Reflections on polar predictability challenges from three events in Korea - July 2025

Polar predictability matters now more than ever. With record sea ice loss, extreme temperature swings, and increasing human activity in these regions, accurate forecasts affect human and maritime safety, ecosystem management, and global climate understanding. Three recent events held in Korea during 16–25 July—WAMC 2025, the AntClimNow Predictability Workshop, and “Advances in Sea Ice Forecasting and Modelling” sessions at BACO-25—provided fresh perspectives on these challenges.PCAPS steering group members Clare Eayrs (KOPRI) and Vicki Heinrich (BoM) actively engaged in these recent events.

Attendees of the AntClimNow Predictability Workshop, Busan, 20 July 2025. Photo courtesy of Clare Eayrs.

At the BACO-25 conference, Clare chaired sessions on “Advances in Sea Ice Forecasting and Modelling” and presented the PCAPS ORCAS Task Team’s work on targeted observations to support AI-driven sea ice predictions, demonstrating PCAPS’s integrated approach to improving polar forecasts. Both Clare and Vicki contributed to discussions on current challenges in predicting Antarctic climate at the AntClimNow Predictability Workshop—where Clare serves as Co-Chief Officer. The workshop aimed to identify key uncertainties, foster collaboration, and chart priorities for advancing Antarctic climate prediction. Clare also presented a poster outlining PCAPS goals and structure at WAMC 2025, where discussions centred on rapid Antarctic climate change, atmospheric circulation patterns, and advancing climate modelling capabilities alongside observational methods. The poster can be downloaded here.

Predictability challenges

Complexity across scales

Polar regions face inherent predictability constraints that stem from observational challenges related to the vast expanse and the extreme, dynamic nature of these environments. 

The BACO-25 sea ice forecasting sessions directly mapped onto PCAPS PREDICT priorities around fidelity. Session 1 tackled modelling and forecasting across multiple time and space scales. Session 2 examined forecast performance and extreme events. Session 3 explored AI-based prediction systems, connecting to PCAPS’s focus on emerging technologies for enhanced prediction capabilities.

The AntClimNow Predictability Workshop revealed the complexity beneath these constraints. Discussions highlighted processes like climate regime shifts, stratospheric-tropospheric coupling and ocean memory effects as key sources of predictability and uncertainty. Participants advocated for probabilistic rather than deterministic approaches for extreme events, a pragmatic recognition that polar prediction must embrace uncertainty.

From science to operations

WAMC's operational focus grounded these scientific challenges in real-world consequences. Presentations demonstrated how predictability research must serve concrete needs, from station operations to aircraft landings and ship navigation. This operational perspective resonated with PCAPS's emphasis on fidelity, actionability and impact, ensuring scientific advances translate into practical benefits for diverse user communities.

Sea ice as the connecting thread

Sea ice emerged as the unifying thread connecting all these challenges. Sea ice is hard to predict because it sits at the intersection of atmosphere, ocean, and ice processes that interact across multiple scales. Critical observational gaps persist, especially for sea ice thickness and under-ice conditions, while models struggle with summer sea ice representation and smaller spatial scales.

Yet, sea ice prediction remains critically important precisely because of these connections. Sea ice influences climate through heat exchange and albedo effects, shapes ecosystems by providing habitat for key species, and determines human activities through navigation and logistics requirements. The BACO-25 sessions showcased emerging solutions that acknowledge this complexity: improved data assimilation methods, enhanced model physics, and innovative machine learning approaches that combine process-based understanding with data-driven techniques.

Integration as the path forward

These events demonstrated that advancing polar predictability requires more than better models or more data—it demands integrated approaches that connect scientific understanding with operational needs, embrace uncertainty while pushing for greater precision, and recognise that the greatest challenges often lie at the interfaces between different systems and scales.

User-centred design philosophy

Even within the research community, different disciplines have vastly different predictability needs across time and space scales. Vicki’s facilitation of discussions at the AntClimNow Predictability Workshop highlighted how PCAPS’s integrated environmental forecasting value cycle approach can map these cross-cutting needs, directly supporting PCAPS SERVICES objectives. 

WAMC's operational focus provided concrete examples of this challenge in action, from station operations requiring hourly weather updates to shipping routes needing seasonal ice forecasts. This diversity of needs reinforces why PCAPS SUSTAINABILITY emphasises informed decision-making to enhance human safety and reduce environmental risks across polar regions. The co-production mechanisms and stakeholder-driven service design that both events highlighted are essential for translating scientific advances into actionable guidance.

Future Directions

Clear priorities emerged from these discussions. 

• Targeted observations that serve multiple purposes in both polar regions: model improvement, real-time operational forecasting, and scientific understanding. The timing is ideal for PCAPS to facilitate community-driven observatory design through upcoming initiatives like IPY5 and Antarctica InSync, creating observational networks that integrate the priorities for polar prediction. 

• Enhanced integration across scientific disciplines to include diverse stakeholders and knowledge systems.  PCAPS promotes this approach through its PARTNERSHIPS objective and by supporting PCAPS INCLUSIVITY goals through capacity development and inclusive participation that ensures polar predictability research serves all communities.

From the perspective of the PCAPS ORCAS Task Team, these insights directly guide our work on observational requirements for AI prediction systems, ensuring that improved observations feed back into enhanced machine learning predictions through continuous observing system design optimisation.

The PCAPS ORCAS framework demonstrates the interconnected cycle between data collection and analysis, machine learning approaches, and operational outcomes.

The convergence of these three events feels symbolic of a broader convergence in the field: a recognition that the future of polar predictability lies in building integrated systems that connect science to service, uncertainty to decision-making and global research to local needs.