Author:
Kucik Andrzej,Stokholm Andreas
Abstract
AbstractFor 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.
Publisher
Springer Science and Business Media LLC
Reference32 articles.
1. Perovich, D. et al. Arctic Report Card 2020: Sea Ice https://repository.library.noaa.gov/view/noaa/27904 (2021).
2. Constable, A. J. et al. Cross-Chapter Paper 6: Polar Regions 2319–2368 (Cambridge University Press, 2022).
3. Bekkers, E., Francois, J. F. & RojasRomagosa, H. Melting ice caps and the economic impact of opening the northern sea route. Econ. J. 128(610), 1095–1127 (2017).
4. Boutin, G., Williams, T., Rampal, P., Olason, E. & Lique, C. Impact of wave-induced sea ice fragmentation on sea ice dynamics in the MIZ. Technical report, Copernicus GmbH (2020) (Accessed 01 Mar 2023).
5. Saldo, R. et al. AI4Arctic/ASIP Sea Ice Dataset-version 2. https://data.dtu.dk/articles/dataset/AI4Arctic_ASIP_Sea_Ice_Dataset_-_version_2/13011134. (2020).
Cited by
5 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献