Early‐to‐Late Winter 20th Century North Atlantic Multidecadal Atmospheric Variability in Observations, CMIP5 and CMIP6.

The strong multidecadal variability in North Atlantic (NA) winter atmospheric circulation is poorly understood and appears too weak in climate models. Recent research has shown peak atmospheric multidecadal variability over the NA in late winter, particularly March, linked to Atlantic multidecadal variability (AMV) of the ocean. Here a range of NA atmospheric circulation indices are assessed to provide a comprehensive picture of early‐to‐late winter low‐frequency variability and its representation in the latest generation of climate models (Coupled Model Intercomparison Project Phase 6 [CMIP6]). As found for CMIP5, CMIP6 models exhibit too‐weak multidecadal NA atmospheric variability compared to reanalysis data over the period 1862–2005. Consistent with previous research, the eastern part of the NA westerly jet (U700NA) exhibits peak low‐frequency variability in March. However, for NA‐wide jet speed and the NAO, low‐frequency variability and model‐reanalysis discrepancies are strongest in January and February, associated with too‐weak NA ocean‐atmosphere linkages. Plain Language Summary: The strong low‐frequency multidecadal variability of large‐scale atmospheric circulation in the North Atlantic (NA) drives variability in climate conditions over western Europe. Reliably reproducing this variability in climate models is of high importance for estimating future climatic conditions. However, a current issue in climate modeling is that models generally exhibit weak winter multidecadal variability compared to estimates based on observationally constrained reconstructions (reanalyses). In particular recent research has shown that strong low‐frequency atmospheric variability and associated discrepancies between models and reanalyses are most pronounced in late winter, particularly March. The first step in this study is to establish that the current generation of climate models still exhibit the long‐standing issue that simulated multidecadal variability of NA atmospheric circulation is weaker than in reanalyses. A subseasonal analysis shows that, consistent with previous research, the eastern part of the NA westerly jet exhibits peak low‐frequency variability in March. However, for NA‐wide indices, tropospheric westerly jet speed and the NAO, low‐frequency variability is strongest in January and February. Evidence is shown suggesting that these inter‐index differences in subseasonal variability are associated with differences in monthly correlations with NA subpolar gyre SSTs. Key Points: Climate models exhibit systematically weak winter multidecadal North Atlantic (NA) atmospheric variability relative to reanalysis outputThe subseasonal timing of 20th century winter variability is sensitive to the choice of atmospheric circulation indexOnset of strong variability occurs earliest for NA jet speed (December), followed by NAO (January) and eastern‐NA jet speed (February) [ABSTRACT FROM AUTHOR]