Serial Clustering of North Atlantic Cyclones and Wind Storms: A New Identification Base and Sensitivity to Intensity and Intra-Seasonal Variability

Abstract:

Time-clustered mid-latitude winter storms are responsible for a large portion of the overall windstorm-related damage in Europe. Thus, its study entails a high meteorological interest, while its outcome can result in a crucial utility for the (re)insurance industry.

In addition to existing cyclone-based studies, here we use an event identification approach based on surface near wind speeds only, to investigate windstorm clustering and compare it to cyclone clustering. Specifically, cyclone and windstorm tracks are identified for winter 1979-2013 (Oct-Mar), to perform two sensitivity analyses on event-clustering in the North Atlantic using ERA-Interim Reanalysis. First, the link between clustering and cyclone intensity is analysed and compared to windstorms. Secondly, the sensitivity of clustering on intra-seasonal time scales is investigated, for both cyclones and windstorms.

The wind-based approach reveals additional regions of clustering over Western Europe, which could be related to extreme damages, showing the added value of investigating wind field derived tracks in addition to that of cyclone tracks.

Previous studies indicate a higher degree of clustering for stronger cyclones. However, our results show that this assumption is not always met. Although a positive relationship is confirmed for the clustering centre located over Iceland, clustering off the coast of the Iberian Peninsula behaves opposite. Even though this region shows the highest clustering, most of its signal is due to cyclones with intensities below the 70^th percentile of the Laplacian of MSLP.

Results on the sensitivity of clustering to the time of the winter season (Oct-Mar) show a temporal evolution of the clustering patterns, for both windstorms and cyclones. Compared to all cyclones, clustering of windstorms and strongest cyclones culminate around February, while all cyclone clustering peak in December to January.