- July marks the midpoint of the summer sea ice melt season, during which ice declines rapidly under the almost constant Arctic sun, and melt ponds form on ice floes. Scientists study melt ponds to better understand sea ice dynamics and to help forecast the annual sea ice minimum in September.
- Sea ice melt ponds absorb more solar energy than bare ice or snow but far less than open ocean, which complicates the story of summer sea ice melt. But in the past, scientists lacked the tools to accurately measure melt pond impacts.
- New remote sensing data look at sea ice from different satellite resolutions. High-resolution imaging sends a clear picture of melt ponds in a very small area, while low-resolution imaging produces a fuzzy picture of them across the entire Arctic.
- A combination of satellite data and on-the-ice observations now help scientists track the evolution of sea ice melt ponds and their influence on the overall decline of sea ice in the Arctic.
July is the peak of melt season in the Arctic. The sun is high and shines for almost 24 hours a day. Air and water temperatures rise above freezing. Rapid midsummer melt is therefore both unremarkable and unsurprising. But it is also an unreliable indicator of just how much ice will cover the Arctic Ocean when the September annual sea ice minimum arrives.
Now a combination of satellite and on-the-ice observations are helping scientists better understand summer sea ice behavior and track the evolution of melt ponds atop the ice, so as to more accurately forecast what extent will remain frozen at the September minimum.
In separate research, scientists have recently demonstrated how to see details on sea ice, combining high- and low-resolution satellite images to get a clearer picture of melt ponds and thereby test how well climate models are capturing melt pond evolution.
Uncovering melt ponds with satellites
Ice and snow reflect sunlight, keeping the Arctic cool, while large dark areas of open ocean absorb abundant solar energy, which triggers more ice melt and delays the autumn refreeze.
Melt ponds sitting atop the sea ice absorb more energy than fresh snow, but less than open water. So, understanding their evolution could help scientists better forecast future seasonal sea ice extent. But melt ponds have always been hard to see — especially on satellite images.
Passive microwave data can’t distinguish between melt ponds and open ocean because those data are low-resolution and cover a broad area, explains University of Washington atmospheric scientist Cecilia Bitz. But even high-resolution satellite imagery, like that captured by NASA’s Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2), struggles to see melt ponds, especially those smaller than 2 meters (6.6 feet) across.
“ICE-Sat2 was definitely not intended to measure melt ponds,” notes Ellen Buckley, assistant professor in the University of Illinois Urbana-Champaign Earth Science and Environmental Change Department. But fortunately, the satellite has a green laser capable of penetrating the water column to measure both the surface and bottom of a melt pond.
In a study of multiyear sea ice, recently published in The Cryosphere, Buckley and her team combined multiple data sets from ICESat-2, the European Space Agency’s Copernicus Sentinel-2 satellite and commercial satellites to chart the evolution of melt ponds over the course of the 2020 melt season. The team found that the melt pond fraction (the percentage of ponded sea ice) remained below 5% in early June, then increased rapidly (peaking June 24), before slowly decreasing through early July (when melt ponds drained). The fraction remained low through the fall freeze up. Pond depth increased steadily through June and peaked July 16.
“Our results demonstrate that by combining high-resolution passive and active remote sensing we now have the ability to track evolving melt conditions and observe changes in the sea ice cover throughout the summer season,” the paper says.
Ice campaigns and satellites give a clearer picture
As summer sea ice melts and becomes too unstable for field research, remote sensing becomes increasingly important. But satellite data still often misidentify melt ponds as open water and fail to distinguish among multiple types of ice.
That problem is being addressed in part by University of Hamburg sea ice research scientist Niels Fuchs, whose photogrammetry technique stitches together high-resolution images of melt ponds to measure their size, bathymetry, volume, surface elevation above sea level and evolution over time.
In a recent paper in The Cryosphere, Fuchs combined satellite images with pictures collected from aerial surveys during the yearlong Multidisciplinary drifting Observatory for the Study of Arctic Climate (MOSAiC) polar expedition. He analyzed more than 1,600 melt ponds on the MOSAiC floe as it drifted in the Central Arctic near the North Pole.
Most Arctic satellite data tend to be lower-resolution, like NASA’s Moderate Resolution Imaging Spectroradiometer, which can’t see at night or through the clouds often blanketing the Arctic in summer. Sea ice scientists, Fuchs says, can instead use high-resolution satellite data to show the color of melt ponds as well as the reflectivity in the different wavelengths and bands to show how that evolves from first melt in June to first freeze in August or September.
