The history of the Baltic Sea level observations in the Gulf of Finland at coastal hydro-meteorological stations extends over many tens and even hundreds of years. Sea level at Kotlin Island near St.-Petersburg has been measured since 1703. In the last century, sea level observations in the Gulf of Finland were taken at more than 40 stations. As a result of such long time series, rich observational material has been collected. However, this material has not been used widely, mainly for two reasons. First, in the former Soviet Union, the data on Baltic Sea level observations were classified as confidential. Second, most of the data collected before the wide use of computers, is stored on paper drums. Therefore the data from Baltic sea level observations of the Russian National Oceanographic Data Centre (RNODC) is represented in electronic form only for the period from 1977-78 until present [Oleinikov et al, 2001].

Subdivisions of the NW branch of FSRHEM gather the observations of sea level in the Gulf of Finland from the Russian hydro-meteorological stations. The State Oceanographic Institute (SOI) has been responsible for quality control (periodical inspections) and collection of sea level data from these stations for many years. For this reason the SOI has at its disposal a rich archive of historical sea level data for the Gulf of Finland (on paper drums).

It is clear that flood risk analysis for the Gulf of Finland will become more comprehensive as more sea level data are used. Therefore, within the framework of the present project, significant efforts will be directed towards sea level data collection and transformation of the early data into electronic form.

It is planned to create a relational database on Baltic sea level (BSLDB). Besides its use in the present project, BSLDB will also be of intrinsic value to other projects on the Baltic Sea. The digitized sea level data will be provided to RNODC (one of the end-users of the project - see the letter from N. Mikhailov in Appendix 5), thus expanding a number of potential contributors to investigations of flooding in the Baltic Sea.

Sea level observations at RNODC are now stored in a specific file system [Oleinikov et al, 2001]. It is planned to develop the relational data model and the database on Baltic sea level. The methodology of this relational database will be passed to RNODC, where it can be used for other seas.

It is proposed to place the BSLDB on the SOI server in the ORDBMS PostgreSQL, under the Linux operating system. The architecture of the ORDBMS PostgreSQL allows to apply this DBMS in the client-server environment in the multi-user mode. The ORDBMS PostgreSQL is oriented to the TCP/IP protocol – it can be a local network or Internet. The client applications can be developed in a user-friendly environment, such as the Windows environment. Using an ODBC driver, it is possible to work with a database created in the PostgreSQL using spreadsheet programs, such as Microsoft Access and Excel. It is essential, that the license permission on the usage of the ORDBMS PostgreSQL it is not required (free-ware). At the same time, the ORDBMS PostgreSQL can successfully compete with certain commercial DBMS. It is planned to create a graphic interface to work with the BSLDB. The ERWIN software will be applied as a toolkit during a design stage for the development of a logical scheme and a physical model of the BSLDB.

It is necessary to note that the digitized sea level data will be used as soon as they are prepared, regardless of the readiness of BSLDB.

Logical steps in the creation of BSLDB:

• Compilation of a detailed catalogue of all historical data from Baltic sea level stations in the Gulf of Finland and gathering of metadata using the library archives of SOI and RNODC.
• Development of conceptual, logical and physical models of the Baltic sea level data and metadata.
• Purchase and installation of the application server in SOI. Installation of the Linux operating system and the ORDBMS PostgreSQL software on the server. Connection of ORDBMS to the institute’s network.
• Creation of the BSLDB on the SOI server on the platform of ORDBMS PostgreSQL.
• Creation of the user interface to this database.
• Filling of the database with sea level metadata.
• Preparation of sea level data from the coastal station network (scanning of the printed materials with historical sea level data using optical character recognition of the scanned material, creation of digital datasets, visual control of the level data in digital datasets, comparison with original data in printed form, editing, conversion between data formats).
• Filling of the database with monthly mean, annual mean and extremal Baltic Sea level data for all historical periods of measurements. Filling of the database with data of hourly and fixed-interval sea level measurements for specific time periods (periods of floods etc.).
• Preparation of a manuscript with the description of the BSLDB.
• Implementation of digitized historical Baltic Sea level time-series and methodology of implementation of BSLDB for RNODC.

(SOI, CDBHI)

The quality of flood forecasts in Saint-Petersburg is expected to improve if sea-level measurements at the network of hydrometeorological stations in the Gulf of Finland, and especially near Gogland and Moschny Islands (see the map in Appendix 2) are assimilated in numerical models. However, sea level near Gogland is now measured with the help of a wood level plane (depth-gauge), while there are no sea level observations at Moschny (see Yu. Malashin’s letter in Appendix 5). At other Russian stations in the Gulf of Finland, measurements of a sea level are carried out with the help of a level plane or using a float-type tide gauge placed in a well.

