Recalibration of extreme multi-decadal trends in the North Atlantic Oscillation.

The historical variability of the winter mean North Atlantic Oscillation (NAO) has featured periods with large multi-decadal trends which are not well represented by coupled general circulation models (CGCMs), consistent with a lack of autocorrelation in the winter mean NAO index series. Post-processing reddening methods are proposed, using stochastic model theory to make the autocorrelation structure of the CGCM NAO index match that of the observed NAO. Using CGCMs from the Coupled Model Intercomparison Project Phase 6 (CMIP6), these recalibration methods are shown to successfully improve the autocorrelation structure of the NAO and in turn the simulation of extreme trends. The 1963-1993 NAO trend is the maximum 31-year trend in the historical period, but without reddening the CGCMs underestimate the likelihood of this trend by a factor of ten. CMIP6 future projections show a small systematic increase in long-term (2024-2094) NAO ensemble mean trends relative to the magnitude of the radiative forcing from 0.09 to 0.16 hPa/decade (range for low to high radiative forcing scenarios). This range is doubled after reddening, becoming 0.24 to 0.35 hPa/decade. There is also a related shift in the distribution of extreme 31-year NAO trends, which is more clearly apparent after reddening. Near-term projections of the next 31 years (2024-2054) are less sensitive to the future scenario. After reddening they still show weak-to-no forced trend in the models but have a 74% larger ensemble range (around +/- 1 standard deviation per decade). This level of internal variability could increase or decrease regional climate change signals in the Northern Hemisphere by magnitudes that are greatly underestimated when using raw climate model output.