Pan-Arctic sea ice thickness has been monitored by the European Space Agency SAR altimeter CryoSat-2 since 2010. However, sea ice thickness observations are so far restricted to Arctic winter months (Oct-Apr) because conventional processing techniques fail to work when meltwater ponds accumulate at the ice surface during summer months. Conventional radar waveform classifiers cannot differentiate melt pond-covered ice floes from leads and the ponds introduce a bias to the radar range measurement. Our current understanding of basin-scale sea ice melting patterns during summer are limited to weakly constrained ice-ocean model simulations, at a time when ice thickness observations would be most valuable for sea ice forecasting.

Here we will present a new method for generating valid sea ice thickness observations in Arctic summer months (May-Sept) over the ten-year period between 2011 and 2020. We use 100s of optical and SAR images coinciding with CryoSat-2 orbits to train a 1D convolutional neural network for separating altimeter returns from melting ice floes and leads. Pan-Arctic sea ice radar freeboards are mapped from the difference between floe and ocean surface elevations. A set of radar echo simulations are then performed to quantify the range bias caused by melt ponds dominating the reflected altimeter signal. Corrected sea ice freeboards are converted to thickness by accounting for the ice density and the residual snow load accumulated on the ice. We find that CryoSat-2 sea ice thickness observations capture the spatial patterns of sea ice thickness in July-September measured by airborne EM-bird sensors on Alfred Wegener Institute campaigns. The seasonal cycle of CryoSat-2 observations also matches well to ice draft observations from the Beaufort Gyre Exploration Program mooring sonars between May and Sept in 2011-2018 and to the Pan-Arctic Ice Ocean Modeling and Assimilation System (PIOMAS) reanalysis. Finally, we will demonstrate how summer sea ice thickness observations have a longer ‘memory’ than ice concentration observations, controlling the persistence of ice area anomalies over many months. CryoSat-2 summer sea ice thickness observations can therefore benefit seasonal forecasts of sea ice extent, skilfully predicting ice extent at lead times up to 10 months, both on a pan-Arctic and regional basis.

Sea-ice thickness is a key factor and indicator in understanding the impact of the global climate change. Deriving basin-wide sea-ice thickness estimates from satellite laser and radar altimetry relies on freeboard measurements. The freeboard-to-thickness conversion in turn requires information of snow mass and the density of the sea-ice layer that have unknown spatio-temporal variabilities and trends directly translating into the uncertainty of decadal sea-ice thickness data records. In addition, inter-mission biases arise from, e.g., different sensor types and frequencies as well as varying footprint sizes affected by surface roughness across regions and seasons. Therefore, carrying out validation and inter-calibration studies is crucial for reliable and continuous observation of the Earth’s cryosphere. To achieve this, it is beneficial to have simultaneous measurements of freeboard, snow depth, and sea-ice thickness, which provide reference data for both direct satellite observations and geophysical target parameters.

Here, we present Alfred Wegener Institute’s (AWI) IceBird program, which is a series of fixed-wing aircraft campaigns to measure Arctic sea ice and to monitor its change. During two late-winter campaigns in the western Arctic Ocean in 2017 and 2019, we have carried out surveys with the unique scientific instrument configuration including an airborne laser scanner (ALS) for surface topography and freeboard measurements, a tethered electromagnetic induction sounding instrument (EM-Bird) for total (snow+ice) thickness measurements, and an ultrawideband frequency-modulated continuous-wave microwave radar to measure snow thickness. Therefore, we are able to observe all three bounding interfaces in the sea-ice–snow system in high resolution along survey tracks on regional scales.

During the ship-based drift expedition Multidisciplinary drifting Observatory for the Study of Arctic Climate (MOSAiC) between October 2019 and September 2020, helicopter surveys were carried out in high spatio-temporal resolution throughout the year, including the polar night, to measure freeboard and roughness with the ALS both in local grid pattern and in larger scale. Coincident EM-Bird ice thickness data and information from snow measurements on the ground will help linking these parameters and monitor them and their effect on satellite retrievals for a full seasonal cycle.

The individual parameters are important for describing and monitoring the state of the Arctic sea ice and validating retrievals from satellite data, but combined they offer further possibilities to characterise sea ice. By assuming isostatic equilibrium, we are able to estimate up-to-date bulk density values for different sea-ice types from the IceBird data and to derive a parametrisation of sea-ice bulk density based on sea-ice freeboard. These data allow us to explore spatio-temporal variations in sea-ice parameters observable from space and to evaluate the validity of the freeboard-to-thickness conversion in satellite altimetry through comparison against dedicated satellite overpasses and orbit collections.

