Consistent atmospheric circulation responses due to Arctic sea ice loss between prescribed sea ice simulations and single long control simulations

<p align="justify">This study investigates the differences in atmospheric responses to Arctic sea ice anomalies between simulations from six (atmospheric-only) models contributing to the Polar Amplification Model Intercomparison Project and one long control simulation (piControl from CMIP6) from the same six (coupled) models. We perform a composite analysis between years of low and high Arctic sea ice extent in the piControl and consider four different types of experiment in the PAMIP where only the sea ice concentration is changed (pdSST-futArcSIC, pdSST-futBKSeasSIC, pdSST-futOkhotskSIC and pdSST-pdArcSIC) to examine the associated atmospheric circulation changes owing to an Arctic sea ice loss. A negative change in the North Atlantic Oscillation (NAO) pattern emerges in winter, linked to the so-called stratospheric pathway, and is mainly due to the sea ice anomaly in the Barents-Kara Seas. The results in the PAMIP experiments support these findings, except that the intensity is lower than in the piControl composite analysis. This work highlights that the atmospheric circulation responses to Arctic sea ice loss in a long control simulation (CMIP6) show similarities with the responses of a coordinated set of numerical model experiments with prescribed sea ice (PAMIP). However, the atmospheric responses in the numerical models with prescribed sea ice display weaker changes than in CMIP6. The role of the atmosphere-ocean coupling and of the initial sea ice condition could be the main reasons for this difference in intensity.</p>