The Baltic Sea water level varies over a wide range of temporal and spatial scales. Its spatial variability is mainly influenced by prevailing winds and river run-off, producing a mean sea-surface slope from the east to the Danish Straits in the west [Alenius et al., 1997, 2000]. Studies of secular variations of relative sea level reveal that the dominant factor is the post-glacial rebound (land uplift), contributing to an average decrease of 4.5 mm/year. Analysis of two 100-year-long sequential time-series from the Stockholm station showed that this apparent rate of uplift has slowed down during the last century by about 1 mm/year. This change was attributed to the effect of the eustatic sea level rise, from glacial increase in the amount of the seawater (from melting ice) and thermal expansion in the sea. The value of 1 mm/year is also close to the estimates of a global mean sea level rise during the last century. The long-term sea level trend in the Baltic Sea varies with geographical location because of the non-uniform rebound in Fennoscandia. This sea level change was estimated to be between +1.7 mm/year in the south-eastern Baltic and –9.4 mm/year in the north-western Gulf of Bothnia.

Interannual sea level variations are quite obvious from the long time-series at the Stockholm station. Spectral analysis of tide gauge records from Finnish and Swedish stations shows this variability to be aperiodic. Recent findings indicate these variations to be mostly due to sea level changes in the Kattegat strait, which connects the Baltic Sea to the ocean, as well as due to changes in freshwater supply to the region. The remaining 20 to 50% of the total interannual variability is due to the varying winds, air pressure and changes in water density. The spectra of the sea levels, recorded at various locations in the Baltic Sea, show a dominant peak at the annual period. The amplitude of this annual oscillation varies between 9.5 and 12 cm along the Finnish coast of the Baltic Sea. However, there is no clear physical explanation about the cause of this periodic process. According to simple model calculations, 50–80% of the annual sea level variations are introduced into the Baltic from the North Sea. There are no significant peaks in the short-term (up to several months) sea level spectra, so the spectrum of the daily mean values is mostly white, while tidal sea level variations are relatively small in the Baltic Sea [Lyakhin et al., 1997; Alenius et al., 2000]. Most of the high frequency changes are therefore governed by synoptic winds, air pressure changes and water current variability. The Baltic Sea eigen-oscillations give rise to the commonly observed high-frequency peaks in the sea level spectra, at about 0.5–1 cycles/day.

One of the major issues of current environmental research is the problem of global warming. This issue occupies scientists and governments, not only because of the warmer climate, but also because of measurable increase in sea level and its variability due to melting ice, thermal expansion of sea water, and from changes in prevailing winds. At most locations, these processes cause only a gradual inundation of sea shores. However, over a period of years and decades, such systematic trends may lead to increased susceptibility to flooding and can have serious consequences. It is therefore important to obtain estimates of the relevant time scales for sea level rise to critical values in the Gulf of Finland. This information can be used to determine the coastal locations that should be protected by dams. Such calculations will be carried out in this project using multi-year analyses of coastal and altimetry data in the Gulf of Finland. Another, and probably more important factor of long-term trends in sea level in the Gulf of Finland is vertical movement of its shores. Available levelling data along the coast of the Gulf of Finland will be used to estime such vertical motion.

St-Petersburg's Palace Square submerged during the flood of Nov. 7, 1724

In the Baltic Sea the most intensive sea level oscillations occur in the Eastern Gulf of Finland (EGF). Storm winds over the Baltic Sea are capable to drive large volumes of water into the shallow head of the Gulf of Finland. Development of a reliable system for prediction of sea level oscillations is a current task for many locations along the EGF. The most important among them is St. Petersburg, whose low-lying regions, which include its central historical areas, are subject to annual flooding. Floods cause a significant economic and social damage to this city, sometimes leading to human fatalities. Forecasts of extreme water levels are also very important for the Leningrad Nuclear Power Station. Officially, flooding starts in St. Petersburg when its water level exceeds 160 cm above the Kronshtadt zero level. When the water level exceeds 3 m, the flood is classified as a catastrophic one. There were about 300 floods in the City's history and 3 of them were catastrophic. The maximum water level rise (421 cm) occurred on 19 November 1824. After flood of 15 October 1955 (293 cm), the project of the St. Petersburg Flood Protection Barrier in Neva Bay was initiated. The construction was started in 1979, but in 1990, when 60% of the work was complete, the project was stopped, as it was concluded that the Barrier may be the main source of poor water quality in the Bay. In 1994, after positive recommendations by the International Commission of Experts, the Russian Government decided to complete the Barrier construction in 2001, but, due to lack of finances, this work is still continuing.

