616
von Dobeneck and Schmieder
Extended Time Series Analyses
Having tuned a rock magnetic record to basic astronomical cycles in the Milankovitch range gives
access to time series analyses into the adjacent
frequency bands. These are essential to investigate
the temporal evolution and physical principles of
proxy response to orbital forcing. Suitable methods
to analyze super- (> 100 kyr) and sub-Milankovitch
« 18 kyr) signal variations with and without the
viewpoint of orbital forcing are presented in the
following.
Evolutionary spectral analysis requires long time
series and is therefore demonstrated for the SUSAS
stack covering the last 1.5 m.y. The high temporal
resolution of the CEARIS stack makes it a suitable
exercise for spectral analyses in the sub-Milankovitch frequency range.
Super-Milankovitch signal variations of the
SUSAS stack
Linear orbital forcing through insolation variations
can only account for proxy oscillations in the precession and obliquity band (Imbrie et al. 1992). The
96 kyr, 127 kyr and 404 kyr eccentricity cycles
modulate the precession amplitude, but produce
negligible insolation changes in their own frequency
range (Fig. 12a). Yet, the SUSAS stack like other
Pleistocene climate records exhibits strong 100 kyr
variance (Fig. 12c).
Many attempts have been made to explain this
'100 kyr cycle problem', frequently invoking nonlinear processes channeling energy into the 100 kyr
band (for a summary see Imbrie et aI., 1993). In
rejecting the assumption that variations in insolation alone are responsible for the observed climatic
changes, Muller and MacDonald (1995) explain the
100 kyr cycle by inclination changes of the Earth's
orbital plane relative to the symmetry plane ofthe
solar system. Liu (1992; 1995) concludes that frequency variations of the obliquity cycle can give rise
to strong 100 kyr forcing of climate. Like various
others, Raymo (1997) favors the idea that eccentricity modulation of the precession index amplitude
generates quasi-periodic 100 kyr ice age cycles.
The climate changes from ice shield growth to
decay, when summer insolation exceeds a certain
threshold value. A long-term reduction of the atmospheric pC0 2 level (Saltzman and Verbitsky
1993), eventually caused by tectonic processes
(Raymo et al. 1988; Raymo and Ruddiman 1992),
is thought to have induced a steady rise ofthe insolation threshold for large-scale melting in Quaternary times. This may explain, why late Pliocene to
early Pleistocene paleoclimate records exhibit
mainly obliquity and precession related variance
(e.g., Raymo et al. 1989; Bloemendal and
deMenocal 1989; deMenocal 1995), while late
Pleistocene climate fluctuations, in spite of similar
insolation variations, are dominated by 100 kyr
cycles (e.g., Hays et al. 1976; Ruddiman et al.
1989). The exact timing ofthis mid-Pleistocene climate transition and the question whether it was a
gradual ('mid-Pleistocene evolution' , Ruddiman et
al. 1989) or a sudden change (,mid-Pleistocene revolution', Berger et al. 1994) have been subject of
many studies (Shackleton and Opdyke 1976; Pisias
and Moore 1981; Prell 1982; Ruddiman et al. 1989;
Berger and Jansen 1994b; Mudelsee and
Stattegger 1997) and attempts to model or characterize the transition with different statistical techniques (e.g., DeBlonde and Peltier 1991; Park and
Maasch 1993; Mudelsee and Schulz 1997).
Fig. 11. Temporal analysis of varying frequency content of the SUSAS stack. From bottom to top: Lowpass filtering of the normalized susceptibility stack reveals three distinct periods with different base levels and transitions
centered at around 0.95 and 0.6 Ma. Extracting the basic Milankovitch cycles by bandpass filtering depicts frequency shifts within the eccentricity band and fading of the obliquity related variance in early Pleistocene.
Precessional amplitudes are generally low due to bioturbation damping, particularly in periods oflower sedimentation rates. The envelopes were derived by Hilbert transformation (Bendat and Piersol 1986). For the evolutionary
spectral analysis (see text) 500 kyr sections of the highpass filtered SUSAS signal (purple) progressing in 10 kyr
steps are tapered and transformed into the frequency domain. Related to the center of each section, the resulting
spectra are color-coded and form a so-called 'spectrogram'. The colored horizontal bars represent the limits used
in the bandpass analyses below.
von Dobeneck and Schmieder
Extended Time Series Analyses
Having tuned a rock magnetic record to basic astronomical cycles in the Milankovitch range gives
access to time series analyses into the adjacent
frequency bands. These are essential to investigate
the temporal evolution and physical principles of
proxy response to orbital forcing. Suitable methods
to analyze super- (> 100 kyr) and sub-Milankovitch
« 18 kyr) signal variations with and without the
viewpoint of orbital forcing are presented in the
following.
