Global climatology of cyclone clustering in present and future climates

Cyclone clustering, the rapid succession of multiple extratropical cyclones at a confined region during a short period of time, causes a large fraction of European weather extremes like wet spells and wind extremes. The idea that several cyclones follow a similar track dates back to the concept of cyclone families of Bjerknes and Solberg. To investigate the dynamical causes of cyclone clustering, it is necessary to diagnose the occurrence of cyclone clustering and to determine their characteristics. However, so far, most diagnostics focused either on local impact or on a statistical analysis of storm recurrence. While the first cannot be applied globally, the latter is difficult to relate to individual events. We therefore use a novel method to globally detect cyclone clustering that is closer to the original concept of Bjerknes and Solberg, where extratropical cyclones follow similar tracks within a given time period.Using this new cyclone clustering diagnostic based on spatio-temporal distance between cyclone tracks, we analyse cyclone clustering globally in Era-Interim for the period 1979 until 2016 as well as for 10 CMIP6 climate models. We separate the cyclone clusters into two types: one representing the classical bjerknes-type clusters, and one representing more stagnant clusters. We find that cyclone clustering mainly occurs along the climatological storm tracks, with the Bjerknes-type more common at the western side of the storm tracks, while the stagnant-type of cyclone clusters occurs more downstream. In general, clustered cyclones are stronger than non-clustered cyclones, especially for the Bjerknes-type. CMIP6 models feature a slight bias towards an equatorward shift of the storm tracks, and this bias is stronger for clustered cyclones. This relatively larger bias suggests that the clustered cyclones are important for correctly simulating the storm tracks. In a future climate cyclone clustering occurs more poleward. Furthermore, the average number of storms per cluster decreases in a future climate, though the mean intensity of the cyclones that are clustered increases slightly.