly conditions in the Arctic. I’m aiming to have the “February” post up by the end of this month.
My current entry features an animation showing monthly mean snow depth from the ERA5‑Land reanalysis during 2025 (January through December). This dataset is closely related to the standard ERA5 reanalysis produced by ECMWF, but it focuses specifically on land surface conditions at a higher spatial resolution of 0.1° by 0.1° (latitude by longitude) (~9 km). That finer grid helps better resolve coastlines, topography, and other important land surface characteristics. I’m using a sequential color scheme here (“bone” from matplotlib), with the scale saturating at 0.5 meters to emphasize variability at lower latitudes where snow depths are generally much smaller compared with places like the Greenland Ice Sheet and the Antarctic Ice Sheet. I also intentionally reduced the contrast for continents and coastlines, so the focus stays on the seasonal cycle of snowpack, particularly the differences between the Northern and Southern Hemispheres. There are also some interesting regions where snow persists year-round outside of the major ice sheets, though you can still see summer melt even across parts of western Greenland. Keep in mind that this shows monthly means, so it does not capture short-lived synoptic events where snowfall may have occurred for only a day or two.
I always enjoy visualizations that highlight the seasonality of meteorological and climatological variables, since so much can be explained by Earth’s axial tilt and the distribution of land versus ocean between the hemispheres. Snow cover plays an important role in the climate system because of its high albedo (reflectivity), meaning it reflects a large fraction of incoming sunlight back to space. Snow also strongly influences local weather and climate conditions, including boundary layer temperatures like through cold air damming events or nighttime radiational cooling. The amount of water stored in the snowpack (often measured as snow water equivalent) is critical for millions of people because it acts as a natural freshwater reservoir that feeds rivers used for drinking water and irrigation. It also supports ecosystems by influencing soil moisture, vegetation health, and the timing of plant growth.
However, human-caused climate change is reducing snow cover in many regions of the world. The largest declines occur during Northern Hemisphere spring, with substantial losses across Eurasia and North America by June as warmer temperatures lead to earlier melt. This reduction contributes to further Arctic amplification through the snow‑ice‑albedo feedback, where the loss of bright snow exposes darker land surfaces that absorb more sunlight. In mountainous regions, declining snowpack at higher elevations also reduces water availability later in the year and can leave landscapes drier and more susceptible to wildfire. Recent examples include the extreme warmth and early snowmelt that contributed to widespread wildfires across Siberia in 2020 and in Canada in 2023. Overall, understanding the seasonality, variability, and long-term trends in snow is an important part of monitoring changes across the climate system.
For January 2025, conditions across the Arctic featured unusually warm temperatures and very low sea ice, including low values for total extent, thickness, and volume. Some of the largest temperature anomalies occurred across eastern Siberia, the Canadian Arctic Archipelago, and the Barents Sea region. In fact, temperatures reached record-high levels across large portions of Greenland, especially along the western coast, with departures exceeding 5 °C above the 1981–2010 average. These conditions were partly linked to the large-scale atmospheric circulation, including a persistent blocking pattern and the associated phase of the N(AO). At the same time, temperatures were much colder than average across parts of western Alaska, western Siberia, and over Scandinavia. Sea ice has remained very low this winter, with numerous new daily record lows during the freeze season. The average sea-ice thickness near the North Pole also continues to reach record-low levels according to PIOMAS. The annual maximum sea-ice extent will likely occur soon, although it is still a bit early to predict its final ranking. That said, it will probably end up within the ten lowest maximum extents in the observational record.
Other Blogs (Monthly):
My visualizations:
The views presented here only reflect my own. These figures may be freely distributed (with credit). Information about the data can be found on my references page and methods page.