Separation of the Baltic Sea water level into daily and multi-weekly components

• The water level time series in the Baltic Sea is split into daily and weekly parts.
• Local storm surges have the exponential distribution and contain all outliers.
• Background variations in the water level have a Gaussian distribution.
• The range of local flooding in different coastal sections is quantified.

Storm surges and changes in the water volume of the entire sea, with typical time scales about a day and a few weeks, respectively, are the largest contributors to the water level variations at the eastern Baltic Sea coasts. Our analysis employs time series of sea levels numerically reconstructed using the RCO (Rossby Center, Swedish Meteorological and Hydrological Institute) ocean model for 1961–2005. The distribution for the weekly-scale water level, defined as a running average over a certain time interval, has an almost Gaussian shape. For the 8-day average the distribution of the residual, interpreted as the frequency of occurrence of local storm surges of different height, almost exactly matches the exponential distribution that can be considered as reflecting the time between events of the underlying Poisson process. The distribution of the total water level contains a few outliers that often do not match the classical statistics. All extreme values (outliers) of water level are a part of the exponential distribution of storm surges for averaging intervals longer than about 3 days. Such separation of phenomena on different temporal scales is universal for the entire eastern Baltic Sea coast. The slopes of the exponential distribution for low and high water levels are different, vary markedly along the study area and provide a useful quantification of different coastal sections with respect to the probability of coastal flooding.