The technique in question is data assimilation, the mathematical machinery by which numerical models are continuously corrected with real-world observations. The team focused on sea ice concentration, the fraction of ocean surface covered by ice, as measured by passive microwave satellite sensors and processed with the NASA Team algorithm at the National Snow and Ice Data Center. These observations were fed into the sea ice component of the Geophysical Fluid Dynamics Laboratory’s global ice-ocean model using the Data Assimilation Research Testbed, a community software facility. At the heart of the system sits an Ensemble Adjustment Kalman Filter, a statistical method that blends a spread of model simulations with observations, weighting each according to its uncertainty. With a horizontal localization radius of roughly 170 kilometers, the filter directly updated the concentration of each of the model’s five sea ice thickness categories, while the aggregate thicknesses of ice and snow adjusted implicitly in response. The result was a reanalyzed sea ice state, spanning 1992 to 2017, that reflects both the physics of the model and the reality captured from orbit.

To test whether this improved initial state actually translates into better forecasts, the researchers ran two parallel suites of reforecast experiments with the SPEAR_MED prediction system, part of GFDL’s Seamless System for Prediction and EArth System Research. SPEAR couples the Modular Ocean Model version 6, the Sea Ice Simulator version 2, an atmospheric model, and a land model, all sharing a nominal one-degree resolution for ocean and ice. The baseline suite, called SPEAR, was initialized the standard way: ocean conditions came from an ocean data assimilation experiment, while the atmosphere was nudged toward the NCEP Climate Forecast System Reanalysis and sea surface temperature toward NOAA’s Optimal Interpolation product. The second suite, called SICDA, was identical in every respect except that its sea ice initial conditions came from the new assimilation product. Each experiment comprised fifteen ensemble members, initialized on the first day of every month from 1992 through 2017 and integrated for a full year, with the analysis focused on the first 45 days.

The comparison of initial conditions alone was striking. The standard SPEAR initial state overestimated Antarctic sea ice extent in most seasons, with the largest positive biases in winter and spring across the Weddell, Amundsen-Bellingshausen, Indian, and West Pacific sectors, while underestimating summer extent in the West Pacific and Ross sectors. After assimilating sea ice concentration, these biases shrank substantially in nearly all months and sectors. The wintertime and springtime positive biases in the Weddell and Ross sectors, regions notorious for the open-water polynyas that the unperturbed model simulated too actively, were almost entirely eliminated. The assimilated state also tracked the observed year-to-year variability of regional sea ice extent more faithfully, showing higher detrended anomaly correlation coefficients, and it reduced the root-mean-square error of grid-cell-level concentration almost everywhere, with the greatest gains in the peripheral zone where ice grows and retreats seasonally.

An important and somewhat cautionary side effect emerged in the ice thickness fields. Because the underlying model carried a systematic positive bias in concentration, the assimilation tended to reduce concentration, and thickness followed in the same direction. The assimilated initial state was thinner than the standard one nearly everywhere, with exceptions in the interior Weddell and Ross sectors, where the ice actually thickened. Lacking direct observations of Antarctic sea ice thickness, the team benchmarked their results against two independent model-based references, the Global Ice-Ocean Modeling and Assimilation System and the Southern Ocean State Estimate, both of which reproduce the broad patterns of Antarctic ice volume despite their own biases. The comparison revealed that the four datasets disagree considerably on the magnitude and even the timing of the annual maximum in ice volume, underscoring how poorly constrained Southern Hemisphere ice thickness remains.

When the two forecast suites were evaluated against satellite observations, the assimilation paid clear dividends. Measured by the detrended anomaly correlation coefficient of sea ice concentration, SICDA outperformed SPEAR in nearly all initialization months, with the largest gains in austral spring, followed by winter and summer, and only minor differences in autumn. Both dynamical systems initially trailed a statistical benchmark, a damped anomaly persistence forecast that extrapolates the most recent observed anomaly while accounting for its expected decay, but they overtook it within the first week when averaged across the year. Notably, the skill advantage of the assimilated forecasts decayed gradually with lead time rather than vanishing immediately, and the magnitude of improvement was larger in the Antarctic than in parallel Arctic experiments, where the greatest benefit had previously been found in summer rather than spring.

