Using Rock Magnetic Proxy Records for Orbital Tuning and Extended Time Series Analyses
603
quality hysteresis data from miniature 10-20 mg
samples (Flanders 1988; von Dobeneck 1996).
Typical data sampling densities for high-resolution rock magnetic core-logging are 0.5-2 measurements/cm, which is more than the effective spatial
resolution of most sensors. This over-samplingresuits in smooth, well-defmed and complete records
- an important prerequisite for the performance of
signal analysis techniques. In this paper we present
susceptibility records acquired with a 'Bartington
F type' spot sensor at a 1 cm spacing.
Orbital Tuning Strategies
Milankovitch's (1941) visionary idea of synchronizing geological records to orbital variations has
evolved into a widely accepted chronostratigraphic
method (Imbrie and Imbrie 1980; Imbrie et al. 1984;
Martinson et al. 1987). Applied to records of isotopic or lithologic characteristics, 'orbital tuning' can
yield unmatched precision and consistency in dating Quaternary sediment series. Ideally it takes just
three steps to align a proxy record to its orbital
pacemaker:
- establish a rough initial chronology and detect
spectral characteristics of Milankovitch cyclicity,
- choose an astronomic target curve and shift it by
postulated or empirically determined phase lags,
- filter the proxy signal in the appropriate frequency
band and match it with its model record.
In practice this approach bears numerous pitfalls. Rock magnetic records, like most proxy signals, depend more on regional sedimentation conditions than on global equilibria. They are not, in principle, safe bets for orbital tuning purposes and their
patterns can be anything from a 'SPECMAP (Imbrie et al. 1984) double' to 'random wiggles'. Cyclostratigraphic age models critically rely on the
initial and thus very consequential age model. Simple peak-to-peak correlation of proxy and target
records is highly susceptible to erroneous matches
(Shackleton et al. 1995), which have their origin in
ordinary stratigraphic problems such as undetected
hiati or turbidites, condensed or extended sections,
bioturbation, lithologic and diagenetic effects. Algorithms such as signal slotting (Thompson and
Clarke 1993) or complex demodulation (Pisias et
al. 1990) have been used to find the statistically
most probable proxy-target-correlation. Yet a combination of magneto- and biostratigraphic, eventually radiometric ages should be the most reliable
basis to start from. Specific regional settings and
scientific objectives may require adaptation of orbital tuning schemes into flexible, case-wise adapted strategies.
Two contrasting examples illustrate the scope
of possible approaches at different interpretation
levels. They are based on two series of
FS METEOR gravity cores collected within a longterm Quaternary South Atlantic research project
(SFB 261) at the University of Bremen. The coring sites are located in the subtropical and in the
western tropical South Atlantic. The two study areas differ greatly in size as well as in prevailing
sedimentation conditions and rates. The first example outlines the possibilities and limits of signal correlation and orbital tuning in low accumulation environments. Covering nearly the full Quaternary,
these susceptibility records are rewarding targets
for evolutionary spectral analysis. The particular
interest of the second example lies in a much higher
sedimentation and signal detail opening the possibility of analyzing periodic sub-Milankovitch signal
components. We will primarily focus at the methodical aspects ofthese magnetic time series analyses.
Subtropical South Atlantic
The Subtropical §.outh ~t1antic §.usceptibility
(SUSAS) Stack stretches over the total width of
the pelagic South Atlantic (Fig. 1) at around
25-35°S, connects its four major basins (Argentine,
Brazil, Angola and Cape Basin) and crosses three
submarine ridges (Rio Grande Rise, Mid Atlantic
and Walvis Ridge). Via two deep water passages
in the western South Atlantic, Vema and Hunter
Channel, carbonate undersaturated Antarctic Bottom Water flows northward into the Brazil Basin,
from where only a minor part reaches the Angola
Basin through the Romanche Fracture Zone. Today, the slightly carbonate supersaturated North
Atlantic Deep Water (NADW) flows southward at
depths below 2000 m and controls the bottom water conditions in the Angola Basin. Therefore, the
carbonate lysocline is much deeper in the Angola
Basin (4700 to 4900 m) than in the Argentine, Bra-
603
quality hysteresis data from miniature 10-20 mg
samples (Flanders 1988; von Dobeneck 1996).
