Mediterranean Drying by a Positive North Atlantic Oscillation Trend over the Last 65 Years Is an Extreme Outlier in the CMIP6 Multimodel Ensemble

Abstract The Mediterranean region has experienced declining precipitation during the cool season since the 1950s, which is broadly consistent with climate model future projections. The observed drying is focused on midwinter and the central to western Mediterranean. Observed winter drying in the western Mediterranean is at the very edge of the CMIP6 ensemble range of historical simulations for the same time period and, in the central to eastern Mediterranean, is in the lowest percentiles of model ensemble distributions. Moisture budget analysis shows the drying to be dynamic, driven by changes in circulation, rather than thermodynamically related to changes in humidity. An observed positive trend in the North Atlantic Oscillation drives the drying by placing subsiding air over the Mediterranean augmented by dry northerly advection. The more modest radiatively forced drying in the CMIP6 multimodel ensemble is also dynamic but related to a high pressure anomaly over the eastern North Atlantic with a different pattern to the North Atlantic Oscillation. A small number of individual runs of some CMIP6 models do show trends in Mediterranean drying, the North Atlantic Oscillation, and moisture budget similar to that observed. However, the observed Mediterranean drying and North Atlantic Oscillation trends are at the very limit of what climate model simulations can do from the combination of variability and forced response over this time period. The observed drying is either (i) an extremely rare manifestation of natural multidecadal variability akin to that in models, (ii) a result of variability not captured in models, or (iii) a forced response that models are missing. Significance Statement Drying of the Mediterranean region over the past decades has had significant impacts on water resources, agriculture, and fire. The drying has impacted the whole region but is most extreme in winter and the west. Understanding its causes will help improve predictions of whether this drying trend will continue, stop, or reverse. The drying is driven by a long-term trend to a more positive North Atlantic Oscillation that places northerly, subsiding drying airflow over the Mediterranean. Climate models simulate only weak Mediterranean drying as a response to radiative forcing, driven by different changes in atmospheric circulation. Individual runs of some models can come close to matching observations, but this is rare within the multimodel ensemble. We conclude the observed drying is either a case of internal variability that models struggle to capture or a forced response they are missing.