Voinov and Zakharchuk: Large-Scale Variations of Sea Level in the Laptev Sea
27
advance. Contributions of the components 0 1 and QI does not exceed 0.4 cm, and they are free
from aliasing for frequencies of the long-period constituents. Contributions of N2 and M4 peaks
in the daily mean are respectively 0.8 cm and about 1 cm at Preobrazhenia, but are of no
practical importance for the lack of aliasing in the long-tidal constituents. Contribution of all the
semidiurnal or diurnal constituents is less than 0.1 cm.
Gaps in data were revealed at some points (Peschaniy Cape, Andrey Island, Terpiay-Tumsa
Cape, Dunay Island), and in two cases this gap was up to 1 year long. To a first
approximation, these gaps were omitted in the harmonic analyses. In the subsequent analysis
the gaps were restored using the predicted long-period tides, and the harmonic analyses was
recirculated.
The non-tidal oscillations and the long-period tides were separated with the aim to study the
large-scale level variations. The predicted long-period tide was obtained on the basis of 10
constituents with the most significant magnitude ( Mn, Sa, Ssa, Sta, Msm, Mm, Msf, Mf, Mtm
and Msw). Non-tidal (residual) values of the daily mean sea level were derived after subtraction
the predicted long-period tide values from the initial mean daily data. As it is known, the
constituents Sa (365.26 days) and to a smaller degree Ssa (182.62 days) also are determined
mainly by hydrometeorological means such as change of temperature and salinity of water and
by various processes in the atmosphere. However, the tidal component of the tide Sa is almost
inseparable from annual variations of hydrometeorological factors. However, physical
mechanisms causing the Sa and Ssa constituents remain in the aggregate fixed in time, and
interannual variations of amplitudes and phases of these constituents are, therefore, not great.
Hence, the Sa and Ssa constituents are predictable for most practical purposes.
Spectra for initial series of the level oscillations were obtained using the orthodox method of
spectral analysis by computing the cosine transforms of the autocorrelations (Jenkins and
Watts, 1969). Amplitude and phase of more than 4000 harmonics were obtained in the result of
the Fourier analysis for each of the residual series. Variance of the global oscillations of the sea
level (interannual variability) was calculated by taking the sum of each of the variance from I to
22 Fourier harmonics. The variance of the oscillations for the periods less than I year was
derived after subtracting the variance of global oscillations from the variance of the residual
series.
For complete investigation of the oscillations of the synoptical scale (periods from days to
months) the residual series were divided into semi-annual successive sections (46 sections of
realization for each of the station). The Fourier analysis was carried out for each section. Twodimensional distributions of temporal variations of the amplitudes of the synoptical scale
oscillations were constructed on the basis of this analysis. We excluded the oscillations with
periods more than half a year from the residual series by polynomial smoothing for
investigation of temporal variability of the variance of the synoptic scale level oscillations. The
thus obtained series were divided into monthly successive realizations. The variance was
calculated for each realization.
The cross- spectral analysis was carried out between sea level oscillations of synoptical scale
at different stations in the Laptev Sea for estimating parameters of low-frequency waves.
Spatial distribution of the phases indicates that these oscillations represent mainly progressive
waves in the range less 60 days. Estimating the wave parameters was performed for typical and
uniform cases in the Laptev Sea using the results of the Fourier analysis. Calculation of wave
propagation velocities was done using the formula
V=l!'t
(2)
where I is the distance between stations, 't is the delay period in days equal ,1(jl*Tl360°, ,1(jl is
27
advance. Contributions of the components 0 1 and QI does not exceed 0.4 cm, and they are free
from aliasing for frequencies of the long-period constituents. Contributions of N2 and M4 peaks
in the daily mean are respectively 0.8 cm and about 1 cm at Preobrazhenia, but are of no
practical importance for the lack of aliasing in the long-tidal constituents. Contribution of all the
semidiurnal or diurnal constituents is less than 0.1 cm.
Gaps in data were revealed at some points (Peschaniy Cape, Andrey Island, Terpiay-Tumsa
Cape, Dunay Island), and in two cases this gap was up to 1 year long. To a first
approximation, these gaps were omitted in the harmonic analyses. In the subsequent analysis
the gaps were restored using the predicted long-period tides, and the harmonic analyses was
recirculated.
The non-tidal oscillations and the long-period tides were separated with the aim to study the
large-scale level variations. The predicted long-period tide was obtained on the basis of 10
constituents with the most significant magnitude ( Mn, Sa, Ssa, Sta, Msm, Mm, Msf, Mf, Mtm
and Msw). Non-tidal (residual) values of the daily mean sea level were derived after subtraction
the predicted long-period tide values from the initial mean daily data. As it is known, the
constituents Sa (365.26 days) and to a smaller degree Ssa (182.62 days) also are determined
mainly by hydrometeorological means such as change of temperature and salinity of water and
by various processes in the atmosphere. However, the tidal component of the tide Sa is almost
inseparable from annual variations of hydrometeorological factors. However, physical
mechanisms causing the Sa and Ssa constituents remain in the aggregate fixed in time, and
interannual variations of amplitudes and phases of these constituents are, therefore, not great.
Hence, the Sa and Ssa constituents are predictable for most practical purposes.
Spectra for initial series of the level oscillations were obtained using the orthodox method of
spectral analysis by computing the cosine transforms of the autocorrelations (Jenkins and
Watts, 1969). Amplitude and phase of more than 4000 harmonics were obtained in the result of
the Fourier analysis for each of the residual series. Variance of the global oscillations of the sea
level (interannual variability) was calculated by taking the sum of each of the variance from I to
22 Fourier harmonics. The variance of the oscillations for the periods less than I year was
derived after subtracting the variance of global oscillations from the variance of the residual
series.
For complete investigation of the oscillations of the synoptical scale (periods from days to
months) the residual series were divided into semi-annual successive sections (46 sections of
realization for each of the station). The Fourier analysis was carried out for each section. Twodimensional distributions of temporal variations of the amplitudes of the synoptical scale
oscillations were constructed on the basis of this analysis. We excluded the oscillations with
periods more than half a year from the residual series by polynomial smoothing for
investigation of temporal variability of the variance of the synoptic scale level oscillations. The
thus obtained series were divided into monthly successive realizations. The variance was
calculated for each realization.
The cross- spectral analysis was carried out between sea level oscillations of synoptical scale
at different stations in the Laptev Sea for estimating parameters of low-frequency waves.
Spatial distribution of the phases indicates that these oscillations represent mainly progressive
waves in the range less 60 days. Estimating the wave parameters was performed for typical and
uniform cases in the Laptev Sea using the results of the Fourier analysis. Calculation of wave
propagation velocities was done using the formula
V=l!'t
(2)
where I is the distance between stations, 't is the delay period in days equal ,1(jl*Tl360°, ,1(jl is