The improvements extended to the sea ice edge, arguably the most operationally relevant forecast target for ships navigating ice-infested waters. The team quantified edge skill with the Spatial Probability Score, normalized against a trend climatology so that values below one indicate skill exceeding climatology. In winter and spring-initialized forecasts, the standard SPEAR system showed essentially no edge skill, with normalized scores at or above one throughout the 45-day window, reflecting large initial errors in edge position. SICDA transformed this picture: its scores were significantly lower than SPEAR’s for the entire 45 days in winter and about 40 days in spring, and it began beating the damped persistence benchmark around day 15 in winter. Averaged across all twelve initialization months, SICDA maintained significantly better edge skill than SPEAR for 35 days in the Antarctic, compared with only 22 days in the Arctic, and it outperformed the benchmark after roughly day 10 annually, a feat the standard system never achieved in the Southern Ocean.

The study is candid about its one notable failure mode. In forecasts initialized in December and January, SICDA initially led SPEAR but lost ground after about 20 days, with the degradation concentrated along the Antarctic coast and in the interior Weddell and Ross sectors. The mechanisms differ between these regions. Along the coast, the assimilated state featured lower concentration and thinner ice, which through the ice-albedo feedback absorbed more solar radiation, warmed the sea surface by up to 0.8 degrees Celsius relative to the standard forecasts, and accelerated bottom melt, compounding the initial deficit. In the interior Weddell and Ross sectors, by contrast, the assimilated state carried more extensive and thicker ice, which slowed an already sluggish model melt rate and allowed positive concentration biases to grow. The authors suggest that adjusting highly coupled fields such as concentration and thickness through assimilation can introduce inconsistencies that only surface at longer lead times, a caution relevant to any data assimilation system.

The broader implications reach beyond the Southern Ocean. Previous assessments of subseasonal prediction systems in the S2S database found that only one system outperformed simple benchmarks for Antarctic ice edge forecasts, and that all systems fared worse in the Antarctic than the Arctic, a gap attributed largely to deficient initial conditions. By showing that assimilating a single, widely available satellite observable can close much of that gap, the study offers a relatively inexpensive path to better polar forecasts, and the authors note that their assimilated system performs comparably to the leading European Centre for Medium-Range Weather Forecasts system in both hemispheres. The work also highlights a persistent observational deficit: without ground truth for ice thickness, it remains impossible to confirm whether the assimilation adjusts volume in the right direction, strengthening the case for expanded thickness measurements from satellites, aircraft, and ships. As Antarctic sea ice enters what may be a new, lower state, the ability to anticipate its fluctuations weeks in advance has shifted from an academic curiosity to a practical necessity, and this study shows that the tools to meet that challenge are already within reach.

Subject of Research: Improving subseasonal Antarctic sea ice forecasts through satellite sea ice concentration data assimilation

Article Title: Subseasonal forecast improvements from sea ice concentration data assimilation in the Antarctic

Article References: Zhang, Y.-F., Bushuk, M., Winton, M., Gregory, W., Hurlin, B., Jia, L., & Lu, F. (2026). Subseasonal forecast improvements from sea ice concentration data assimilation in the Antarctic. The Cryosphere, 20(10), 5867-5882. https://doi.org/10.5194/tc-20-5867-2026

Image Credits: AI Generated

DOI: 10.5194/tc-20-5867-2026

Keywords: Antarctic sea ice, data assimilation, subseasonal forecasting, sea ice concentration, sea ice edge, SPEAR model, GFDL, The Cryosphere, sea ice thickness, Kalman filter, Southern Ocean, polar prediction

The tool is the Snow Microwave Radiative Transfer model, known as SMRT, which has been developed and refined by the snow remote-sensing community since 2015. Originally built to simulate passive microwave emission from snow, SMRT evolved into a general-purpose framework capable of handling a wide range of cryospheric environments and sensors. It already included a module for conventional low-resolution-mode altimetry, the older style of radar altimetry in which the satellite illuminates a broad, pulse-limited footprint. What it lacked was support for the delay-Doppler, or synthetic aperture radar, processing mode that most modern sensors, including CryoSat-2 and Sentinel-3, actually use. The new version, SMRT 1.7, fills that gap with a dedicated SAR altimetry module described in the journal Geoscientific Model Development.

