Christopher Barrell

1 Introduction The Arctic marine environment is inherently a coupled system, with the atmosphere, ocean and sea ice all interacting and significantly affecting development over a range of timescales—from hours to decades. To predict these interactions, in particular the feedbacks between weather systems and the ocean/sea-ice surface, requires a coupled environmental prediction system.

Conditional Sampling by Sea Ice Extent It is evident that the sea ice in these climate simulations is a major source of variability and error. This prompts the question: are the models more accurate when there is less sea ice? If true, this would provide confidence in future projections when less sea ice is expected (e.g., IPCC, 2021). We address this question by conditionally sampling the historical simulations and analyzing model performance during “low” and “high” ice years.