The modern method of measuring sea level using a depth pressure recorder is not used at the Russian station network in the Gulf of Finland. In the framework of this project, it is planed to introduce such pressure sensor close to a coastal hydrometeorological station located on the Gogland island.

Stages of work

Second project year

• Purchase of the sea level gauge GNU-3 using hydrostatic pressure sensors and special-purpose software and hardware (computer, power source, modem Fastpac M1203a, cables).
• Laboratory tests of the GNU-3 during one month at CDBHI (Obninsk).
• Delivery of the equipment to the experimental marine base of SIO in Gelendjik by a car.
• Training of young participants of the project to work under water with the aqualung.
• Installation of the sea level gauge in the sea.
• Organization of on-line transfer of information from the gauge to the coastal computer or Internet.
• Test measurements and calibration of sea level.
• Processing of the sea level data received.
• Training of the young participants in data processing.
• Preparation of the report with description of the field work.
• Delivery of the equipment from Gelendjik to Obninsk by a car.
• Participation in cruise to Gogland island for an acquaintance and spade-works. Partial rent of a vessel (there is no regular shipping from Saint-Petersburg to Gogland island).

• Delivery of the equipment from Obninsk to Saint-Petersburg by a car.
• Rent of a vessel for expedition to Gogland island.
• Sea-bottom survey near Gogland island and selection of the gauge installation site.
• Installation of the gauge on the sea bottom.
• Organization of automatic transfer of sea level data by cellular communication to the computer of the end-user in Saint-Petersburg.
• Preparation of a manuscript with a description of the methodology of the deployment of sea level gauges using hydrostatic pressure sensors.
• Transfer of the methodology and the working sea level gauge to the end-user (SPCHEM).
• Delivery of the methodology to the Estonian partner (MSI).

(SOI, CRIGASC)

Verification of heights of benchmarks near sea level gauges along the Russian coast of the Gulf of Finland and gauge datums was not carried out for a long time. Heights of benchmarks on the Gogland island are completely absent in the Main geodetic archive of Russia. These benchmarks were installed on the Gogland island by Finland before 1941.

In this project it is planned to carry out field levelling works (in the first and third project years) in order to raise quality of sea level measurements at Russian stations in the Gulf of Finland to the international standard. With a help of the modern equipment (GPS-receivers Trimble 5700) a conjunction of heights of benchmarks near sea level gauges to the world satellite Global Positioning System will be done. Then with a help of the geodetic levelling instrument DINI-22, a levelling of the gauge datums to the benchmarks heights will be carried out.

It is important to note that repeat fixes of heights of fundamental benchmarks using GPS-receivers will allow monitoring vertical land motion. According to certain scientists (http://www.vesti.ru/news.html?id=48551) such motion can be significant around St. Petersburg.

(SPPU)

The Baltic Sea model BSM5, which is the core of the automated flood forecasting system in St. Petersburg, was developed with the modelling system CARDINAL and will be used for different tasks. One of them is to perform numerical model studies of sea-level extremes in the EGF. Statistical assessment of extreme levels along the EGF coast will be aided with hydrodynamic modelling. A number of numerical experiments will be done with BSM5. Atmospheric input in the model will be based on data on extreme winds for this area and estimated return periods. Such data are available from studies made at the Main Geophysical Observatory in St. Petersburg [Borisenko et al., 1997]. These data were already used for estimation of extreme water levels near the Leningrad Nuclear Power Station and in St. Petersburg [Klevanny, 2004]. In this project such evaluations will be done for the complete EGF, based on recent developments in the parametrization of sea bottom drag under extreme storm conditions.

(SPPU, RSMHU)

Improvements in the accuracy of operational flood forecasting are of significant importance for St. Petersburg and enterprises located along the coast of the EGF. In this relation, a number of measures are planned, in order to take into consideration the data from new automated gauges being installed in different points in the EGF. Joint efforts of project partners show considerable promise of extending the amount of data which can be used as a basis for modernization of the operational Flood Warning System by sensitivity analysis and re-validation. Methods of data assimilation (coastal and satellite altimetry) will be investigated, in order to improve the quality of the operational sea-level forecasts that are based on numerical model simulations. It is expected that implementation of observational sea-level data will noticeably increase the forecasting quality in comparison with that when only the forecasts of wind and pressure over the Baltic Sea are taken into account, as is the case at present. A great importance will be given to development of the Flood Warning System, in the context of planned completion of the St. Petersburg Flood Protection Barrier in 2011. Without such warning system, the Barrier will not operate properly.

