Reflections from Arctic Partnership Week 2024: The growing role of AI in sea ice prediction in a rapidly changing Arctic
As the Arctic undergoes rapid transformation with unprecedented sea ice retreat, it has become both a stark symbol of climate change and an urgent reminder of our need for accurate prediction systems. During the 2024 Arctic Partnership Week (10-12 December, Busan), discussions were centred on sustainable solutions for the changing Arctic. PCAPS Steering Group member Clare Eayrs from the Korea Polar Research Institute participated in two key sessions during “Knowledge Day”: the “Arctic Youth Dialogue” and a KMI-Arctic Frontiers session on “Sustainable Arctic Seas: AI and Environmental Protection.”
Clare Eayrs presenting at the 2024 Arctic Partnership Week. Photo courtesy of the Korea Maritime Institute.
The Arctic Youth Dialogue showcased the next generation of polar researchers. Presentations covered a remarkable range of topics, from ice decline and ecological consequences to Indigenous rights and software innovation. Drawing from her extensive involvement with APECS and early career leadership roles in both the Polar Prediction Project (the precursor to PCAPS) and the SCAR Scientific Research Programme AntClimNow, Clare presented on "Strengthening Polar Science Through Early Career Networks and Mentorship," sharing how these networks foster scientific collaboration and professional development. It was great to have APECS Vice-President Conor Savage among the workshop participants to underscore the importance of early career networks in polar science.
PCAPS is committed to developing the next generation of polar researchers through dedicated early career researcher (ECR) events and opportunities to participate in interdisciplinary task teams. In a recent blog post, our PCAPS Communication Fellows Amanda Skogjordet and Linn Bolin Haakenstad outlined their experiences in PCAPS as ECRs. More opportunities for ECRs via PCAPS are planned for 2025 and beyond. Polar ECRs are encouraged to register for the PCAPS mailing list to stay updated on the latest opportunities.
Building on these themes of innovation and fresh perspectives, the KMI-Arctic Frontiers session on "Sustainable Arctic Seas: AI and Environmental Protection" revealed how AI is revolutionising environmental prediction, with ECRs playing key roles in developing new technological approaches. In a panel session chaired by Anu Fredrikson (Executive Director of Arctic Frontiers), Clare Eayrs, Tore Hattermann (Section Leader for Oceanography at the Norwegian Polar Institute), and Rolf Rødven (Executive Secretary of AMAP) explored how machine learning approaches can address rapid changes in the Arctic. Clare’s presentation, developed from the outcomes of an International Sea Ice Modelling Workshop held in Korea in 2023, demonstrated how machine learning is transforming every aspect of the sea ice modelling workflow, from parameter tuning and bias correction to the development of complete AI model systems.
Participants in the KMI-Arctic Frontiers Seminar (Photo credit: Jenny Turton, Arctic Frontiers),
The discussions during the week highlighted both opportunities and challenges for advancing AI applications in sea ice prediction. At the forefront is the dependence on high-quality training data, raising crucial questions about both the quantity and quality of available datasets. Equally important is the need to build confidence in AI systems, which requires maintaining physical understanding through interpretable algorithms. This can best be achieved through cross-sector collaboration between sea ice scientists, AI experts, and stakeholders.
Panelists in the KMI-Arctic Frontiers discussion (Photo credit: Jenny Turton, Arctic Frontiers),
PCAPS is developing a new task team to address some of the emerging challenges in implementing AI systems for polar prediction. The proposed initiative will focus on optimising observational data support for AI systems, establishing requirements for model initialisation and evaluation, assessing the physical consistency of AI predictions, and providing recommendations for strengthening observing systems for improved forecast initialisation.
Building on the discussions at Arctic Partnership Week, where ‘cooperation’ emerged as a central theme across all sessions, this task team will emphasise cross-sector collaboration and early career involvement, bringing together traditional scientific understanding with innovative AI approaches. This spirit of cooperation – between nations, across sectors, and across career stages – will be essential for addressing the complex challenges of polar prediction. More information on the new task team will be available soon - check the PCAPS website regularly for updates.