Insights on past and future sea-ice evolution from combining observations and models

We discuss the current understanding of past and future sea-ice evolution as inferred from combining model simulations and observations. In such combined analysis, the models allow us to enhance our understanding behind the observed evolution of sea ice, while the observations allow us to assess how realistically the models represent the processes that govern sea-ice evolution in the real world. Combined, observations and models thus provide robust insights into the functioning of sea ice in the Earth's climate system, and can inform policy decisions related to the future evolution of the ice cover. We find that models and observations agree well on the sensitivity of Arctic sea ice to global warming and on the main drivers for the observed retreat. In contrast, a robust reduction of the uncertainty range of future sea-ice evolution remains difficult, in particular since the observational record is often too short to robustly examine the impact of internal variability on model biases. Process-based model evaluation and model evaluation based on seasonal-prediction systems provide promising ways to overcome these limitations.