Abstract Wintertime North American cold spells often co‐occur with European wind extremes, forming so‐called pan‐Atlantic extremes. In Part 1 of this paper, we identified regionally varying relationships between the two sets of extremes in ERA5 reanalysis data. We further analyzed how CMIP6 climate models simulated the past and future surface and tropospheric circulation patterns of the pan‐Atlantic extremes. In this second part of the study, we interpret the pan‐Atlantic extremes and their representation in models through the lens of dynamical circulation pathways. Using a novel clustering method that captures the evolving North Atlantic atmospheric circulation during North American cold events, we identify five pathways in reanalysis data. Three of the pathways correspond to distinct spatial and temporal patterns of co‐occurring North American cold events and European extreme wind days (EWDs). Of these three pathways, two resemble the positive and negative phases of the Arctic Oscillation (AO+ and AO−), while the third resembles a propagating wave train (WT). The four CMIP6 models generally reproduce the large‐scale features and frequencies of the pathways, the major exception being the WT pathway. Model projections under SSP5‐8.5 show a general weakening of the circulation anomalies associated with all pathways, but little evidence of changes to their frequency. Our qualitative conclusions are robust to changing the approach for defining the pathways and to projecting models onto reanalysis data to define the pathways.
Leeding et al. (Fri,) studied this question.