Variability of Sea Surface Heights in the Baltic Sea: An Intercomparison of Observations and Model Simulations

Abstract

Sea level changes in the Baltic Sea are dominated by internal, short-term variations that are mostly caused by the ephemeral nature of atmospheric conditions over the Baltic area. Tides are small and their influence decreases from western parts of the Baltic Sea to the Baltic Proper. Superimposed to the large short-term sea level changes (up to few decimeters from day to day) are seasonal and interannual variations (centimeters to decimeters). This study focuses on the comparison of sea surface heights obtained from observations and from a high resolution oceanographic model of the Baltic Sea. From this comparison, the accuracy of the modeled sea surface variations is evaluated, which is a necessary precondition for the further use of the oceanographic model in geodetic applications. The model reproduces all observed Baltic sea level variations very reliably with an accuracy of 5 to 9 cm (rms) for short-term variations (up to 2 months) and 8 cm (rms) for long-term variations (>2 months). An additional improvement of the model can be attained by including long-period sea level variations of the North Sea. The model performs well also in the case of extreme sea level events, as is shown for a major storm surge that occurred at the southern coast of the Baltic Sea in November 1995.

Keywords:

• Baltic Sea
• sea level variations
• oceanographic model
• tide gauge
• satellite altimetry
• storm surge

Acknowledgments

This investigation was supported by the German Federal Ministry of Education and Research as part of the German Climate Research Programme (project BASEWECS; FKZ 01LD0025). Furthermore, the authors would like to thank the Leibniz-Institut für Ostseeforschung Warnemünde (IOW), especially S. Krüger and W. Roeder, for the fruitful cooperation and technical assistance in maintaining the sea level measurement system at the MARNET stations. Wind data from the MARNET station Darss Sill were also provided by the IOW.

Notes

*Applying a Fourier transform to the time series, the total variance was divided into contributions from different frequency bands (low/high-frequency, periods ≥/<2 months, respectively).

*The total standard deviations of SLA differences (cf. Figure 7) can be considered the rss of different components.