Explore the vast range of titles available.

We present an initial assessment of all-season Arctic sea ice thickness estimates from ICESat-2 by combining freeboard retrievals with all-season SnowModel-LG snow loading. ICESat-2 captures the key regional and seasonal patterns of Arctic sea ice variability and shows good agreement with CryoSat-2 all-season estimates, including regional patterns of inter-annual variability in summer ice thickness. ICESat-2 shows consistently thicker ice compared to CryoSat-2 across the western coastal Arctic, while CryoSat-2 shows some periods of thicker ice across the Central Arctic, largely consistent with winter thickness biases. Validation against upward-looking sonar moorings, IceBird-2019 airborne observations and MOSAiC buoy data highlights generally strong performance across a range of conditions, although seasonal biases linked to snow loading, freeboard differences and ice density assumptions persist. The SnowModel-LG and NESOSIM snow accumulation models perform well across the validation datasets, but do not consistently add skill beyond the modified Warren climatology. Experimental ICESat-2/CryoSat-2 dual altimetry winter snow depths show strong performance relative to existing products and future work should extend these into summer for further assessments. Overall, our analysis supports the viability of an all-season ICESat-2-derived thickness record.

Lead/floe discrimination is essential for calculating sea ice freeboard and thickness (SIT) from radar altimetry. During the summer months (May–September) the classification is complicated by the presence of melt ponds. In this study, we develop a neural network to classify CryoSat-2 measurements during the summer months, building on the work by Dawson et al. (2022) with various improvements: (i) we expand the training dataset and make it more geographically and seasonally diverse, (ii) we introduce an additional radar detectable class for thinned floes, (iii) we design a deeper neural network and train it longer and (iv) we update the input parameters to data from the latest publicly available CryoSat-2 processing baseline. We show that both the expansion of the training data and the novel architecture increase the classification accuracy. The overall test accuracy improves from 77 ± 5 % to 84 ± 2 % and the lead user accuracy increases from 82 ± 10 % to 88 ± 5 % with the novel classifier. When used for SIT calculation, we observe minor improvements in agreement with the validation data. However, as more leads are detected with the new approach, we achieve better coverage especially in the marginal ice zone. The novel classifier presented here is used for the Summer Sea Ice CryoTEMPO (CryoSat Thematic Product).

Arctic sea ice is diminishing with climate warming 1 at a rate unmatched for at least 1,000 years 2 . As the receding ice pack raises commercial interest in the Arctic 3 , it has become more variable and mobile 4 , which increases safety risks to maritime users 5 . Satellite observations of sea-ice thickness are currently unavailable during the crucial melt period from May to September, when they would be most valuable for applications such as seasonal forecasting 6 , owing to major challenges in the processing of altimetry data 7 . Here we use deep learning and numerical simulations of the CryoSat-2 radar altimeter response to overcome these challenges and generate a pan-Arctic sea-ice thickness dataset for the Arctic melt period. CryoSat-2 observations capture the spatial and the temporal patterns of ice melting rates recorded by independent sensors and match the time series of sea-ice volume modelled by the Pan-Arctic Ice Ocean Modelling and Assimilation System reanalysis 8 . Between 2011 and 2020, Arctic sea-ice thickness was 1.87 ± 0.10 m at the start of the melting season in May and 0.82 ± 0.11 m by the end of the melting season in August. Our year-round sea-ice thickness record unlocks opportunities for understanding Arctic climate feedbacks on different timescales. For instance, sea-ice volume observations from the early summer may extend the lead time of skilful August–October sea-ice forecasts by several months, at the peak of the Arctic shipping season. Deep learning and numerical simulations of CryoSat-2 radar altimeter data are used to generate a pan-Arctic sea-ice thickness dataset for the Arctic melt period.

Because of a spring predictability barrier, the seasonal forecast skill of Arctic summer sea ice is limited by the availability of melt‐season sea ice thickness (SIT) observations. The first year‐round SIT observations, retrieved from CryoSat‐2 from 2011 to 2020, are assimilated into the GFDL ocean–sea ice model. The model's SIT anomaly field is brought into significantly better agreement with the observations, particularly in the Central Arctic. Although the short observational period makes forecast assessment challenging, we find that the addition of May–August SIT assimilation improves September local sea ice concentration (SIC) and extent forecasts similarly to SIC‐only assimilation. Although most regional forecasts are improved by SIT assimilation, the Chukchi Sea forecasts are degraded. This degradation is likely due to the introduction of negative correlations between September SIC and earlier SIT introduced by SIT assimilation, contrary to the increased correlations found in other regions. Plain Language Summary The dramatic decline of Arctic sea ice, especially in summer, has received a lot of attention. The ability to better predict Arctic summer sea ice several months ahead of time will help decision making on protecting local communities and ecosystems and regulating economic activities in the Arctic. Climate dynamical models have shown reasonable skill in predicting Arctic summer sea ice on seasonal timescales, but also contain considerable errors. Integrating observed sea ice thickness conditions into the model in the summer melt season has a large potential to reduce such errors. The prediction skill of summer Arctic sea ice initialized before June 1st is found to be notably lower than that initialized afterward, which is known as the spring predictability barrier. Hence constraining initial conditions post‐June has great implications for summer Arctic sea ice predictions. This study combines a new year‐round satellite sea ice thickness observational product with the sea ice and ocean dynamical model at GFDL and examines its impact on the seasonal prediction of Arctic sea ice. We find that the prediction skill has been improved in general, although some uncertainties exist due to the limited temporal availability of the observations. Key Points The representation of Arctic sea ice volume anomalies is significantly improved by assimilating year‐round sea ice thickness (SIT) observations from CryoSat‐2 Arctic summer sea ice prediction skill is generally improved when initial conditions are constrained by satellite SIT observations The Arctic summer sea ice in the 2010s decade is particularly hard to predict due to anomalously low correlation between volume and extent