The albedo of sea ice melt ponds is a crucial part of the summer sea ice story. As melt ponds form on fresh ice, they appear bright blue and are easily detected by satellites. But as the melt season progresses, ponds become deeper and darker as the ice below thins. So, satellites struggle to discern late melt season ponds from the ocean. Some types of sea ice — such as brash ice, the floating wreckage of floes that have broken apart — also tend to confuse satellite imaging.
Putting melt ponds into models
Fuchs and University of Washington polar research scientist Melinda Webster were co-authors on polar observation and modeling research associate Sanggyun Lee’s recent paper in Remote Sensing of Environment that looked at how accurately three different sets of optical satellite data captured melt pond fraction.
“We still have so much to learn with the physical processes of melt pond evolution and how to represent them in models,” Webster says. “We’re still at this stage that we need to evaluate how good a climate model represents and simulates melt ponds.”
The authors found the different data sets were at times in agreement, especially early in the melt season, but started to diverge as the melt season progressed, reflecting the complicated evolution of summer sea ice melt. Each of the three data sets captured only a part of the melt season well. “It’s really difficult to retrieve melt ponds accurately from medium-resolution satellite imagery. Especially because a lot of things look like a melt pond at that resolution,” Webster explains.
High-resolution satellite images best capture melt ponds, but the spatial coverage is very limited, so it’s difficult to get a pan-Arctic view of them. She adds, “You can go for more accuracy with limited spatial coverage, or you can go with less accuracy with more broad coverage of the Arctic.”
A better grasp of the dynamic behavior of sea ice may come from combining satellite and on-the-ice observations to track seasonal melt pond evolution. Fuchs says science will eventually get to a place where the physical processes of melt pond evolution are well understood and included fully in climate models, but likely not until satellite retrieval algorithms can provide a better view of pan-Arctic melt pond coverage.
While we may not know precisely how the 2024 melt season will end up this year, melt pond behavior this spring and summer suggests that the Arctic won’t break a record for the lowest sea ice extent this year, as it did in 2012 and 2020.
When sea ice reached its second-lowest extent in 2020 — part of the long-term trend of warmer air, higher ocean temperatures and declining sea ice — an unusually warm spring north of Alaska sparked an early melt and earlier formation of melt ponds on ice floes. In 2012, extensive melt, followed by a massive August storm, smashed thin sea ice into fast-decaying fragments.
This year, temperatures north of Alaska were below normal and early spring temperatures across the Arctic were low relative to recent record years. A cool spring discouraged widespread early melt pond formation, which could be an indicator of sea ice extent in the fall.
Even so, researchers say the sea ice is unlikely to return to levels observed in the early satellite record: The 17 years since 2007 have seen the lowest sea ice extents since recordkeeping began in 1979. And it seems like 2024 will likely continue that trend.
“Arctic sea ice this summer is similar to the past decade or so,” Bitz says, “which is to say that it is much lower than the previous 35 years, but not unusual for the ‘new normal.’”
Banner image: Marcel König and Natascha Oppelt (both from the University of Kiel, Germany) measuring melt pond optical properties during RV Polarstern campaign PASCAL, north of Svalbard in June 2017. Image courtesy of Niels Fuchs, University of Hamburg.
Citations:
Buckley, E. M., Farrell, S. L., Herzfeld, U. C., Webster, M. A., Trantow, T., Baney, O. N., … Lawson, M. (2023). Observing the evolution of summer melt on multiyear sea ice with ICESat-2 and Sentinel-2. The Cryosphere, 17(9), 3695-3719. doi:10.5194/tc-17-3695-2023
Fuchs, N., Von Albedyll, L., Birnbaum, G., Linhardt, F., Oppelt, N., & Haas, C. (2024). Sea ice melt pond bathymetry reconstructed from aerial photographs using photogrammetry: A new method applied to mosaic data. The Cryosphere, 18(7), 2991-3015. doi:10.5194/tc-18-2991-2024
Lee, S., Stroeve, J., Webster, M., Fuchs, N., & Perovich, D. K. (2024). Inter-comparison of melt pond products from optical satellite imagery. Remote Sensing of Environment, 301, 113920. doi:10.1016/j.rse.2023.113920
FEEDBACK: Use this form to send a message to the author of this post. If you want to post a public comment, you can do that at the bottom of the page.
Credits
