Fróðskaparrit - 01.01.1975, Blaðsíða 70

78
Sca Level Fluctuations in Tórshavn, Preliminary Results
Nautical Department of the Danish Meteorological Institute
(DMI). The gauge digitized the measured values and stored
them on papertapes, which were changed each month by the
Harbor master of Tórshavn. Edited data were supplied to the
author by the DMI.
The air pressure data also were supplied by the DMI, they
had been recorded at the Meteorological station of Tórshavn
approximately 2 km’s distant from the gauge and had a samp-
ling time interval of 3 hours.
Parts of the analysis require that there be no gaps in the
time series. In the sea level series there were two consecutive
registrations missing out of a total of about 30.000. The values
of these were interpolated and filled in. The meteorological
series was much less complete, as about 2 % of the registrations
were missing, rather evenly distributed in the period. These
values also were interpolated and inserted at those parts of the
analysis where it was necessary. It is not felt that these mani-
pulations could have any influence on the results presented.
Methods of analysis
We shall assume the sea level h to vary in time according
to the equation
h(k) = i c• • cos(2• n■ f • • k* At + ip.) + h_(k) + h„(k) (1)
•j_ -*■ l r K
At being the sampling time interval and h(k) the height of sea
level at the time k • At. The first term on the right is the har-
monic part containing the astronomical frequencies fj . hp is
the sea level height due to air pressure and hR is the rest, which
we may here consider as noise.
Various methods exist for extracting the parameters fp c^
<p; from a tidal series. The method mainly employed in this
work is rather unconventional, firstly as it used the Power
Density Spectrum (PDS) — thereby making no assumptions
as to what frequencies are present — and secondly because the
PDS is estimated by the Maximum Entropy Method (MEM)
rather than by the traditional use of either the Fast Fourier
Transform or the Autocorrelation function.