Flood of 8-9 January, 2005, in St-Petersburg

Forecasting the water levels in the Eastern Gulf of Finland (EGF) is the official task of the North-West Regional Administration of Hydrometeorological Service of Russia (NWHMS). Until 1967, these forecasts were solely based on empirical methods, using the observed water level in Tallinn. Since 1967 and untill 1997, when there was a threat of flooding, a one-dimensional model of the Baltic Sea was used, with advance time of 12 hours. This model was developed at the Leningrad Branch of the State Oceanographic Institute, with its grid consisting of only 200 points. In the early days, atmospheric pressure forecasts were made using the methods from the Moscow Hydrometeorological Center. These were later obtained from the European Center in Bracknell, with a horizontal grid resolution of 2.5° and a time step of 12 hours. The main source of errors was this large time interval for meteorological data that are used to force the sea level models.

As a result of completion of several Russian-Netherlands projects in 1998-2004, a new and fully automated flood forecasting system, with an advance time of 48 hours, was developed. This system is based on the two-dimensional model of the Baltic Sea BSM5 and a high-resolution regional model of the atmosphere HIRLAM, which is operational at the Swedish Meteorological and Hydrological Institute (SMHI) in Norrkoping. The water exchange between the Baltic and the North Seas is prescribed according to the forecasts made with a large-scale model of the North and Baltic seas (BSHcmod), which works operationally at the Federal Maritime and Hydrographic Agency in Hamburg.

Flood occurrences in the coastal area of the Baltic Sea in Estonia, though not very frequent, cause considerable damage. It has been estimated that because of the global warming, the flood occurrence frequency can increase substantially. During the last flood event on 8-9 January 2005, the water level increased by almost 3 m on the west coast of Estonia and flooded several coastal towns. The overall material damage from this event was estimated to be about 1 billion EEK (100 million US$). After regaining its independence in 1991, many coastal areas of Estonia became construction sites for civil projects, and the risk of flooding is often completely ignored.

Flood in the coastal area of Estonia, 2005

Large storm surges can be disastrous for many locations along the Baltic coast, but the statistics of extreme sea levels in the Baltic Sea are poorly known. In this project, we propose to use statistical analysis methods and numerical models to investigate extreme sea levels in the Baltic Sea and in the Gulf of Finland, based on the observations at coastal stations and on altimeter data from several satellites (TOPEX/POSEIDON, Jason-1, ERS-1/2, ENVISAT and Geosat Follow-On). Recent developments in the fields of remote sensing (such as altimetry) and ocean modelling have expanded considerably the opportunities for research about extreme sea levels, making it very timely to improve the methods for estimation of the risk of floods and their repeat periods. Until recently, storm surge models were calibrated using only coastal data. Introduction of spatial information from satellites makes it possible to calibrate such models for the complete area in question. The more traditional methods of estimating the risk of floods from coastal sea level data can now be significantly expanded using new data processing methods, numerical models and altimetry.

This project will include an investigation of interannual sea level variability in the Baltic Sea, with a special focus on the extreme sea level rise and flooding in the coastal areas of the Gulf of Finland and Eastern Baltic Sea. The projected results from this project are important socially – by increasing the awareness of the authorities and the population about the risk of flooding, as well as economically – by avoiding or mitigating the potential for flood damage to harbors, coastal structures and navigation.

Our main goal is to improve the quality (reliability, advance time and precision) of operational flood warning in terms of upgrading the existing flood forecasting technology, while taking into account complementary information, including additional coastal observational and satellite altimetry data, as well as new information about the storm surge dynamics. Among the improvements to the forecasting system, a provision is made for development of a new refined model of the Neva Bay and the Eastern Gulf of Finland, in order to study the influence of the St. Petersburg Protection Barrier on the local water quality.

Woodcut from a 1607 pamphlet depicting the aftermath of the flood in the Levels around the Severn Estuary

Statistical assessment of extreme water levels along the EGF coast will be supplemented with hydrodynamic modelling. A number of numerical experiments will be done with the Baltic Sea model BSM5. Atmospheric input in the model will be based on data on extreme winds for this area. Such data are available from studies made at the Main Geophysical Observatory in St. Petersburg.

