Possibility of statistical correction of hydrodynamic-numerical model results using the example of the storm surge of 1872

Life at the coasts of the Baltic Sea was and is characterized by storm surge events (e. g. Petersen & Rhode 1979). In order to improve the data basis and thus the decision basis for flood protection, the integration of information from historical events is recommended (e. g. in DWA M-552). Information on water levels, especially from historical events, is often available only locally. Therefore, in order to obtain a valid basis for the design at coastal sections without observed information about water levels, a hydrodynamic simulation for spatial and temporal information extension is often indispensable.

In November 1872, there was an extraordinary storm surge in the area of the German and Danish Baltic Sea coast of dimensions never reached since (e. g. Jensen &Töppe 1986; 1990). The water levels were above all previously known values - even though similar storm surge catastrophes have been reported from time to time in the last 1000 years. The storm surge, which is nevertheless often referred to as "singular" (or outlier), represents the beginning as well as the greatest challenge of modern coastal protection. Besides the exceptionally high water levels, the first measurement of the water levels, as well as the detailed description of the genesis, the course and the consequences are the unique aspects of this storm surge (e. g. in Baensch 1875). However, the available hydrographs are only locally available and may be associated with uncertainties. In order to obtain a complete picture of the water levels of the storm surge of 1872 the event was simulated using an existing hydrodynamic-numerical model of the Baltic Sea. Simulation, especially of extreme events, is often accompanied only by compromises in the accuracy of the results. In order to achieve results that satisfy high qualitative demands, model-generated data can be corrected locally using a statistical correction function (called bias correction), which can then be inversely distance-weighted and applied to the coastline. Using the example of the storm surge of 1872 at the southwestern Baltic Sea, the statistical correction of model results was demonstrated to be a suitable as post-proceeding for the optimization of model results. The assumptions made with this approach were tested and found to be valid. Thus, a semi-statistical and semi-hydrodynamic data set for the coastline of the Baltic Sea in the observation area was generated, which validly reflects the locally available water level observations during the storm surge and satisfies high demands. This can be used as a basis for optimization for coastal protection and disaster management for the German Baltic Sea coast.