Dynamical and thermodynamical contributions to the mid-latitude atmospheric response to Arctic sea ice decline

Previous climate model studies have shown that Arctic sea ice decline can solely affect weather and climate at lower latitudes during the cold season. However, the mechanisms beneath this linkage are poorly understood. Whether sea ice loss have had an influence on the lower latitudes climate over the past decades is also uncertain (Barnes and Screen 2015). The goal of this work is to better understand the relative contributions of dyncamical and thermodynamical changes in the atmospheric response to Arctic sea ice loss, which have been suggested to oppose each other (Screen 2017). We conducted two sets of sensitivity transient experiments that allow to isolate the effect of Arctic sea ice decline on the mid-latitudes from other climate forcings, using the climate model CNRM-CM6 (Voldoire et al. 2019) in a coupled configuration or with an atmosphere-only. The first set of experiments, that is part of the European H2020 PRIMAVERA project, consists of a 100-member ensemble in which sea ice albedo is reduced to the ocean value (PERT) in the fully coupled CNRM-CM6, and which is compared to a 1950 control run (CTL) (Haarsma et al. 2016). This yields idealised ice-free conditions in summer and a more moderate sea ice reduction during the following months. The second set of experiments, that is part of the CMIP6 Polar Amplification Model Intercomparison Project (PAMIP, Smith et al. 2019), consists of a 300-member ensemble in which the atmospheric component of CNRM-CM6 is forced by sea ice anomalies associated with a future 2°C warming (FUT) and present day sea surface temperatures (SSTs). These are compared to experiments in which the atmosphere is forced by present-day sea ice conditions (PD) and the same SSTs. To extract the dynamical component of the response in the two sets of experiments, we use a dynamical adjustment method (Deser et al. 2016) based on a regional reconstruction of circulation analogs. We focus on three mid-latitudes regions in which a significant near-surface temperature response has been identified, namely North America, Europe and central Asia. We show that the cooling occurring over central Asia in both sets of experiments is dynamically-induced through an intensification of the Siberian High, and that opposed temperature responses over North America between the two sets of experiments could be explained by opposed dynamical components occurring in response to the imposed Arctic sea ice decline. Finally, we discuss whether different dynamical and thermodynamical contributions in the PAMIP multi-model experiments could explain the multi-model differences in the atmospheric response to sea ice loss.