A number of numerical experiments will be carried out, focused on acquisition of kinematic characteristics of water particles motion in different parts of the Gulf of Finland under various weather conditions. A methodology of separation of particles motion into those produced by long-wave and drifting factors is to be worked out based on features of particle orbits, amplitude/phase relationships between horizontal and vertical movements, temporal and spatial variability, vertical profiles of currents, their correlation with wind, etc. In this way, the contribution of drifting and long-wave factors to sea level change can be assessed at different stages of a storm surge and in various parts of the area. This will then be used for improvements in forecasting procedures.

Numerical experiments will be accomplished with various simplified scenarios, followed by statistical analysis of the model results. Such simplified models allow to investigate different factors, such as barotropic/baroclinic conditions and various forms of friction, topography, nonlinearity, and boundary conditions. This analysis will provide estimates of contributions of these mechanisms to the flooding, with and without the protective barrier. The complete solution will be obtained, for comparison with partial simplified cases and to elucidate the relative roles of natural and man-made impacts on flooding in St. Petersburg.

(SPPU)

Simulation of dispersion of pollutants in Neva Bay will be done with the updated model of Neva Bay and EGF. The numerical method, which is used in the model to simulate advection of pollutants (the most difficult problem from the numerical point of view) has the 3rd order of accuracy and model validation proved such simulations to be of high quality. While the Neva Bay has been studied during several decades, it is realized now that even more refined and detailed numerical models are needed for reliable forecasts of influence of various hydrotechnical construction projects on water quality in the Bay. The Bay is shallow and there are many small-scale objects, which impact significantly on water dynamics. These are narrow navigation channels, different dams, including 12 dams of the Flood Protection Barrier, and underwater protection lines made from piles and stones in the previous centuries. In recent years, bathymetry and the coast of Neva Bay were changed considerably by hydrotechnical works and natural processes. These changes are not reflected in the existing maps. Accurate simulations of dispersion of pollutants discharged from outfalls in Neva Bay require in many cases a small grid size, at least in the vicinity of the outfalls. Therefore development of a newly refined and accurate model of the Neva Bay and EGF will need much effort and an access to a powerful computer.

With this new model and a computer, it will be possible to fulfil a large number of numerical experiments aimed to evaluate the changes in water dynamics and concentrations of pollutants under different weather conditions and in the presence of different hydrotechnical constructions.

(MSI, SIO, SOI, CRIGASC, HMTI)

Sea level data that are in the tails of statistical distributions have been traditionally modeled with a Gaussian distribution. This inherent assumption of many statistical calculations can be dangerous for applications such as damage detection, which deal mostly with those extreme data points that are not accurately modelled by the Gaussian assumption. Extreme value statistics (EVS) focuses on modelling these extreme events without knowing their parent distributions. Modelling the tails simplifies the decision-making (establishment of decision boundaries) to some extent because the extreme values follow one of the three EV distributions: Gumbel, Weibull, or Frechet. In this Project, risk analysis and damage detection will be reworked to take advantage of these extreme value distributions. A programming technique for solving the nonlinear optimization problems will be adopted to estimate the parameters of the extreme value distributions. In particular, no specific type of the distribution for each sample data needs to be known a priori. The effectiveness of EVS will be tested on numerical examples. Thresholds obtained from the actual distribution, the best-fit normal distribution and the extreme value distributions will be compared. It will be shown that the best-fit extreme distribution can be determined a posteriori by applying the indicated parameter estimation technique to real sample data.

Gumbel, Weibull, and Frechet Density Functions

Normally sea level variations are treated as stationary stochastic time series, which include floods events, as a Poisson-type process. Actually, the probability of flood is seasonally modulated with a maximum risk occurring in Fall and in Winter. Occurrence of floods also has climatic variability. There were periods of 5-7 years and probably one period in 19 years (1803-1821) when St-Petersburg did not experience any floods. For an overview of occurrences of very dangerous floods, see Table 7.1.

A fast and efficient modelling algorithm will be used to carry out multi-year simulations of sea level variations in the Gulf of Finland. Forced by prescribed (from reanalysis) wind forcing, this model will allow to extend the statistics of sea level data in the Baltic Sea - so called discrete Monte-Carlo statistical model.

Probability distribution of maximum sea levels in Neva Bay

The long-term sea level trends in the Baltic Sea will be analysed using the historical archive of sea level data, as well as data of all available levelling for the Eastern Gulf of Finland. It is important to make estimates of the time scales for sea level rise to critical values in the Gulf of Finland, in order to determine the coastal locations that need to be protected by dams.

