184 Encyclopedia of the Solar System
FIGURE 4 CO 2 concentrations (top) and temperature variations
(bottom) over the last 420,000 years as obtained from ice cores at
Vostok, Antarctica (data from Petit et al., 1999). The approximate
100,000-year period of the ice ages is evident, although many
shorter period fluctuations are superimposed within the record.
Prominent ice-age terminations occurred at∼410, 320, 240–220,
130, and 15 ka in the past. Also note the correlation between
temperature and CO 2 concentration during these cycles, which
shows the influence of changes in the greenhouse effect on ice
ages. The vertical line at the right side of the top plot shows the
increase in CO 2 caused by humans between∼1800 and 2006.
between glacial and interglacial periods, causing migration
of coastlines by hundreds of kilometers in some regions.
The time history of temperature, ice volume, and other
variables can be studied using stable isotopes of carbon,
hydrogen, and oxygen as recorded in glacial ice, deep-sea
sediments, and land-based records such as cave calcite and
organic material. This record shows that glacial/interglacial
cycles over the past 800,000 years have a predominant pe-
riod of∼100,000 years (Fig. 4). During this cycle, glaciers
gradually increase in volume (and air temperature gradu-
ally decreases) over most of the 100,000 year period; the
glaciers then melt, and the temperature increases over a
relatively short∼5000 year interval. The cycle is thus ex-
tremely asymmetric and resembles a saw-tooth curve rather
than a sinusoid. The last ice age peaked 18,000 years ago
and ended by 10,000 years ago; the modern climate cor-
responds to an interglacial period. Analysis of ancient air
trapped in air bubbles inside the Antarctic and Greenland
ice sheets shows that the atmospheric CO 2 concentration
is low during ice ages—typically about 200 ppm—and rises
to∼280 ppm during the intervening interglacial periods
(Fig. 4).
Ice ages seem to result from changes in the strength of
sunlight caused by periodic variations in Earth’s orbit, mag-
nified by several of the feedbacks discussed in Section 5.2.
A power spectrum of the time series in Fig. 4 shows that
temperature, ice volume, and CO 2 vary predominantly on
periods of 100, 41, 23, and 19 thousand years (ka; the sum-
mation of sinusoids at each of these periods leads to the saw-
tooth patterns in Fig. 4). Interestingly, these periods match
the periods over which northern hemisphere sunlight varies
due to orbital oscillations. The Earth’s orbital eccentricity
oscillates on periods of 100 ka, the orbital obliquity (the
tilt of Earth’s rotation axis) oscillates on a period of 41 ka,
and the Earth’s rotation axis precesses on periods of 19 and
23 ka. These variables affect the difference in sunlight re-
ceived at Earth between winter and summer and between
the equator and pole. In turn, these sunlight variations de-
termine the extent to which snowpack accumulates in high
northern latitudes during winter, and the extent to which
this snowpack resists melting during summer; glaciers build
up when snow that falls during winter cannot melt the fol-
lowing summer. The idea that these orbital variations cause
ice ages has become known as the Milankovitch theory of
ice ages.
By themselves, however, orbital variations are only part
of the story. Sunlight variations due to the 100 ka eccentric-
ity variations are much weaker than sunlight variations due
to the 41, 23, and 19 ka obliquity and precession variations.
Thus, if the orbit-induced sunlight variations translated di-
rectly into temperature and ice variations, ice ages would
be dominated by the 41, 23, and 19 ka periods, but instead,
the 100 ka period dominates (as can be seen in Fig. 4). This
means that some nonlinearity in the climate system ampli-
fies the climatic response at 100 ka much better than at the
shorter periods. Furthermore, the observed oscillations in
CO 2 between glacial and interglacial periods (Fig. 4) in-
dicates that ice ages are able to occur partly because the
greenhouse effect is weak during ice ages but strong during
interglacial periods. Most likely, atmospheric CO 2 becomes
dissolved in ocean water during ice ages, allowing the atmo-
spheric CO 2 levels to decrease; the ocean then rejects this
CO 2 at the end of the ice age, increasing its atmospheric
concentration. Recent analyses of Antarctic ice cores show
that, at the end of an ice age, temperature rise precedes
CO 2 rise in Antarctica by about 800 years, indicating that
CO 2 variation is an amplifier rather than a trigger of ice-age
termination. Interestingly, however, both of these events
precede the initiation of deglaciation in the northern hemi-
sphere. These observations suggest that the end of an ice
age is first triggered by a warming event in the Antarctic
region; this initiates the process of CO 2 rejection from the
oceans to the atmosphere, and the resulting increase in the
greenhouse effect, which is global, then allows deglaciation
to commence across the rest of the planet. The ice-
albedo and water-vapor feedbacks (Section 5.2) help am-
plify the transition. However, many details, including the