Models and observations agree on strong reductions of midlatitude cold extremes

An apparent increase in winter cold extremes over the Northern Hemisphere midlatitudes has been reported over recent decades, in contrast to the robust decreases projected by climate models. Some have attributed this model-observation discrepancy to Arctic-induced changes in atmospheric circulation that are poorly captured by models, but this remains controversial. In this study, we search for evidence of a model-observation discrepancy by directly comparing recent trends in midlatitude cold extremes between observations and large ensemble simulations with historical forcing. We find that the frequency and intensity of midlatitude cold extremes have strongly decreased since 1990 and the magnitudes are consistent with model trends. While some regions show weak, non-significant increases in cold extremes in recent observations, the spatial distribution of trends and the percentage of land with an increase in cold extremes are consistent with internal variability seen in individual model realizations. Our results provide strong evidence against claims of a recent increase in midlatitude cold extremes or of a model-observation discrepancy.