No abstract

For maritime navigation in the Arctic, sea ice charts are an essential tool, which still to this day is drawn manually by professional ice analysts. The total Sea Ice Concentration (SIC) is the primary descriptor of the charts and indicates the fraction of ice in an ocean surface area. Naturally, automating the SIC chart creation is desired. However, the optimal representation of the corresponding machine-learning task is ambivalent and discussed in the community. In this study, we explore the representation with either regressional or classification objectives, each with two different (weighted) loss functions: Mean Square Error and Binary Cross-Entropy, and Categorical Cross-Entropy and the Earth Mover’s Distance, respectively. While all models achieve good results they differ as the regression-based models obtain the highest numerical similarity to the reference charts, whereas the classification-optimised models generate results more visually pleasing and consistent. Rescaling the loss functions with inverse class weights improves the performance for intermediate classes at the expense of open water and fully-covered sea ice areas.

The elevation of ice sheets response dynamically to climate change and satellite altimetry is the preferred tool for evaluating the ice sheet-wide changes. In-situ validation are needed to ensure the quality of the observed elevation changes, but the coast is most often the limiting factor for the amount of in-situ data available. As more and more tourists are accessing the ice sheets, citizen science might provide the needed in-situ data in an environmental and cost-efficient way. Here, we investigate opportunistic kinematic-GPS profiles across the Greenland ice sheet, collected the American-Icelandic Expedition on the Greenlandic icecap 2018. First, the collected GPS-data are tested against widely used NASA Operation IceBridge airborne lidar-scannings, and shows good agreement, with an accuracy of 11 cm. The main difference is attributed to changes in the compaction of the snow as encountered while driving, as well as changing tire pressures. The kinematic-GPS data is then used for satellite validation by inter-comparing it with data from ESA's CryoSat-2 mission. Here, a bias in the two records of 89 cm is observed, with the Cryosat-2 observation originating from the subsurface of the ice sheet. This points to surface penetration of Ku-band radar on the Greenland ice sheet, and the observed magnitude is in accordance with the literature. Finally, we assess the long-term durability of citizen science kinematic-GPS data, when compared to a profile obtained in 2005 near Kangerlussuaq, West Greenland. Here, the records show an average ice elevation decreased of 9 meters and with peaks at 25.7 meters. This result show how kinematic-GPS data can be used to see the full impact of climate change by repeat measurements. Thereby are citizen science kinematic-GPS data shown to be a highly versatile approach to acquire high-resolution validation data for satellite altimetry, with the added benefit of potentially direct sampling properties of the surface and firn, when applying traditional airborne platforms. Thereby linking up with citizen-science expeditions is truly a beneficial way of providing cost-efficient satellite validations and may also have a societal impact by involving more in the climate monitoring of ice sheets.

No abstract