Satellite observations of sea ice freeboard are integral to the estimation of sea ice thickness. It is commonly assumed that radar pulses from satellite‐mounted Ku‐band altimeters penetrate through the snow and reflect from the snow‐ice interface. We would therefore expect a negative correlation between snow accumulation and radar freeboard measurements, as increased snow loading weighs the ice floe down. In this study we produce daily resolution radar freeboard products from the CryoSat‐2 and Sentinel‐3 altimeters via a recently developed optimal interpolation scheme. We find statistically significant (p < 0.05) positive correlations between radar freeboard anomalies and modeled snow accumulation. This suggests that, in the period after snowfall, radar pulses are not scattering from the snow‐ice interface as commonly assumed. Our results offer satellite‐based evidence of winter Ku‐band radar scattering above the snow‐ice interface, violating a key assumption in sea ice thickness retrievals. Plain Language Summary Arctic sea ice thickness is often estimated using radar pulses from satellite‐mounted Ku‐band altimeters, which retrieve the radar freeboard. This is a proxy for the height of the ice surface above the waterline. Ku‐band radar pulses are widely assumed to penetrate through the overlying snowpack and reflect from the top of the sea ice. This means that increased snow loading on a sea ice floe is expected to reduce its radar freeboard, as the snow weighs the sea ice down. We produce daily resolution pan‐Arctic radar freeboard data sets from CryoSat‐2 and Sentinel‐3 retrievals. Using these new products, we find that an increased snow load often increases the radar freeboard, suggesting that the radar pulses are not reflecting off the ice surface. This could explain why satellite‐based sea ice thickness estimates don't always match in situ observations. Key Points We reveal synoptic‐scale positive correlations between snow accumulation and Ku‐band radar freeboard change These correlations indicate that the conventional assumption of full radar‐wave penetration of the snowpack does not always hold true This sensitivity in freeboard estimates to snow accumulation introduces synoptic variability into satellite estimates of sea ice thickness

The Northwest Passage through the Canadian Arctic Archipelago (CAA) provides a shorter transit route connecting the Atlantic Ocean to the Pacific Ocean but ever-present sea ice has prevented its practical navigation. Sea ice area in the northern route of the Northwest Passage on 30 September 2024 fell to a minimum of 4x10.sup.3 km.sup.2 or â¼3 % of its total area the lowest ice area observed since 1960. Here, we investigate the processes responsible for the record low sea ice area in 2024 and show it was driven by a perfect sequence of thermodynamic and dynamic forcing events acting on an increasingly less resilient ice cover. Specifically, multi-year ice (MYI) only made up â¼10 % of total sea ice area at the start of the melt season that was characterized by an atmospheric circulation pattern that brought warm southerly air directly into the middle of CAA. This resulted in a record summer air temperature anomaly of 2.1 °C that drove rapid melt and limited the import of ice from higher latitude regions to 50 % of the 2016-2024 mean. Finally, positive air temperature anomalies upwards of 12 °C persisted into October, extending the melt season further and delaying freeze-up by 1 month, compared to the 1991-2020 baseline. Overall, a series of specific and cascading thermodynamic and dynamic processes is required to melt all ice in the northern route of Northwest Passage as it did in 2024, therefore ice conditions along this route will likely continue to remain highly variable during the transition to a summertime sea ice free Arctic.

Pan-Arctic sea ice thickness has been monitored over recent decades by satellite radar altimeters such as CryoSat-2, which emits Ku-band radar waves that are assumed in publicly available sea ice thickness products to penetrate overlying snow and scatter from the ice–snow interface. Here we examine two expressions for the time delay caused by slower radar wave propagation through the snow layer and related assumptions concerning the time evolution of overlying snow density. Two conventional treatments introduce systematic underestimates of up to 15 cm into ice thickness estimates and up to 10 cm into thermodynamic growth rate estimates over multi-year ice in winter. Correcting these biases would impact a wide variety of model projections, calibrations, validations and reanalyses.

