Verification of 6 Vz - months’ operational forecasts 
17 
It turned out that Mix is the best choice in periods in which RV(Kombi, DMO) is lower than 25%. The 
surge predictand delivered operationally what the development phase had promised. Kombi would 
need another 30% RV to meet the expectations arisen from the development data. Follow-up investi 
gations will show whether there is too much overfitting in the Kombi equations or whether the result will 
come closer to 60% with a prolonged verification period. 
Fig. 4: RMSEof wind surge forecasts 15.07.2009-28.01.2010 {DMO_2Dv4: 23770 cases; BSH-MOS (Kombi): 20911, MMM: 2937}. 
RV of MMM: 28,5%. 
Finally, the question remains as to the quality of BSH-MOS forecasts (here: Kombi) in comparison with 
independently made man-machine mix predictions (Fig. 4). To begin with, it should be noted that a 
period of 6V2 months is rather short and, accordingly, 4 daily MMM reference values may cause major 
differences from one lead time to the next. As 2 of the 6V2 verification months are relatively calm 
summer months, the RMSE level in Fig. 4 is a few centimetres lower than that in Fig. 2. 
Apart from that, the behaviour of MMM is similar to that in Fig. 2, with the exception of the jumps at 
6 and 14 h. Beyond 17 hours prediction time, it is hardly possible to improve DMO (2Dv4) by human 
intervention. However, it should be noted that the performance of BSH-MOS is better than that of MMM 
without major outliers, and it is only beyond 30 h prediction time that it cannot improve DMO further. 
The reason is that, due to technical problems with certain issue times, DMO cannot be used by 
regression for the longest lead times. This will be corrected in connection with the next software update. 
It may be concluded from these findings that the automation potential of water level forecasts for the 
German coast is very good. Especially the hourly updates to the forecasts based on latest gauge data 
and four daily prediction runs of the modelling chain covering all pilotage waters and harbours in the 
German Bight would be beyond the capacity of a man-machine mix. Here, future use of BSH-MOS 
might lead to a quality boost.