Access to more powerful computers will make it possible to develop more refined models of Neva Bay and EGF and to make a number of numerical experiments aimed to evaluate the influence of St. Petersburg flood Protection Barrier on water quality in the Bay in greater detail than before [Klevanny, 1999, 2003].

Improvement of infrastructure of sea-level observation and information systems and installation of modern tidal stations, development of methodology and technology of sea level observation will be done for better early flood forecast system. Development of sea level database and implementation of statistical methods will benefit flood risk estimation.

Knowledge gained from new research, enhanced observational network and modelling activity in this project will contribute significantly to capacity building for flood prediction and mitigation in Russia, Estonia and other countries in this region.

The use of altimetry data from multiple satellites will provide new tools and techniques for analyzing trends, seasonal cycle, interannual variability and extreme sea level events in the Eastern Baltic Sea and in the Bay of Finland and their relation to observations from coastal stations and to atmospheric forcing data, both on short and long time scales. The altimetry data will also be used for verification and calibration of regional tidal and circulation models in the complete area in question, thus contributing to the analyses of physical mechanisms responsible for such variability.

A stochastic model for the sea level space-time variability will be constructed, on the basis of which a method to estimate the return periods of extreme sea level events will be developed. Long-term historical sea level time series measured in Finland, Estonia and Russia will be analyzed from the point of view of extreme sea level events and their possible causes.

Newer and more powerful computers will allow the use of ocean models, which calculate circulation and sea level features with a precision suitable for engineering applications. Physical models, such as POM (Princeton Ocean Model), can make reliable predictions of sea level changes due to tides and storm surges. Such models are used already for short-term forecasting. However, due to computational limitations, numerical models are rarely used to obtain multi-year statistics of flood events.

In this project, we propose to develop a fast and efficient modelling algorithm, which will allow us to carry out multi-year simulations of sea level variations in the Gulf of Finland. This fast “model” will use a database of prepared sea level response to prescribed wind forcing. A finite number of “typical” sea-level response scenarios will be calculated using the method of empirical orthogonal functions. A similar approach has been developed by Harper and McMonagle [1983] using the so called discrete Monte-Carlo statistical model.

Application of effective numerical model with elements of satellite data assimilation is the next problem to which the project is addressed. Improvement of technology of operational flood forecasting, which is one of the project tasks, is of big importance for St. Petersburg and enterprises located along the coast of the EGF. Among them are Leningrad Nuclear Power Station and new ports in Ust’-Luga, Primorsk, Vyborg, Vysotsk. Successful operation of St. Petersburg Flood Protection Barrier, which completion is planned in 2008, also needs a reliable Flood Warning System. Installation of new automated gauges for recording of water levels in different points in the EGF would be very desirable. This provides a basis for the implementation of a data assimilation module (e.g., using a Kalman Filter) in the Flood Warning System. At present, as it was indicated above, flood forecasts are based mainly on the forecasts of wind and pressure over the Baltic Sea, while tide gauge observations and altimetry data were not taken into account. Experience gained with similar systems in other countries proved that data assimilation improves the accuracy of water level forecasts.

Aiming at development and improvement of predicting technologies based on existing flood forecasting system, we plan a ‘structural’ analysis of storm surges in the Gulf of Finland and related floods on its shores. The main goal of this analysis is to estimate the relative contributions to coastal water uplift from two major factors: (1) wind-driven drift and (2) long-wave effects. It is proposed that such an estimate can be implemented using analyses of kinematic characteristics (displacements, trajectories, vertical profile of velocity, etc.) of water parcels from numerical experiments. Using the high-quality flood forecasts in St. Petersburg by the CARDINAL modelling system, we will consider the calculated three-dimensional components of water particles velocity and displacement to specify the structure of water motion produced by either of these factors. It is expected that such analysis will allow to assess the relative contributions of these factors in resulting sea level changes, depending on specific weather conditions and for different parts of the Gulf of Finland. These results and understanding of the processes involved are expected to be important for further development of storm surge and flood predictive methodology.

Tsar Peter the Great saves the drowning during the flood of 1724

To gain a better understanding of other factors contributing to storm surges in St. Petersburg, an in-depth investigation will be useful. According to the theory, sea-level variations during St. Petersburg floods may be due to a number of causes, such as wind-driven currents, gravitational, gradient-vorticity and topographic Rossby waves, Neva River discharge, etc. These will be studied and compared to numerical model experiments. The mechanism of energy transfer from air to water by anemobaric forces, as well as the role of the resonance effects will also be investigated under the specific conditions that are prevalent in the Gulf of Finland.