The CryoSat-2 radar altimeter and ICESat-2 laser altimeter can provide complementary measurements of the freeboard and thickness of Arctic sea ice. However, both sensors face significant challenges for accurately measuring the ice freeboard when the sea ice is melting in summer months. Here, we used crossover points between CryoSat-2 and ICESat-2 to compare elevation retrievals over summer sea ice between 2018–2021. We focused on the electromagnetic (EM) bias documented in CryoSat-2 measurements, associated with surface melt ponds over summer sea ice which cause the radar altimeter to underestimate elevation. The laser altimeter of ICESat-2 is not susceptible to this bias but has other biases associated with melt ponds. So, we compared the elevation difference and reflectance statistics between the two satellites. We found that CryoSat-2 underestimated elevation compared to ICESat-2 by a median difference of 2.4 cm and by a median absolute deviation of 5.3 cm, while the differences between individual ICESat-2 beams and CryoSat-2 ranged between 1–3.5 cm. Spatial and temporal patterns of the bias were compared to surface roughness information derived from the ICESat-2 elevation data, the ICESat-2 photon rate (surface reflectivity), the CryoSat-2 backscatter, and the melt pond fraction derived from Sentinel-3 Ocean and Land Color Instrument (OLCI) data. We found good agreement between theoretical predictions of the CryoSat-2 EM melt pond bias and our new observations; however, at typical roughness <0.1 m the experimentally measured bias was larger (5–10 cm) compared to biases resulting from the theoretical simulations (0–5 cm). This intercomparison will be valuable for interpreting and improving the summer sea ice freeboard retrievals from both altimeters.

Pressure ridges, formed by sea ice deformation, affect momentum transfer in the Arctic Ocean and support a larger biomass than the surrounding-level ice. Although trends in Arctic sea ice thickness and concentration are well documented, changes in ridge morphology remain unclear. This study provides airborne-based evidence of a shift towards a smoother ice surface, with fewer pressure ridges and reduced surface drag, attributed to the loss of old ice. Furthermore, an increase in seasonal ice cover enhances overall deformation in the Arctic and acts as a negative feedback mechanism on pan-Arctic ridge morphology: the greater the proportion of seasonal ice, the higher the pan-Arctic mean ridge rate, dampening an overall decline in ridges with age. While thinner and less frequent ridges benefit industries such as shipping, these changes are likely to have profound impacts on the energy and mass balance and the ecosystem of the Arctic Ocean. Pressure ridges, a characteristic feature of Arctic sea ice, play an important role in the ecosystem but pose challenges to shipping. Here the authors use aircraft measurements to document a decline in both the frequency and height of these pressure ridges in recent decades.

The EU and ESA plan to launch a dual-frequency Ku- and Ka-band polar-orbiting synthetic aperture radar (SAR) altimeter, the Copernicus Polar Ice and Snow Topography Altimeter (CRISTAL), by 2027 to monitor polar sea ice thickness (SIT) and its overlying snow depth, among other applications. However, the interactions of Ku- and Ka-band radar waves with snow and sea ice are not fully understood, demanding further research effort before we can take full advantage of the CRISTAL observations. Here, we use three ongoing altimetry missions to mimic the sensing configuration of CRISTAL over Arctic sea ice and investigate the derived snow depth estimates obtained from dual-frequency altimetry. We apply a physical model for the backscattered radar altimeter echo over sea ice to CryoSat-2's (CS2's) Ku-band altimeter in SAR mode and to the SARAL mission's AltiKa (AK) Ka-band altimeter in low-resolution mode (LRM), and then we compare it to reference laser altimetry observations from ICESat-2 (IS2). ICESat-2 snow freeboards (snow + sea ice) are representative of the air–snow interface, whereas the radar freeboards of AltiKa are expected to represent a height at or close to the air–snow interface, and CryoSat-2 radar freeboards are expected to represent a height at or close to the snow–ice interface. The freeboards from AltiKa and ICESat-2 show similar patterns and distributions; however, the AltiKa freeboards do not thicken at the same rate over winter, implying that Ka-band height estimates can be biased low by 10 cm relative to the snow surface due to uncertain penetration over first-year ice in spring. Previously observed mismatches between radar freeboards and independent airborne reference data have frequently been attributed to radar penetration biases, but they can be significantly reduced by accounting for surface topography when retracking the radar waveforms. Waveform simulations of CRISTAL in its expected sea ice mode reveal that the heights of the detected snow and ice interfaces are more sensitive to multi-scale surface roughness than to snow properties. For late-winter conditions, the simulations suggest that the CRISTAL Ku-band radar retrievals will track a median elevation 3 % of the snow depth above the snow–ice interface because the radar return is dominated by surface scattering from the snow–ice interface which has a consistently smoother footprint-scale slope distribution than the air–snow interface. Significantly more backscatter is simulated to return from the air–snow interface and snow volume at Ka band, with the radar retrievals tracking a median elevation 10 % of the snow depth below the air–snow interface. These model results generally agree with the derived satellite radar freeboards, which are consistently thicker for AltiKa than CryoSat-2, across all measured snow and sea ice conditions.