Stages of work

• Formation of sea level time series based on historical data, including satellite altimetry data
• A programming technique for solving nonlinear optimization problems will be adopted to estimate the parameters of the extreme value distributions.
• Estimation of flood recurrence periods for coast of Gulf of Finland
• Multi-year numerical simulations of sea level variations in the Gulf of Finland
• Analysis of seasonal and climatic modulations of flood recurrence.
• Estimation of long-term sea level trends in the Baltic Sea based on historical data base and satellite altimetry data

(SIO, IOS)

Several sources of altimetry data will be used in this project. These include AVISO (ERS-1/2, ENVISAT, TOPEX/POSEIDON, Jason-1), NASA Pathfinder (TOPEX/POSEIDON, Jason-1, ERS-1/2, GFO), DEOS/NOAA RADS, and US Navy (GFO). Of particular interest (though not only) for this project are the collocated NASA Pathfinder tidalist data (include ocean tides), which allow detailed analysis of tidal constants in TOPEX/POSEIDON and Jason-1 data, as well as the use of local (vs global) high-resolution tidal models for detiding ERS, Envisat and GFO data. All such data will be extracted for the Baltic Sea and edited to remove erroneous outliers that are not of geophysical origin.

(SIO, IOS)

While the complete signal must be included in analyses of extreme sea levels, detiding of altimetry data allows separating the tidal effect from other physical mechanisms, as well as for process studies and verification of numerical models. The two-prong approach adopted in this project follows closely the procedures used at IOS. TOPEX/POSEIDON and Jason-1 tidal constituents are computed using a 2-step harmonic analysis of edited sea levels. These constituents are accurate enough to compute sea-level anomalies that do not show anomalous signals at aliased periods, even in the shallow seas. Other altimeter data are detided with the help of a numerical tidal model, with minor constituents computed using inference. Given the different characteristics and duration of various altimeter missions, customized procedures may have to be used for some (especially short-duration) altimetry data sets.

(SIO, IOS)

Harmonic analysis of TOPEX/POSEIDON and Jason data yields directly the annual and semiannual harmonics, while binning in monthly means provides the varying seasonal cycle for each data set. After removing the tides and mean seasonal cycle, the along-track residual time series will be processed to extract extreme values (scaled with standard deviation) which would correspond to specific geophysical events, such as storms. The detided altimeter data will also be combined with similarly processed sea-level data from coastal stations to map monthly and seasonal sea-level anomalies. Objective analysis or Kriging (with topography-orientated elliptical scaling) will be used to map the combined data onto a regular grid. Relative accuracy of such maps is expected to be between 1 and 5 cm, depending on local data coverage in each season (number of cycles) and occurrence of episodic events, such as storms. A similar analysis will be designed for extreme sea level events that have large-scale signature.

(MSI, SIO, IOS)

It is planned to make estimates of spatial variability of sea level in Gulf of Finland and to calculate the last decadal trend and variability on interannual time scales. After removal of these and seasonal signals, the anomalies in altimetry data will be examined for sea level extreme events that are correlated with atmospheric or remote forcing in the proper Baltic Sea. These data will be combined with coastal station data for extreme values statistical analysis, which was explained in Task V.

(SIO, RSHU, HMTI)

Verification and calibration of numerical models will be accomplished using coastal tide gauge and satellite altimetry data, as well as their computed statistics.

(SOI, SIO, MSI, RSHU)

The information on project status, development and results will be transferred directly to the principal end-users of the project and, with their help, to the Baltic Sea user groups and societies. In the case of Estonia, Estonian Environmental Centre is identified as an end-user. In Russia, Head of the Leningrad District Hydrometeorology Center and the Administration of the Flood Protection Barrier, under the Ministry of Natural Resources of Russian Federation, are the end-users of this project, ensuring reception and transfer of the project information and results to communities on the Russian coast of the Baltic Sea and to general public. Development of sea level data base will be coordinated with RIHMI-WDC. Regular consultations with the end-users will be made through the project coordinator and directors as well as during project meetings.

(SPSU, SOI, SIO, MSI, RSHU, IOS)

Regular training of young participants of the project (students, researchers) will take place during project activities by participation and exchange of persons from each institution on work performed for particular tasks. These will be in the form of short-term working visits to counterparts. Participation in international meetings and in writing of peer-reviewed publications will also be considered as training for the young researchers presenting their scientific results.

(all co-directors)

A training workshop “Flood Risk Analysis for the Gulf of Finland and Saint Petersburg” is planned in the second quarter of the first year of the Project. This workshop will serve to introduce the project initial results and to update the participating scientists about the project development and outstanding research, as well as provide the forum for consultation with invited experts from other similar projects in Europe.

(SIO)

Project web site will be created and maintained during the duration of the project in order to make its results visible to a broad scientific community, while allowing additional communications between the various centres participating in the project. Modern web technology (database back-end and PHP or ASP scripting) will support distant database management and data exchange.