With longer time periods, larger scales in space are also involved. This means that if we look at events lasting about 3 weeks, then the exceptional regime in the atmosphere is not at the local or meso-scale, but at the planetary scale. Mailier et al. (2006) revealed that the large-scale atmospheric circulation pattern controls the speed and the path of existing cyclones. As the Baltic Sea region lies at the end of the North Atlantic see more storm track, serial clustering of cyclones in this area is common, but it is also important that the serial clustering of mid-latitude cyclones

is particularly associated with strong systems (Mailier et al., 2006 and Vitolo et al., 2009). Therefore, we find that the actual cause of the sea level extremes in 1967 and 2005 could be the properties of a series of cyclones crossing the Baltic Sea, rather than the parameters of a single cyclone causing a particular storm surge flooding coastal areas. The clustering of cyclone tracks in time and space does not have a very high probability, but produces extreme cases that do not belong to the ensemble of high storm surges. In other words, certain (to some extent, similar) trajectories of cyclones with certain periodicities in a given timespan give rise to extreme sea levels that are real outliers in the ensemble of extreme cases. This conclusion is supported by the

series of higher-than-normal sea levels oscillating before and after the main extreme event, but also by the fact that there was always more than one deep cyclone during the approximately two-month Selleck CP 868596 period that surrounded the highest sea level events. The exact characteristics and sequence of the cyclones need further research, as the more than just chance clustering of cyclones does not provide sufficient evidence for the causality of the forcing. But at the local scale, the propagation of these cyclones merely generates a wind system that changes in speed and direction, and the estimation of these winds and

their evolution, preconditioning and conditioning of sea level extremes also require refining and downscaling of the wind pattern (see Figures 4 and 5). Ensemble hydrodynamic modelling of the sea (using ROMS, HIROMB, HBM, NEMO, etc.) could provide important information about the response of the sea system Sclareol and would help to define the framework for atmospheric forcing and uncertainty of sea level extremes, as well as the necessary preconditions for sea level extremes. Analysis of two extreme storm surges and the relevant forcing of cyclonic activity permits the definition of the basic parameters of cyclones and their series causing extreme sea levels along northern Baltic coasts. The authors wish to thank Olga Zolina, Irina Rudeva and Sergei Gulev for making the Northern Hemisphere cyclone database available, Marko Zirk for preparing the Baltan65 + pressure maps, and the two anonymous reviewers for their helpful comments.