Evolutionary spectral analysis requires long time
series and is therefore demonstrated for the SUSAS
stack covering the last 1.5 m.y. The high temporal
resolution of the CEARIS stack makes it a suitable
exercise for spectral analyses in the sub-Milankovitch frequency range.
Super-Milankovitch signal variations of the
SUSAS stack
Linear orbital forcing through insolation variations
can only account for proxy oscillations in the precession and obliquity band (Imbrie et al. 1992). The
96 kyr, 127 kyr and 404 kyr eccentricity cycles
modulate the precession amplitude, but produce
negligible insolation changes in their own frequency
range (Fig. 12a). Yet, the SUSAS stack like other
Pleistocene climate records exhibits strong 100 kyr
variance (Fig. 12c).
Many attempts have been made to explain this
'100 kyr cycle problem', frequently invoking nonlinear processes channeling energy into the 100 kyr
band (for a summary see Imbrie et aI., 1993). In
rejecting the assumption that variations in insolation alone are responsible for the observed climatic
changes, Muller and MacDonald (1995) explain the
100 kyr cycle by inclination changes of the Earth's
orbital plane relative to the symmetry plane ofthe
solar system. Liu (1992; 1995) concludes that frequency variations of the obliquity cycle can give rise
to strong 100 kyr forcing of climate. Like various
others, Raymo (1997) favors the idea that eccentricity modulation of the precession index amplitude
generates quasi-periodic 100 kyr ice age cycles.
The climate changes from ice shield growth to
decay, when summer insolation exceeds a certain
threshold value. A long-term reduction of the atmospheric pC0 2 level (Saltzman and Verbitsky
1993), eventually caused by tectonic processes
(Raymo et al. 1988; Raymo and Ruddiman 1992),
is thought to have induced a steady rise ofthe insolation threshold for large-scale melting in Quaternary times. This may explain, why late Pliocene to
early Pleistocene paleoclimate records exhibit
mainly obliquity and precession related variance
(e.g., Raymo et al. 1989; Bloemendal and
deMenocal 1989; deMenocal 1995), while late
Pleistocene climate fluctuations, in spite of similar
insolation variations, are dominated by 100 kyr
cycles (e.g., Hays et al. 1976; Ruddiman et al.
1989). The exact timing ofthis mid-Pleistocene climate transition and the question whether it was a
gradual ('mid-Pleistocene evolution' , Ruddiman et
al. 1989) or a sudden change (,mid-Pleistocene revolution', Berger et al. 1994) have been subject of
many studies (Shackleton and Opdyke 1976; Pisias
and Moore 1981; Prell 1982; Ruddiman et al. 1989;
Berger and Jansen 1994b; Mudelsee and
Stattegger 1997) and attempts to model or characterize the transition with different statistical techniques (e.g., DeBlonde and Peltier 1991; Park and
Maasch 1993; Mudelsee and Schulz 1997).
Fig. 11. Temporal analysis of varying frequency content of the SUSAS stack. From bottom to top: Lowpass filtering of the normalized susceptibility stack reveals three distinct periods with different base levels and transitions
centered at around 0.95 and 0.6 Ma. Extracting the basic Milankovitch cycles by bandpass filtering depicts frequency shifts within the eccentricity band and fading of the obliquity related variance in early Pleistocene.
Precessional amplitudes are generally low due to bioturbation damping, particularly in periods oflower sedimentation rates. The envelopes were derived by Hilbert transformation (Bendat and Piersol 1986). For the evolutionary
spectral analysis (see text) 500 kyr sections of the highpass filtered SUSAS signal (purple) progressing in 10 kyr
steps are tapered and transformed into the frequency domain. Related to the center of each section, the resulting
spectra are color-coded and form a so-called 'spectrogram'. The colored horizontal bars represent the limits used
in the bandpass analyses below.