About a third of Greenland’s total ice losses come from the Northwest sector, a sector that includes a large number of marine-terminating outlet glaciers, which have all experienced widespread retreat triggered by ocean-induced melting. Accurately estimating the mass balance of this region using satellite altimetry has therefore important implications and requires robust estimates of surface elevation change. Airborne campaigns across the Greenland Ice Sheet provide an extensive dataset to validate satellite radar altimetry estimates of surface elevation and elevation change.
Here, we measure changes in surface elevation in the Northwest sector of the Greenland Ice Sheet from CryoSat-2 between July 2010 and July 2021 and find that the surface has lowered at a rate of 21.9 ± 1.1 cm/yr on average over this period, with rapid thinning at the ice sheet margins at a rate of 46.9 ± 5.9 cm/yr. To validate this dataset, we compare our satellite radar rates of elevation change to airborne laser altimetry data from NASA’s Operation IceBridge. Overall, we find a good agreement between the two datasets with a mean difference of 6.7 cm/yr and standard deviation of 73 cm/yr. Differences between satellite radar and airborne laser rates of elevation change are larger in the coastal SARIn region of the sector with a mean difference of 9.5 cm/yr, while difference are almost negligible in the interior LRM region with differences of -0.7 cm/yr. In addition, we examine ASIRAS Ku-band radar and ALS laser data collected during the ESA CryoVEx campaigns of 2012, 2014 and 2016 in the Northwest sector to further investigate differences between satellite and airborne measurements of surface elevation.
We further compute mass change from our validated dataset of CryoSat-2 surface elevation change, and we show that the Northwest sector lost 456 ± 5.7 Gt of ice between July 2010 and July 2021. By partitioning our mass balance estimates into sub-regions of the Northwest sector, we can compare our mass balance estimate to independent estimates from gravimetry and the mass budget method across different spatial scales. We show that our altimetry estimate is the least negative across all regions, the gravimetry and mass budget estimates alternating in recording the largest ice losses. Our comparisons reveal that the spatial pattern of differences between mass balance estimates is complex, suggesting that discrepancies between techniques do not solely originate from one single region or technique. However, airborne campaigns such as ESA CryoVEx and NASA Operation IceBridge campaigns play an important role in investigating the existence of potential biases in satellite radar altimetry estimates of surface elevation change and their subsequent impact on mass balance.

Snow on the sea ice is an element that is still very poorly known and poorly observed. However, it plays an important role in climate change due to its albedo, and in the dynamics of the sea ice due to its insulating properties. This lack of knowledge is also one of the main factors limiting the measurement of sea ice thickness by altimetry.
A method to measure this snow depth was first proposed in [Guerreiro et al. 2016]. It is based on the comparison of two altimetry frequencies, the Ku frequency on board CryoSat-2, and the Ka frequency from Saral. Indeed, these frequencies have different snow penetration properties. This observation was one of the reasons for adding a second frequency to the Copernicus CRISTAL polar altimeter project.
Following this study, an Altimetric Snow Depth (ASD) product was developed in the Arctic as part of the ESA CryoSeaNice project. In the framework of the ESA CS+AO project, an equivalent snow depth product has also been computed in Antarctica. These products are presented in [Garnier et al. 2021].
The first objective of this presentation is to show the relevance of this approach and the contribution of these data to the measurement of sea ice thickness.
To do so, the ASD data are compared with different datasets from different sources: space data obtained with the AMSR-2 passive radiometer, model data and in-situ data.
The AMSR-2 data are currently the only other snow depth observations that provide monthly estimates of snow depth on the polar ice pack. The first version (Meier et al, 2018) available on the NSIDC website (https://nsidc.org/data/AU_SI12/versions/1) has the disadvantage that it only covers first-year ice. The Bremen AMSR-2 v1.0 product (Rostosky et al, 2018) is calculated on multi-year ice but only for the months of March and April (during the Operation Ice Bridge campaigns). Only the NSIDC product covers the Southern Hemisphere, but this is complete because in this region the pack ice is mostly first-year ice.
Furthermore, in the Arctic, we will compare the ASD data with the Warren99 modified climatology (Arctic only), the PIOMAS model, the NESOSIM model (Arctic only) and the CMEMS LIM-3 sea ice model. The performance of these snow depth solutions will be evaluated by comparison with: 1) several Operation Ice Bridge (OIB) airborne campaigns, 2) the 2017 ESA-CRYOsat Validation Experiment (CryoVex) campaign that includes the KAREN Ka-band airborne altimeter, and 3) Beaufort Gyre Exploration Project (BGEP) data. We will also present the impact of different snow depth solutions on sea ice thickness estimates.
Finally, the important potential of these results for the monitoring of sea ice evolution by the future Copernicus CRISTAL mission will be discussed.
Acknowledgements: This work was supported by the ESA CryoSeaNICE project, the ESA CryoSat+ Antarctica project and the CNES TOSCA CASSIS project.

