Recent work from our group shows that there is a linear correspondence between the 1000-500 hPa column warming in the Arctic in winter and the increase in the Siberian High Index (SHI) in a variety of climate model experiments, such as forced 1) with sea-ice loss (both in atmosphere-only and coupled models) and 2) by nudging the temperature variability in the atmospheric column of the high Arctic to projected temperature variability from future states of the ensemble mean of CESM1-LENS (without any other forcing). We also included an ensemble of AMIP experiments (forced by chronology of SST and sea-ice) as well as AMIP experiments with the addition of nudging the high-Arctic temperature variability in the atmospheric column to that from reanalysis. Only the pure AMIP ensemble is an outlier showing insignificant response in the SHI whereas when the nudging is included, the linear relation is strong. We have now included other PAMIP model experiments in Fig. 4 in Labe et al (2020) and they confirm the earlier result that sea-ice forcing by itself results in warming that is confined to the surface with little change in the Arctic mid-troposphere (~700 hPA) and thus little response in the SHI resulting in insignificant surface winter cooling in Eurasia. Similarly, SST warming outside the Arctic results in weak warming in the mid-troposphere Arctic and thus little change in the SH and weak Arctic-midlatitude teleconnection, albeit in an atmosphere-only model (AMIP).

Here we further explore the available large ensembles including PAMIP, the new collection of initial-condition large ensembles, and CMIP6 to examine connections between different variables, focusing on the connection between mid-tropospheric Arctic warming (T_700), the SHI, strength of the stratospheric polar vortex and the warming in the tropical upper troposphere. This will be supplemented with an analysis of the interannual variability in the seasonal evolution of these variables from late fall through winter. We expect that Arctic to midlatitude teleconnections are intermittent and the aim is to identify “windows of opportunity” when this connection is strong, followed by composite analysis for hypothesis testing. The long-term aim of this line of research is to supplement the data analysis with perturbation experiments to verify mechanisms.