Die Kuste, 81 (2014), 273-290 
279 
Table 1: Amplitude and phase calculated from BSH-HBM output as well as the error compared 
to data taken from SHOM (1982) for M2- and S2-tide at selected stations at the German coast. 
Station 
M2 
amp 
[cm] 
M2 
amp err 
[cm] 
M2 
pha 
[deg] 
M2 
pha err 
[deg] 
S2 
amp 
[cm] 
S2 
amp err 
[cm] 
S2 
pha 
[deg] 
S2 
pha err 
[deg] 
Borkum 
107.6 
2.8 
278 
8 
28.1 
1.0 
338 
5 
Helgoland 
113.2 
4.6 
312 
0 
30.8 
1.9 
13 
-5 
Cult aven 
147.9 
13.5 
340 
-4 
37.1 
2.7 
46 
-7 
Buesum 
154.7 
-1.5 
341 
4 
41.0 
-1.1 
47 
0 
The results in Tab. 1 show that both the amplitude as well as the phase of M2 and S2 is 
represented quite well by the model at die considered stations. Only die amplitude at 
Cuxhaven has a significant error in comparison widi die SHOM data, but it should be 
kept in mind diat diere is also an uncertainty in the observed data. A recent analysis of a 
19 year time series at tide gauge Cuxhaven by BSH (personal communication Patrick 
Goffinet) gave a M2 amplitude of 138 cm, which would reduce the model error to about 
10 cm. Nevertheless the modeled M2 amplitude at Cuxhaven fails the BSH internal quali 
ty criteria at the moment and further calibration work is presendy carried out. 
When validating the water levels in the North Sea, the analysis is restricted to the peak 
values at high and low water because these are the values of highest interest. The exact 
timing of the peak values is not considered by this method. 
Table 2: Bias and bias-corrected root mean square deviation from observations (RMSD) of total 
water level peaks during high and low water at selected stations at the German coast. 
Total water level 
Station 
High water 
Low water 
Bias 
RMSD 
Bias 
RMSD 
[cm] 
[cm] 
[cm] 
[cm] 
Borkum 
-3 
11 
10 
14 
Helgoland 
3 
14 
10 
11 
Cuxhaven 
19 
18 
5 
14 
Buesum 
15 
18 
6 
17 
Whereas the bias’ shown in Tab. 2 at the stations Borkum, Helgoland and Cuxhaven are 
mainly explainable by the error in tides, die bias in Buesum can partly be explained by die 
very difficult local topographic conditions around tiiat station. The bias-corrected root 
mean square deviation from observations (RMSD) which is lower titan 20 cm at all sta 
tions, is already sufficiently but titere is of course potential for improvements in die fu 
ture. The RMSD is, however, at the same level as it is in die current operational model 
BSHcmod. 
For station Cuxhaven Fig. 5 shows die frequency distribution of high and low water 
level differences. If die bias is considered, die rate of events which are reproduced in a 
range of +10 cm is 39 % for high water and 49 % for low water. In a range of +20 cm it 
is 69 % for high and 85 % for low water. In die range of +30 cm more than 90 % of 
botii high and low water events are reproduced.