The significance of the delay-Doppler technique lies in how it reshapes the radar echo. Rather than letting the pulse spread across a wide oval footprint, the satellite exploits the Doppler shift of the returning signal to synthetically focus the beam along its flight direction, shrinking the illuminated strip to roughly 300 metres in the along-track direction. This improves the precision of elevation measurements and reduces the blurring effect of sloped or rugged terrain. But the focusing comes at a price: the resulting waveform has a distinctly different shape from the classic ocean-altimetry curve described by Brown’s famous 1977 model, and interpreting it over snow and ice demands purpose-built physics.

Picard and his colleagues, working with partners in Canada, Norway, Italy, and France, took a deliberately modular approach. Their key insight is that the echo can be decomposed into two nearly independent pieces that are later combined. The first piece describes the vertical journey of the wave into the snowpack, capturing the echoes produced by the surface, by internal interfaces between layers, and by volume scattering within the snow itself. The second piece describes the horizontal spreading of the spherical radar wave over the curved Earth and the filtering imposed by the SAR processing. The first component already existed in SMRT; the second required the team to survey the scientific literature and implement not one but eight different delay-Doppler formulations, published between 2004 and 2019 by groups including Wingham and colleagues, Halimi and colleagues, Ray and colleagues, Boy and colleagues, Buchhaupt and colleagues, Dinardo and colleagues, and Landy and colleagues.

Why eight? Because, as the team’s careful review makes clear, there is no single best model. The formulations differ in how they represent surface topography, whether they treat the terrain statistically as a probability distribution of heights or deterministically as a digital elevation model, how they approximate the antenna pattern and the compressed pulse shape, and how they balance analytical elegance against computational speed. Some models, such as those of Dinardo and Buchhaupt, are extremely fast and well suited to physical retracking algorithms that fit observed waveforms in near real time. Others, such as the facet-based model of Landy, can ingest a real digital elevation model and capture the electromagnetic bias caused by skewed slope distributions, but at a far greater computational cost. By bundling all eight into a single plug-and-play framework, SMRT lets researchers compare them directly and pick the right tool for the job.

The technical machinery works in three steps. First, existing SMRT modules compute the backscatter of every layer and interface, drawing on theories such as the Improved Born Approximation for volume scattering and the Geometrical Optics approximation for rough surfaces. A new addition handles coherent, quasi-specular backscatter, which matters when the surface is smooth, as it often is over sea ice, and which depends in a subtle way on the altitude of the sensor. Second, the user-selected delay-Doppler model computes a delay-Doppler map for each interface and for the volume, applying a slant-range correction that migrates echoes arriving at off-nadir angles back to their correct range. Third, the pieces are combined into the final delay-Doppler map, which is summed along the Doppler dimension to yield the waveform a satellite would record.

Verification proceeded on several fronts. Under simplified conditions, all eight models produced waveforms that peaked within the same range gate, with only small differences in the leading and trailing edges attributable to numerical and theoretical approximations. The team’s implementation of the Landy model matched the original Matlab code to within one percent. And in a particularly elegant test, the researchers disabled the slant-range correction to produce pseudo-low-resolution-mode waveforms and compared them against SMRT’s independent LRM module, finding near-perfect agreement with the Wingham18 formulation. These cross-checks matter because a modelling error of even a few percent in the peak position could be misread as a real change in surface elevation.

The real-world test came on the Antarctic ice sheet. The team drove the model with in-situ measurements collected during four field campaigns, including density and specific surface area profiles from snow cores, temperature readings, and photogrammetric estimates of centimetre-scale surface roughness, supplemented by the Reference Elevation Model of Antarctica for large-scale terrain statistics. Comparing the simulations with Sentinel-3 Ku-band waveforms at eighteen sites, they found that the model reproduced the observed latitudinal gradient in waveform amplitude, with stronger echoes over the smooth, wind-scoured plateau near Dome C and weaker ones in the rougher coastal margins. When the poorly constrained radar-scale roughness was optimized at each site, the agreement improved markedly, confirming that surface roughness exerts first-order control on how much of the echo comes from the snow surface versus the snow volume beneath it.

