18 
Summary 
6 Summary 
In the BSH-MOS project, the BSH's physical and statistical expertise in water level forecasting for the 
German North Sea coast has been combined with the MSWR-MOS standard technology, which is 
quite mature by now. The approach of defining the initialisation error of DMO as a predictor has been 
particularly successful, because a measurement-based initial distribution covering the entire North Sea 
area cannot be constructed for the 2D North Sea surge model, which differs in this respect from the 
models used in weather forecasting. Therefore, the predicted distribution from the previous forecast for 
a corresponding period, i.e. a 6-hour prediction in each case, has to be used as initial distribution. 
Also the proposal to base the classification not on seasons, as is usually done, but on HW and LW or 
positive and negative surges led to a major additional improvement as far as developmental results 
are concerned. This would not have been possible by adhering rigidly to the MSWR-MOS standard 
technique, in which a limited number of surge predictors would have been defined and optimised. It will 
have to be investigated in what way these results open up opportunities for automating water level 
forecasting for the German North Sea coast. 
Already now, the MOS equations found and the very good DMO allow the conclusion that complete 
automation may be possible at least in stable or slowly changing weather situations. All other weather 
situations - especially storm surge situations - will require human expertise in a man-machine mix also 
in the future (MÜLLER-NAVARRA, 2008). 
Acknowledgment 
The authors wish to thank Ms. Ingrid Lange (BSH) for her translation into English, and Mr. S. Palkowski 
(BSH) for providing to us the base data for operational use.