Climate Cycles and Climate Transitions . ..
241
Another important feedback mechanism is basal sliding, which can occur in
a one-dimensional ice sheet-bedrock model if thermodynamics is included.
Oerlemans (1982) has shown that such a model can produce free oscillations
with a period of about 100 ka, again for a time constant of E21 = 5 ka.
The case 1 results with a long time constant for bedrock depression
E21 = 30 ka are in good agreement with similar experiments of Pollard (1983,
Figs. 2b, 5) and Deblonde and Peltier (1991, Figs. 2a, c) on the one hand and
the SPECMAP 8 18 0 record on the other hand.
The positive elevation-temperature feedback included in case 2 does not
result in an improved simulation of the observed 8 18 0 record, while in all cases
the ice-calving mechanism does. This means that it is the design of the negative feedbacks that determines how good the fit to the data is, one criterion for
a good fit being how well the rapid deglaciations are simulated.
Case 3 shows that inclination forcing can help to achieve nearly complete
terminations, although the ice-calving mechanism is still required to melt all
ice. If the ice-calving mechanism is invoked, the timing of the terminations is
less satisfactory with inclination forcing than without it.
We think that the use of raw spectra as advocated by Muller and MacDonald
is problematic in comparing observed 8 18 0 data and simulated ice mass. This
is because of the ever-present noise and the quasi-periodic nature of the insolation forcing as well as the inclination forcing. A high standard error in the
amplitudes of the raw spectra does not allow a discussion on whether a peak
near a period of about 100 ka is split or not. From the smoothed spectra of
cases 1 and 3, we conclude that the widths of the peaks for the 8 18 0 data and
the simulated ice mass are comparable, and that there is no excessive power in
the 400-ka band, neither for the forced nor for the relaxation oscillator. However, we admit that a longer time series must be considered to properly assess a
peak near a period of 400 ka.
In both parameterizations for the effect of changes in orbital inclination on
temperature, we assume that times of high accretion are coincident with times
oflow temperature, at variance with Farley (1995), who found a 100-ka cycle of
3He in sediment, but times of high accretion roughly coincident with the interglacials. Marcantonio et al. (1996) even dispute that the 100-ka cycle in IDP
accretion is the cause of the observed helium-3 variability.
The second question that arises is what has actually caused the climate transitions in the late Cenozoic. On the one hand, Muller (1994) suggests that the
sudden onset of the 100-ka cycle might have been caused by an increase in the
amount of IDP or meteoroids at that time. On the other hand, the growth of
large and unstable continental ice sheets is attributed to a slow decrease in the
atmospheric CO2 level, induced by tectonic processes (e.g., Saltzman and
Maasch 1991; Saltzman and Verbitsky 1993; Raymo 1997). Both mechanisms
are still speculative. The full PCM that we use in case 2 is built on the scond
mechanism.
The transitions that are seen in Fig. 13b can be readily associated with the
climate transitions in the late Pliocene and the mid-Pleistocene. The first occurs
241
Another important feedback mechanism is basal sliding, which can occur in
a one-dimensional ice sheet-bedrock model if thermodynamics is included.
Oerlemans (1982) has shown that such a model can produce free oscillations
with a period of about 100 ka, again for a time constant of E21 = 5 ka.
The case 1 results with a long time constant for bedrock depression
E21 = 30 ka are in good agreement with similar experiments of Pollard (1983,
Figs. 2b, 5) and Deblonde and Peltier (1991, Figs. 2a, c) on the one hand and
the SPECMAP 8 18 0 record on the other hand.
The positive elevation-temperature feedback included in case 2 does not
result in an improved simulation of the observed 8 18 0 record, while in all cases
the ice-calving mechanism does. This means that it is the design of the negative feedbacks that determines how good the fit to the data is, one criterion for
a good fit being how well the rapid deglaciations are simulated.
Case 3 shows that inclination forcing can help to achieve nearly complete
terminations, although the ice-calving mechanism is still required to melt all
ice. If the ice-calving mechanism is invoked, the timing of the terminations is
less satisfactory with inclination forcing than without it.
We think that the use of raw spectra as advocated by Muller and MacDonald
is problematic in comparing observed 8 18 0 data and simulated ice mass. This
is because of the ever-present noise and the quasi-periodic nature of the insolation forcing as well as the inclination forcing. A high standard error in the
amplitudes of the raw spectra does not allow a discussion on whether a peak
near a period of about 100 ka is split or not. From the smoothed spectra of
cases 1 and 3, we conclude that the widths of the peaks for the 8 18 0 data and
the simulated ice mass are comparable, and that there is no excessive power in
the 400-ka band, neither for the forced nor for the relaxation oscillator. However, we admit that a longer time series must be considered to properly assess a
peak near a period of 400 ka.
In both parameterizations for the effect of changes in orbital inclination on
temperature, we assume that times of high accretion are coincident with times
oflow temperature, at variance with Farley (1995), who found a 100-ka cycle of
3He in sediment, but times of high accretion roughly coincident with the interglacials. Marcantonio et al. (1996) even dispute that the 100-ka cycle in IDP
accretion is the cause of the observed helium-3 variability.
The second question that arises is what has actually caused the climate transitions in the late Cenozoic. On the one hand, Muller (1994) suggests that the
sudden onset of the 100-ka cycle might have been caused by an increase in the
amount of IDP or meteoroids at that time. On the other hand, the growth of
large and unstable continental ice sheets is attributed to a slow decrease in the
atmospheric CO2 level, induced by tectonic processes (e.g., Saltzman and
Maasch 1991; Saltzman and Verbitsky 1993; Raymo 1997). Both mechanisms
are still speculative. The full PCM that we use in case 2 is built on the scond
mechanism.
The transitions that are seen in Fig. 13b can be readily associated with the
climate transitions in the late Pliocene and the mid-Pleistocene. The first occurs