The study is candid about its limits. The current implementation assumes the medium is homogeneous across a footprint spanning many kilometres, neglects multiple scattering between layers, and covers only unfocused SAR processing, leaving fully focused and interferometric modes for future work. Multiple scattering is likely to become significant at the Ka-band frequency of the upcoming Copernicus CRISTAL mission, where the team’s calculations suggest single scattering accounts for less than half the backscatter in the top metre of Antarctic snow. Even so, the payoff is substantial. With a verified, open-source framework that links detailed snow physics to realistic SAR waveforms, scientists can now attack long-standing problems such as the penetration bias that offsets ice-sheet elevation measurements, the validity of assuming that radar echoes from sea ice originate at the snow-ice interface, and the design of retrieval algorithms for snow depth on sea ice and thickness of lake ice. As new missions like CRISTAL and Sentinel-6 come online, the ability to simulate exactly what their radars should see, given what we know about the snow and ice below, may prove decisive for turning polar radar echoes into climate-grade measurements.

Subject of Research: Simulation of SAR radar altimeter echoes from snow and ice surfaces using the SMRT model

Article Title: Simulating SAR altimeter echoes from cryospheric surfaces with the Snow Microwave Radiative Transfer (SMRT) model version 1.7

Article References: Picard, G., Murfitt, J., Zakharova, E., Zeiger, P., Arnaud, L., Aublanc, J., Landy, J. C., Scagliola, M., & Duguay, C. (2026). Simulating SAR altimeter echoes from cryospheric surfaces with the Snow Microwave Radiative Transfer (SMRT) model version 1.7. Geoscientific Model Development, 19(18), 9103-9129. https://doi.org/10.5194/gmd-19-9103-2026

Image Credits: AI Generated

DOI: 10.5194/gmd-19-9103-2026

Keywords: radar altimetry, SMRT model, cryosphere, Antarctica, Sentinel-3, delay-Doppler, sea ice thickness, ice sheet elevation, snow microwave radiative transfer, waveform retracking, Ku-band radar, CRISTAL

In the Arctic, the transformation is unambiguous and accelerating. The melt season has lengthened by approximately 7.4 days per decade between 1979 and 2024, a shift that compounds year after year as earlier snowmelt and later freeze-up expose dark ocean water to sunlight for longer periods. Winter sea-ice thickness has declined to a mean total of roughly 1.6 metres over 1980 to 2023, down from values that once routinely exceeded three metres in the central Arctic. Summer surface albedo, the fraction of incoming solar radiation reflected back to space, has fallen by about 0.03 per decade over 1979 to 2020, while spring snow depth on the ice has thinned by 2.5 centimetres per decade over the period 1954 to 2024. Each of these trends feeds the others in a self-reinforcing cascade.

The physical mechanism behind this cascade is the ice-albedo feedback. Fresh snow reflects up to ninety percent of incident sunlight, but once the snow melts, bare ice reflects far less, and melt ponds pooling on the surface reflect less still. As the Arctic melt season lengthens, more solar energy is absorbed by the ice-ocean system, warming the upper ocean and thinning the ice from below. Thinner ice breaks up more easily, creating more open water, which absorbs more heat and delays freeze-up further. Satellite records show that the Arctic’s once-dominant multiyear ice, ice that survives at least one summer melt, has been progressively replaced by thinner, more saline first-year ice that melts more readily. A regime shift in Arctic Ocean ice thickness has been documented, and the age structure of the ice pack has shifted decisively toward young ice.

The Antarctic tells a strikingly different story. Antarctic sea ice sits at the edge of a vast, cold continent surrounded by a circumpolar ocean, and its thickness is limited by rapid drift away from the coast and by heavy snow loading that can push the ice surface below sea level, flooding it and forming snow-ice. Unlike the Arctic, the Antarctic record through the satellite era showed a slight overall increase in extent through 2014, followed by abrupt declines, including record lows in 2017 and again in 2023 that have led some researchers to argue the region may have entered a new sea-ice state. Because regional trends in the Antarctic point in different directions in different sectors and different decades, the hemisphere-wide changes in ice properties are smaller than those observed in the Arctic, and the underlying drivers remain contested.

Snow plays a fundamentally different role at each pole. In the Arctic, snow insulates the ice from the cold atmosphere in winter, slowing growth, but its high albedo protects the ice in spring. Declining snow depth therefore removes a protective layer and accelerates surface melt. In the Antarctic, thick snow cover frequently depresses the ice surface below the waterline, and the resulting slush refreezes into snow-ice, adding mass from above. Antarctic snow also modulates the penetration of light into the ice and upper ocean, shaping the timing and productivity of ice-algal blooms that anchor polar marine food webs. Recent work shows that summer snowfall events in the Arctic, increasingly modulated by the Arctic Oscillation, can temporarily brighten the surface and slow melt, while rain-on-snow events darken it and hasten melt onset, making precipitation a critical and underappreciated player in the seasonal ice budget.

