In fact, the tracks of the dangerous cyclones in the research area are not always very straight and they also come from a rather large sector – from the SW to NW (Figure 2). The directions offered by AV2010 have a larger meridional

track component than the average of all the real cases. At both sites, Pärnu and Tallinn, the highest historical sea levels were registered on 8–9 January 2005, at Pärnu since 1923 and at Tallinn since as far back as 1842. From the viewpoint of atmospheric pressure minima, this was only the twelfth cyclone (with minimum pressure of 957.4 hPa) in the 1948–2010 period in this region. On the other hand, Erwin/Gudrun could be called an explosive Verteporfin cyclone or bomb, according to Bergeron’s definition (Roebber 1984), with a maximum Normalised Deepening Rate (NDP) of − 24.5 hPa/24 h during its first day of existence. The second highest storm surge in the area, from 18 October 1967, was caused AZD6244 in vitro by a much longer cyclone, with a minimum pressure of 968.3 hPa. For the October 1967 cyclone, NDP was − 20.9 hPa/24 h – also a very high value. We presumed that the extreme sea levels during the 8–9 January 2005 event were

actually caused not so much by certain parameters of a single cyclone as by the properties of a sequence of cyclones crossing the Baltic Sea that had certain (to some extent similar) trajectories with a certain periodicity over a given time span. In Figure 1 one can follow how an extreme sea level was built up by six progressive secondary sea level maxima, easily detectable over approximately 10 consecutive days before the occurrence of the most extreme sea level, and with a very similar periodicity in both the 1967 and 2005 cases. Looking at the trajectories of the cyclones (Figure 3) and comparing the sequences of cyclones for these two extreme storm surges, in both cases we can point out 5 cyclones that had the lowest air pressure learn more values in the sector 10°E–30°E, 55°N–67°N. There were four cyclones crossing the area that arose one after another and had very similar directions of propagation. Nevertheless, common

to both events were the two very long (in time and space) cyclones generated over the western Atlantic Ocean at latitude ca 40°N. The second of these was generated after the time of the sea level maxima, which indicates the possible serial clustering of cyclones, induced by the time-varying effect of large-scale atmospheric factors on individual cyclone tracks. Figures 4 and 5 show maps of mean sea level pressure for six dates when a strong SW wind reached Pärnu Bay, and a sea level maximum could be detected at Pärnu in either October 1967 or December 2004/January 2005 (see Figure 1). Some of the synoptic patterns recall the ideally circular cyclone shown in AV2010, especially in the case of the main, strong 2005 cyclone (9.01.