> 
Modelling the water level data and calculating the tidal parameters | 31 | 
The analysis is usually carried out with observation series over 5 years. This is 
a shorter period than the 19 years used in detailed tidal analyses in order to 
take into account all relevant astronomical influences (including the nodal 
3ycle). Due to the shorter period, short-term developments (e.g. due to a 
shange in morphology or construction measures in the vicinity of the tide 
gauge) are deliberately given greater consideration. 
The input data is filtered as follows during the operational calculation of the tidal 
parameters: high water and low water data that have received special markings 
during processing by the authorities are not included in the analysis (disturbed 
data). Events for which the heights or intervals deviate from the respective 
mean value by more than three times the standard deviation are also filtered 
out before the analysis. The analysis of the semi-monthly inequality takes place 
in two iterations. After the first iteration, the data that deviates by more than 
three standard deviations from the determined model function is filtered out 
again. In the calculations for the tide tables 2003 to 2019 inclusive, the filtering 
of the observation data was based on 0.6 times the mean tidal range (instead of 
the criterion of three standard deviations). 
3.2 
Reference- 
based calcula- 
tion with sparse 
observation 
data 
A special case arises If there is not enough observation data available for a tide 
gauge over the desired period. In this case, the tidal parameters can be deter- 
mined using a reference-based calculation using water level data from a refer- 
ance tide gauge. The reference gauge must provide sufficient data so that the 
zoefficientS ag ref, Ajrer ANd b; 2: Can be calculated as described above (for 
a better overview, the indices HWH, NWH, HWI, NWI are omitted here). The dif- 
ferences Ayft) = y(t) - Yyrer(f) are determined for all high and low water for which 
data are available from both the reference gauge and the subordinate gauge. 
The least squares method is then applied to the difference 
4 
Ay(t) = y(t) — Yrer(t) = Aag + (has cos(igt) + Ab;sin(iogt)) 
1=1 
in ordert o determine the coefficients Aap, Aa; and Ab; Thus, the coefficients for 
the subordinate tide gauge are 
a9 = AQ9 + Ao,ref 
a; = Ad; + Qpref 
a; = AD; + Diret 
For an independent analysis, at least 500 * number_of_observation_years of 
nigh or low water are required in the operational service, where number_of_ 
observation_years is usually = 5. Ifthere is less data, a reference-based calcu- 
lation for high and/or low water is attempted