Sea ice and its snow cover play key roles in Earth’s climate and satellite altimetry can provide basin-wide estimations of sea ice and snow thickness. These approaches to retrieval from satellite data are based on assumptions about the dominant scattering surfaces of the radar waves in order to estimate the elevations of air/snow and snow/ice interfaces. Often it is assumed that the Ka-band mainly scatters from the air/snow interface and that Ku- and lower frequencies scatter from the snow/ice interface.

In this presentation we summarise data gathered using nadir-looking surface-based radars and coincident field data from multiple campaigns including MOSAiC, CryoVEx and SIPEX 1, conducted in the Arctic and Antarctic between 2007 - 2020. We compare the snow and ice conditions and radar waveforms, across the campaigns. Our results demonstrate the complexity and variability of scattering observed in radar data gathered over 2-18 GHz, Ku-band (13 GHZ) and Ka-band (35 GHz) frequencies. We investigate the effects of the snow and sea ice properties and meteorological conditions as well as radar frequency and polarisation on the waveforms.

We find, for instance, that the snow/ice interface is not usually the dominant scattering surface in the Ku-band, but that it was for a set of very low-density snow pits over Antarctic sea ice. We also show that meteorological conditions in the past, as well as when the snow and ice were sampled, play an important role in the scattering characteristics of the snow and ice as observed by the radars. Lastly we outline the implications of relating data from ground- and satellite-borne instruments, and how these insights might be applied. We discuss barriers to direct application, such as scales and geometries, and how the information gained through these studies and how these relate to analysis of data from current and future missions such as CryoSat-2 and CRISTAL.

Reference and repeat-observations of land-ice topography is critical to identify causal links between climate and ice trends, generate an accurate record of ice mass balance, and quantify land ice contribution to sea level change. Over the last 30 years, radar altimetry has been instrumental in monitoring ice sheets and their contribution to sea level change. Launched in 2010, the European Space Agency Altimetry mission CryoSat-2 was the first radar altimetry mission with a SAR/Interferometry radar altimeter payload. The aim of this new technology was to gain a better insight into the evolution of the cryosphere, in particular over the steep slopes typically found along ice sheet margins and glaciers, where the majority of the mass loss is taking place. CryoSat’s revolutionary design features a Synthetic Interferometric Radar Altimeter (SIRAL), with two antennas for interferometry, the corresponding SAR Interferometer (SARIn) mode of operation increases spatial resolution while also increasing the accuracy of the geo-location by resolving the angular origin of off-nadir echoes occurring over sloping terrain.

While the elevation of the Point Of Closest Approach (POCA), or level-2, is the standard product of the CryoSat-2 mission, the Interferometric mode of CryoSat-2 provides the ability to resolve substantially more than just the elevation at the POCA and thus led to a paradigm shift in radar altimetry observation over land ice. The so-called “swath processing” exploits the fact that over sloping terrain, CryoSat-2 altimeter operates in a manner such that the interferometric phase of the altimeter echoes may be unwrapped to produce a wide swath of elevation measurements across the satellite ground track, well beyond the POCA only. This technique provides the opportunity to increase spatial resolution and to recover elevation over regions where conventional radar altimetry fails; providing an opportunity to monitor land ice trends globally from radar altimetry.

Here I will discuss work done over the last decade to develop and apply swath processing over all components of land ice including the Greenland and Antarctic Ice Sheets, ice shelves, and glaciers worldwide. Swath-based studies have quantified global ice mass balance, sub-glacial lakes activity, ocean melting of ice shelves, contributing to a better understanding of processes linking ice trends to climate forcing. I will also highlight swath elevation as a CryoSat dataset over land ice globally as part of the EOLIS implementation of the CryoTEMPO programme. Finally, I will discuss opportunities to expand the application of swath processing further, in particular in the context of CRISTAL whose primary mode over land ice will be SARIn and with swath processing as a core technology to derive time-dependant ice topography.