Shifts of means are not a proxy for changes in extreme winter temperatures in climate projections

Changes in the severity of extreme weather events under the influence of the enhanced greenhouse effect could have disproportionally large effects compared to changes in the mean climate. Here, we explored the meteorological circumstances of extremes and changes therein using two 49-member climate model ensembles for reference (1961-1990) and scenario (2051-2080) greenhouse-gas concentrations. We have focused on daily-mean surface-air temperatures over the Northern Hemisphere in January. Over large parts of the continents, changes in the one-in-10-year temperature events are influenced at least as much by changes in the shape of the probability distribution functions (PDFs) as by shifts in the mean. In coastal areas, this is largely attributable to changes in the large-scale circulation, for those types of extremes linked to infrequent wind directions. In other areas, the inhomogeneous mean warming, increasing inland and polewards, affects the tails of the local temperature PDFs. Temperature extremes in widely different regions were found to be linked by a large-scale circulation anomaly pattern, which resembles the Arctic Oscillation. In the scenario ensemble, this anomaly pattern favors its positive phase, leading to enhanced probabilities of westerly winds in a belt around the Northern Hemisphere.