Data and definition
11
2. Data and definition
2.1. Data
This monograph was prepared using hydrological
and meteorological data archived at the BSH and
IMGW, and available published data. Mareo-
graphic records were obtained from the water
gauges at Wismar, Warnemünde, Sassnitz,
Swinoujscie, and Kotobrzeg. All time series start
around 1955 and have been analysed up to and
including 2005. For most of the period reviewed,
the German data are available at 1 -hour intervals.
All Polish data recorded during periods with
extreme sea levels (negative and positive storm
surges) have been digitised to hourly values. The
majority of all other data has been digitised to
4-hourly values.
The zero level of tide gauges in Schleswig-Hol
stein and in Poland is PN = NN-500 cm. In Meck
lenburg-Vorpommern, the zero level is
PN = HN-514 cm, but the reference level used
until October 1985 was NN. Therefore, the gauge
station data recorded prior to October 1985 were
corrected to match the currently used definition
(Die Küste: Die Wasserstände an der Küste
Mecklenburg-Vorpommerns von Hans-Joachim
Stigge).
At the Wismar station, there are large data gaps
in 1963. The data from July to September are
missing because the station was out of service
due to maintenance work; these data have been
interpolated using data from the nearby gauge at
Timmendorf. The correlation coefficient between
Timmendorf and Wismar, using the same time
span as in linear regression, i.e. 30 days before
and 30 days after the data gap, is 0.98. The
standard deviation of the differences between
measured and regressed data is 3.8 cm, and the
range is -15.0 cm to +14.0 cm. Also the hourly
data for January and February 1963 at Wismar
and Timmendorf are missing because the record
ing devices did not work due to icing. Therefore,
only single daily values (7:00 CET) and the lowest
and highest values of each month are available.
The lowest value in January was 410 cm, and in
February 450 cm. The available values were line
arly interpolated to hourly values.
There are two gaps in the time series at
Swinoujscie, with three months’ data completely
missing: November 1978, and August/Septem
ber 1982. These values were interpolated with
the values from Kotobrzeg and Sassnitz using a
linear function. The parameters of this function
were calculated using the data of one complete
year preceding the missing month in each case.
The interpolated curve was corrected using a
linear drift, so that there were no jumps between
interpolated and measured values.
At Kotobrzeg, there are several days for which
only a single daily value is available. A total of
805 measurements, corresponding to about
268 days, are missing from the 4-hourly time
series. The missing values were interpolated by
linear interpolation. As the data available all came
from stations located more to the west, the
method did not work satisfactorily for
Swinoujscie.
Extreme values from the digital gauge data may
deviate slightly from published extreme values
because the digital data have a discrete sam
pling interval (1 h or 4 h), and extreme values
from continuous observations may have occur
red between these intervals. There also is a dif
ference between hourly and 4-hourly data. The
difference was determined by subsampling of the
hourly data at Warnemunde. The resulting
4-hourly values were interpolated back linearly to
hourly values. The effect of the reducing opera
tions on the annual mean sea level is very small
(less than 0.1 cm maximum deviation). The
4-hourly data obtained by subsampling showed
equal or higher annual minimum sea levels, with
a maximum difference of 9 cm. The mean differ
ence of all annual values was 1.9 cm with an
standard deviation of 2.4 cm. The difference in
the number of hours per year in which water levels
were below 440 cm ranged from -13 to +9 hours,
with a mean difference of 0.5 ±4.3 hours using the
4-hourly data (see Fig. 2.1.). With the linearly inter
polated data, values ranged from 0 to +18 hours,
the mean difference being 6.3 ±4.4 hours.