Dynamically, both hemispheres are becoming more restless. Sea-ice motion has increased by 0.63 centimetres per second per decade in the Arctic between 1978 and 2024 and by 0.69 centimetres per second per decade in the Antarctic between 1982 and 2024. Faster drift is partly a consequence of thinner, weaker ice that deforms more readily under wind and ocean stress, and partly a response to changing atmospheric circulation patterns. In the Arctic, accelerated drift increases export of ice through Fram Strait into the North Atlantic, draining the ice pack and contributing to the stepwise reduction of multiyear ice area since 1980. Smoother ice with fewer pressure ridges has been observed in a more dynamic Arctic, which reduces surface drag and further enhances drift speeds, another positive feedback loop.

Polynyas, recurring areas of open water within the ice pack, reveal some of the sharpest inter-hemispheric contrasts. Antarctic coastal polynyas, sustained by fierce katabatic winds off the ice sheet, are engines of sea-ice production and of Antarctic Bottom Water formation, the densest water mass in the global overturning circulation. Their occurrence and extent show regionally diverging trends, with emerging long-term trends and interdecadal cycles documented across the continent. In the Arctic, polynyas such as those in the Canadian Arctic Archipelago and the Siberian shelves sustain hyperproductive ecosystems and contribute to intermediate and deep water formation, but their trends differ by region and are tied to distinct atmospheric and oceanic drivers. Offshore polynyas in the Antarctic, including the famous Weddell Polynya of the 1970s and its intermittent modern successors, are linked to Southern Hemisphere climate anomalies and to ocean heat ventilation, and recent extremes in Antarctic sea-ice extent have been modulated by this ventilation of ocean heat.

The consequences ripple far beyond the poles. Arctic amplification, the phenomenon by which the Arctic has warmed nearly four times faster than the globe since 1979, is substantially driven by sea-ice loss and the associated albedo feedback. Observational studies have quantified the radiative heating contributed by vanishing Arctic ice, and the loss of sea ice alters air-sea exchanges of heat, moisture and momentum, with implications for mid-latitude weather patterns. Ecologically, earlier melt onset and longer open-water seasons disrupt the tight phenological coupling between ice algae, zooplankton, fish, seabirds and marine mammals that have evolved around the seasonal ice cycle. In the Antarctic, strengthening snow and ice albedo feedback driven by recent sea-ice loss has now been observed, suggesting the Southern Ocean may be catching up to the Arctic in its climatic significance.

Looking forward, the review highlights major knowledge gaps and calls for joint model-observation efforts to close them. Observationally constrained projections indicate the Arctic could see its first ice-free summer, and even its first ice-free day, before 2030 under low emission scenarios, with the ice-free season projected to extend deep into autumn by century’s end. Antarctic projections remain far more uncertain because current climate models struggle to reproduce the observed variability and recent structural change in the Southern Ocean sea-ice system, and because snow depth, freeboard retrieval and thickness estimates from satellite altimetry carry large uncertainties in the south. The authors recommend coordinated campaigns combining satellite altimetry from ICESat-2 and CryoSat-2, autonomous buoy networks, ship-based observations and improved climate models to constrain snow depth, thickness, albedo and drift together. Only by treating the two polar ice covers as a coupled, hemispherically contrasted system, they argue, can scientists anticipate how the remaining sea ice will behave as the twenty-first century unfolds, and what that behaviour will mean for the climate, ecosystems and communities that depend on it.

Subject of Research: Long-term changes in the physical properties and processes of Arctic and Antarctic sea ice

Article Title: Changes in Arctic and Antarctic sea-ice properties and processes

Article References: Webster, M. A., Arndt, S., Bliss, A., Kacimi, S., Maksym, T., Massonnet, F., Riihelä, A., & Toyota, T. (2026). Changes in Arctic and Antarctic sea-ice properties and processes. Nature Reviews Earth & Environment. https://doi.org/10.1038/s43017-026-00816-9

Image Credits: AI Generated

DOI: 10.1038/s43017-026-00816-9

Keywords: sea ice, Arctic, Antarctic, climate change, albedo, snow depth, sea-ice thickness, polynyas, melt season, satellite observations, cryosphere, polar climate