Typical data sampling densities for high-resolution rock magnetic core-logging are 0.5-2 measurements/cm, which is more than the effective spatial
resolution of most sensors. This over-samplingresuits in smooth, well-defmed and complete records
- an important prerequisite for the performance of
signal analysis techniques. In this paper we present
susceptibility records acquired with a 'Bartington
F type' spot sensor at a 1 cm spacing.
Orbital Tuning Strategies
Milankovitch's (1941) visionary idea of synchronizing geological records to orbital variations has
evolved into a widely accepted chronostratigraphic
method (Imbrie and Imbrie 1980; Imbrie et al. 1984;
Martinson et al. 1987). Applied to records of isotopic or lithologic characteristics, 'orbital tuning' can
yield unmatched precision and consistency in dating Quaternary sediment series. Ideally it takes just
three steps to align a proxy record to its orbital
pacemaker:
- establish a rough initial chronology and detect
spectral characteristics of Milankovitch cyclicity,
- choose an astronomic target curve and shift it by
postulated or empirically determined phase lags,
- filter the proxy signal in the appropriate frequency
band and match it with its model record.
In practice this approach bears numerous pitfalls. Rock magnetic records, like most proxy signals, depend more on regional sedimentation conditions than on global equilibria. They are not, in principle, safe bets for orbital tuning purposes and their
patterns can be anything from a 'SPECMAP (Imbrie et al. 1984) double' to 'random wiggles'. Cyclostratigraphic age models critically rely on the
initial and thus very consequential age model. Simple peak-to-peak correlation of proxy and target
records is highly susceptible to erroneous matches
(Shackleton et al. 1995), which have their origin in
ordinary stratigraphic problems such as undetected
hiati or turbidites, condensed or extended sections,
bioturbation, lithologic and diagenetic effects. Algorithms such as signal slotting (Thompson and
Clarke 1993) or complex demodulation (Pisias et
al. 1990) have been used to find the statistically
most probable proxy-target-correlation. Yet a combination of magneto- and biostratigraphic, eventually radiometric ages should be the most reliable
basis to start from. Specific regional settings and
scientific objectives may require adaptation of orbital tuning schemes into flexible, case-wise adapted strategies.
Two contrasting examples illustrate the scope
of possible approaches at different interpretation
levels. They are based on two series of
FS METEOR gravity cores collected within a longterm Quaternary South Atlantic research project
(SFB 261) at the University of Bremen. The coring sites are located in the subtropical and in the
western tropical South Atlantic. The two study areas differ greatly in size as well as in prevailing
sedimentation conditions and rates. The first example outlines the possibilities and limits of signal correlation and orbital tuning in low accumulation environments. Covering nearly the full Quaternary,
these susceptibility records are rewarding targets
for evolutionary spectral analysis. The particular
interest of the second example lies in a much higher
sedimentation and signal detail opening the possibility of analyzing periodic sub-Milankovitch signal
components. We will primarily focus at the methodical aspects ofthese magnetic time series analyses.
Subtropical South Atlantic
The Subtropical §.outh ~t1antic §.usceptibility
(SUSAS) Stack stretches over the total width of
the pelagic South Atlantic (Fig. 1) at around
25-35°S, connects its four major basins (Argentine,
Brazil, Angola and Cape Basin) and crosses three
submarine ridges (Rio Grande Rise, Mid Atlantic
and Walvis Ridge). Via two deep water passages
in the western South Atlantic, Vema and Hunter
Channel, carbonate undersaturated Antarctic Bottom Water flows northward into the Brazil Basin,
from where only a minor part reaches the Angola
Basin through the Romanche Fracture Zone. Today, the slightly carbonate supersaturated North
Atlantic Deep Water (NADW) flows southward at
depths below 2000 m and controls the bottom water conditions in the Angola Basin. Therefore, the
carbonate lysocline is much deeper in the Angola
Basin (4700 to 4900 m) than in the Argentine, Bra-