626
von Dobeneck and Schmieder
o 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380
250~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~wu~
(ij
200
b
~
150
"
100
:;i .().05
500 ~
¥~
~
~~ 0
450 ~r
~~
~
I i 0.05+-"""r-Ic,..,-~""""""""""'...,.J,...,..,~ ...... r..-1.""'''''''''"r'O;.,...,Lri''f'-.-lr~.-j-.''''''''''''...,...,.J,...'''T'''..,L,..,.......,..1,.-,..,...JI,-,-,rJ.-,.+400 3
,
0
20
40
60
80
100 120 140 160
180 200 220
240 260 280 300 320 340 360 380
...!i
Age [ka)
Fig. 16. (a) Susceptibility record of GeoB 1523-1 (arrows mark twin peaks corresponding to precession harmonics).
(b) idem, lowpass filtered (2 od order Butterworth) to exclude (> 18 kyr, dotted line) and include (> 7 kyr, solid line)
sub-Milankovitch harmonics. (c) The residue of both filtered signals (7 -18 kyr) frequently shows deep minima in
phase with insolation maxima (dotted vertical lines ), followed by twin maxima (shaded as in b) before and after
insolation minima. (d) The millennial band « 7 kyr) signal component appears to be aperiodic, although in some
sections, its envelope (shaded) resembles curve c (see also Fig. 16) (e) Precession index (solid line), shifted by
-1.77 kyr (dashed line), and 15°N July insolation curve (shaded) by Berger and Loutre (1991).
sion. Two further peak chains near 65 kyr and
30 kyr are identified as third-order intermodulations
«1141 + 1123-1119) kyr' and (1141-1123+ 1119) kyr')
within the Milankovitch band. Their important climatic implication will be discussed below.
In essence these results suggest that large parts
of the observed sub-Milankovitch signal components down to periods of at least 10 kyr can be
explained by non-linear climate response to orbital
variations. Below 10 kyr 'subtone dispersion'
comes into play (Fig. 14). Precession and obliquity
cycles are given as trigonometric expansions of
several sinusoidal signals, most ofthem with very
similar frequencies (Berger and Loutre 1992;
Schwarzacher 1993). The frequencies of their second and third order hannonics are still close enough
for their peaks to merge. With increasing order, the
differences become larger and each intennodulation peak splits up into a peak chain, eventually interfering with other peak chains .. . so that even
prominent and significant spectral peaks are no
more safely identified. Considering that for the
CEARIS cores the estimated age uncertainty is in
the order of 2 kyr, there is no sense in searching
for orbital effects much below periods of7 kyr. The
relative increase of the noise component with
higher frequency, resulting from climatic and depositional random processes and stratigraphic error,
is documented by the widening of the standard
deviation band in Fig. 14d.
In view of paleoceanographic interpretation,
signal phase and regularity are almost as important
as frequency and amplitude. A band selective time
domain representation ofthe susceptibility record
von Dobeneck and Schmieder
o 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380
250~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~wu~
(ij
200
b
~
150
"
100
:;i .().05
500 ~
¥~
~
~~ 0
450 ~r
~~
~
I i 0.05+-"""r-Ic,..,-~""""""""""'...,.J,...,..,~ ...... r..-1.""'''''''''"r'O;.,...,Lri''f'-.-lr~.-j-.''''''''''''...,...,.J,...'''T'''..,L,..,.......,..1,.-,..,...JI,-,-,rJ.-,.+400 3
,
0
20
40
60
80
100 120 140 160
180 200 220
240 260 280 300 320 340 360 380
...!i
Age [ka)
Fig. 16. (a) Susceptibility record of GeoB 1523-1 (arrows mark twin peaks corresponding to precession harmonics).
(b) idem, lowpass filtered (2 od order Butterworth) to exclude (> 18 kyr, dotted line) and include (> 7 kyr, solid line)
sub-Milankovitch harmonics. (c) The residue of both filtered signals (7 -18 kyr) frequently shows deep minima in
phase with insolation maxima (dotted vertical lines ), followed by twin maxima (shaded as in b) before and after
insolation minima. (d) The millennial band « 7 kyr) signal component appears to be aperiodic, although in some
sections, its envelope (shaded) resembles curve c (see also Fig. 16) (e) Precession index (solid line), shifted by
-1.77 kyr (dashed line), and 15°N July insolation curve (shaded) by Berger and Loutre (1991).
sion. Two further peak chains near 65 kyr and
30 kyr are identified as third-order intermodulations
«1141 + 1123-1119) kyr' and (1141-1123+ 1119) kyr')
within the Milankovitch band. Their important climatic implication will be discussed below.
In essence these results suggest that large parts
of the observed sub-Milankovitch signal components down to periods of at least 10 kyr can be
explained by non-linear climate response to orbital
variations. Below 10 kyr 'subtone dispersion'
comes into play (Fig. 14). Precession and obliquity
cycles are given as trigonometric expansions of
several sinusoidal signals, most ofthem with very
similar frequencies (Berger and Loutre 1992;
Schwarzacher 1993). The frequencies of their second and third order hannonics are still close enough
for their peaks to merge. With increasing order, the
differences become larger and each intennodulation peak splits up into a peak chain, eventually interfering with other peak chains .. . so that even
prominent and significant spectral peaks are no
more safely identified. Considering that for the
CEARIS cores the estimated age uncertainty is in
the order of 2 kyr, there is no sense in searching
for orbital effects much below periods of7 kyr. The
relative increase of the noise component with
higher frequency, resulting from climatic and depositional random processes and stratigraphic error,
is documented by the widening of the standard
deviation band in Fig. 14d.
In view of paleoceanographic interpretation,
signal phase and regularity are almost as important
as frequency and amplitude. A band selective time
domain representation ofthe susceptibility record