Untersuchungen zur hochauflösenden Wasserstandsvorhersage an der gesamten deutschen Nordseeküste

Sea and estuarine dikes protect more than 2.4 million people in Germany, making them one of the most important coastal protection structures. A failure of these structures would have serious consequences as e.g. during the Hamburg flood in 1962, where one sixth of the city were flooded and more than 300 people lost their lives as a result of several dike failures. Therefore, early recognition of dangers and disaster prevention are elementary to ensure reliable coastal protection. Existing early warning systems for coastal protection in Germany are based on water level measurements and forecasts provided e.g. by the Federal Maritime and Hydrographic Agency in cooperation with the German Weather Service. The predictions of the water level are currently provided selectively for individual tide gauge locations. However, it is known from water level records and their investigations that such tide gauge based water level information is not representative for a larger area, e.g. for the German Bight, or for smaller geographical units, e.g. a Hallig. Thus, local effects and nonlinear interactions can result in spatial water level differences in a range of decimetres along a single coastal section. Especially along complex coastlines, such as the German North Sea with islands, bays, estuaries and tidal flats, a simple interpolation between tide gauge locations is inaccurate. This thesis deals with the development of a new methodology for the prediction of water levels at higher resolution based on existing approaches and models for the entire German North Sea coastline. A hydrodynamic numerical model is used to simulate water levels for the entire coastline of the German North Sea. The modelling is carried out on the basis of currently available bathymetric information, meteorological and astronomical boundary conditions as well as the observed changes in the mean sea level. Next, the water level information is separated into tidal and non-tidal components. The non-tidal residual is applied to derive empirical-statistical models using multiple linear regression relationships. Regression coefficients are derived using meteorological boundary conditions as input. The statistical approaches presented here also aim at incorporating the nonlinear interaction between tide and non-tidal residual into the model chain (tidal synthesis and non-tidal residual prediction). As a result, a first of its kind water level prediction at high spatial and temporal resolution along the entire coastline of the German North Sea (including islands and Halligen, point distance ~1 km, hourly values) is presented. Based on the 2013 storm surge “Xaver“, the procedure was applied practically and then successfully integrated into an operational test operation. This work thus makes a significant contribution to the extension and optimisation of existing early warning systems for coastal protection.