Late Quaternary Climate Variations Reflected in Baltic Sea Sediments - The Baltic Sea Basin - page 105

sediment samples can be used here. The age of the facies boundaries are deter-
mined by interpolation between the age data above and below the boundary. In
a next step, the zonation of the sediment profile at the master station has to be
transferred to contiguous sediment cores along seismic cross-sections using lithos-
tratigraphic correlation methods for continuous logs of facies variables. We used
a numerical method deploying the principle of multiple cross-correlation of sed-
iment physical core logs (MSCL). The software CORRELATOR used (Olea and
Sampson 2002 ) is an implementation of machine correlation that mimics the more
conventional manual correlation of logs, which traditionally involves the simul-
taneous visual inspection of two logs per well, one of which is sensitive to the
amount of shale. Given a stratigraphic interval A in well X, interval A is com-
pared to intervals of the same length in well Y in order to find the interval in
well Y that displays the maximum similarity both in terms of the amount of shale
and in the pattern of fluctuations in the second log that are combined to produce
a weighted correlation coefficient. In addition to the simplicity and efficiency of
the approach, use of a weighted correlation coefficient has the advantage that the
coefficient is an index for the quality of matches. Thus, the weighted coefficient
can be used in combination with a threshold to eliminate correlations of low reli-
ability. The program was originally developed for situations typical in the oil and
gas industry. Yet the method has proved to be robust enough to satisfactorily work
for the circumstances prevailing in marine geology. Having carried out the cor-
relation for a grid of cross-sections the subsurface depths of stratigraphic zone
boundaries can be spatially connected by numerical interpolation methods. In the
result we receive digital elevation models of the subsurfaces of stratigraphic units.
Different thicknesses of stratigraphic layers can now be interpreted in terms of
paleo-dynamics in hydrography and sediment accumulation. Within the basinwide
model detailed studies on the downcore variation of proxy variables can be carried
out using methods of multivariate statistics. The (core-) depth to time transforma-
tion of the data is the main prerequisite for a time series analysis. We used the age
data for the boundaries of physico-stratigraphic units as “anchor point” and interpo-
lated between these points by applying the “piecewise cubic Hermite interpolating
polynomial” method (PCHIP command in MatLab). This method finds values of
an underlying interpolating function P ( x ) at intermediate points providing smooth
interpolation. This space to time transformation is supposed to produce smooth,
monotonic functions honouring all of the tie-points. Time series analysis allows the
extraction of information about these periodic components from time series data.
We applied periodicity analysis based on spectral density estimates by means of
fast Fourier transform (Bloomfield 2000 ) . Results have been additionally enhanced
with the help of “Hamming windowing and zeros padding technique”. All the data
records have been “de-noised” and detrended prior to periodicity analysis. For the
signal-to-noise enhancement, we used a singular spectral analysis (SSA) method
specially designed for noisy and not very long series (see Ghil et al. 2002 ) , which
originated from consideration of the theory of dynamical systems and multivariate
statistics.