Investigating sea ice and snow features using multi-frequency remote sensing observations

The sea ice and its snow cover are integral components of the Earth’s climate system, modifying physical, climatic, and biogeochemical processes. The sea ice constantly deforms due to external forcing by winds, currents, and waves, resulting in a variety of localised, sea ice features. Such features can be driven by divergence-induced deformation resulting in sea ice leads, i.e. cracks between floes with open water; or by convergence-induced deformation where ice floes are pushed against each other, breaks, and where the broken floes can accumulate on top and below the ice, forming ice ridges. The overlying snow cover is just as heterogeneous as the ice cover with significant spatiotemporal complexity owing in part to the heterogeneity of the ice cover, but largely also to the multiplex interactions between the snow, sea ice, ocean, and atmosphere. When considering remote sensing observations acquiring measures of the surface through complex scattering and absorption mechanisms with the atmosphere and surface, the features either at the local scales (such as ridges or leads) or at the scale of the electromagnetic pulses (i.e. the wavelength, such as metamorphism or layers within a snow pack) can significantly alter the returned signals. Furthermore, the spatial scales at which the remote sensing systems observe (e.g., ground, air, or space) lead to significant differences in the observed signals.

This thesis takes on the ambitious goal of evaluating the extent to which such sea ice and snow features impact and limit multi-frequency altimetry observations at different spatial scales, predominantly air- and spaceborne altimetry. This is achieved following three different paths: by (a) using the high-resolution dense surface sampling laser altimeter mission to detect localised deformation features, and evaluate the extent to which such features are detectable from space; (b) using a recent alignment of satellite altimeters observing the sea ice with different wavelengths and systems to evaluate near-coincident dual-frequency observations along orbits, and using a collection of multi-frequency altimeters at monthly scales to evaluate the dominating surface signals; and, (c) evaluating one of these near-coincident dual-frequency orbits from the view-point of airborne altimetry, where the full suite of instrument relevant for understanding impacts of penetration and ice conditions observed from different scales can be compared. A relevant aspect when considering translating between air- and spaceborne observations is the representativeness of such observations and the relevant procedures needed to align such observations, which a part of this thesis is dedicated to.

In the first path, ridges detected from the spaceborne polar photon-counting laser altimetry mission, ICESat-2, are evaluated for the Barents Sea and Fram Strait. Here, a comparison with statistics derived from upward looking sonars is conducted, and large discrepancies are observed when unaccounted for the difference in reference level (i.e. to the local sea ice elevation or the water level), and when using different minimum thresholds for the top of the ridge sail. More than 40% of the ridges detected from space using ICESat-2 were not seen from upward looking sonars using the thresholds currently applied, limiting the statistics of the upward looking sonars to the thicker, and less abundant ridges of the Arctic. Furthermore, a comparison with an airborne under-flight showed ICESat-2 able to capture the magnitude of the ridges, but not the extent or frequency of ridges to the same resolution as airborne observations. In addition, it was noted that different aggregation methods used to derive elevation observations can result in missing or complete loss of identified localised features such as ridges or leads.

In the second path, a first estimate of orbit-wise snow depth estimates is presented for the recent alignment of ICESat-2, and the polar synthetic aperture radar (SAR) altimetry mission, CryoSat-2, known as CRYO2ICE. The difference in wavelength can, based on assumptions of zero and full penetration, provide an estimate of snow depth over sea ice. This is computed for the Arctic sea ice cover during 2020–2022 and presents a first step at producing snow depth along orbits from satellites. Here, relevant aspects such as incomplete radar penetration due to synoptic weather events are discussed, as well as the difficulties with co-locating near-coincident observation essentially observing the sea ice in different ways.

In the third path, the Antarctic under-flight of a CRYO2ICE orbit is evaluated, where a full suite of instruments relevant for characterising microwave penetration into the snow was mounted. Significant scattering was observed from the air-snow interface (the top of the snow cover) at all frequencies used, which is counter to the general assumptions. In particular, it was noted that the contributions are significant and distinguishable, which contradicts the current assumption of a singular dominating scattering surface for airborne observations. These results further highlight the dissimilarities between air- and spaceborne observations and call for a re-evaluation of the available airborne altimetry observations.

In conclusion, this thesis makes important steps toward understanding the limitations of multi-frequency altimetry acquired over sea ice from different spatial scales. The first estimates of along-orbit snow depth using dual-frequency altimetry from the dual-mission campaign, CRYO2ICE, were presented. Such observations were further evaluated using airborne estimates, where the main assumption used in studies was concluded as unjustified,urging the community to further evaluate the airborne studies for the viewpoint that several interfaces and layers can significantly contribute to the received signals and potentially be separated. This assumption is not applicable to satellite systems and presents a significant dissimilarity between how the systems observe the surface. Finally, the thesis showed the limitations of the space-borne altimeters to detect localised sea and snow features and further highlighted the current obstacles present to align statistics of such parameters derived using sensors observing from above- and below the ice. The results of this thesis can be applied to further evaluate the altimeter observations available and help understand the discrepancies we observe between satellite missions, or across spatial scales such as observed by ground-based, air- or spaceborne systems.