The work performed along the fellowship consisted in analysing several datasets of different atmospheric variables in reanalysis (ERA5) and from model outputs (mainly from the Coupled Model Intercomparison Project version 6, CMIP6, and from the Polar Amplification Model Intercomparison Project, PAMIP). These were analysed using a large set of statistical methods in order to identify potential linkages between varying conditions over the Arctic in terms of sea ice concentration and variability in the atmospheric circulation at mid-latitudes. To this aim, the most relevant variables under strutiny were the upper-level zonal winds, sea-level pressure, and low- and mod-tropospheric temperature fields.
Among the PAMIP dataset, the experiments that were particularly relevant for this fellowship were called pdSST_pdSIC, which were forced by present-day conditions of sea surface temperatures (SST) and sea ice cover (SIC), and pdSST_futArcSIC, in which SST are also from present-day climate where SIC is reduced so as to be representative of a globe 2ºC warmer than in the preindustrial period (which will likely occur towards 2040 or 2050). By comparing both datasets, the signal resulting exclusively from sea ice loss on the mid-latitude climate can be isolated: this showed a clear impact from sea ice loss on the frequency of cold air reaching central and southern Europe in winter (under less sea ice, cold snaps tend to occur less frequency over Europe), while a similar impact but on the frequency of COLs reaching the Mediterranean Sea has also been detected. Both results point at a potentially significant alteration to both cold snaps and to heavy rain-producing systems over central and southern Europe. Naturally, these results are resulted to a single forcing (sea ice loss): different impacts can arise either from natural variability or as a forced response.
In parallel, some progress could be done on the identification of the impact from the horizontal resolution on the way models simulate the polar-midlatitude linkages: preliminary results from HighResMIP and PRIMAVERA model outputs suggest that a stronger linkage could exist when the horizontal resolution is increased. However, this pattern is not consistent among the models, and, more importantly, the state-of-the-art simulations from the EERIE project were not still ready at the time of the end of this fellowship, preventing their analysis. This is expected to be done in the upcoming months, even after the fellowship has come to an end.
