Studies of Greenland ice cores have revealed a clear and strong correlation between the dust content of the ice and temperatures deduced from oxygen or hydrogen isotopes [ Dansgaard et al., 1993; Greenland Ice-core Project (GRIP) Members, 1993; Mayewski et al., 1993, 1994]. Cold periods are dusty. The correlation is present on time scales ranging from less than a century to the approximately 100,000 year periodicity of glacial advance and retreat. The Vostok ice core from Antarctica shows the same correlation at the longer time scales that can be resolved in this core [ Lorius et al., 1990; Jouzel et al., 1993]. Sea floor cores from lower latitudes show that the flux of airborne dust to the oceans also correlates with ice ages [ Clemens and Prell, 1990; Rea, 1994]. Dust fluxes to the oceans were large when the high latitudes were cold. The correlation appears to extend throughout the globe. The troposphere was relatively dusty during glaciations.
Perhaps there was enough dust in the atmosphere during glaciations to increase the concentration of cloud condensation nuclei. Dust particles might have served directly as condensation nuclei or might have served as nuclei for the growth of sulfate particles [ Hudson, 1993]. More cloud condensation nuclei would cause cloud droplets to be smaller and more numerous, other things being equal, increasing cloud albedo and cooling the climate [ Ackerman et al., 1994]. Meanwhile, smaller cloud drops settle more slowly, yielding less precipitation and less washout of dust. Increased aridity might result, destroying vegetation cover and promoting the release of dust. A dust, cloud, climate, vegetation feedback might have caused the ice ages by amplifying the small forcing attributable to orbital perturbations and changes in greenhouse gas concentrations.
Apart from global cooling, increased cloud cover
[4]
would serve to
promote the accumulation and advance of ice. Ice accumulates when
winter accumulation of snow exceeds summer melting. Because warm air
can hold more water vapor, glacial growth is favored by warm, but
freezing winters and cool summers. As Figure 1 shows, cloud forcing at
high latitudes is positive in winter and negative in summer. More
clouds should therefore promote glacial advance. Because cloud cover
is already large over the high latitude oceans it cannot increase very
much during an ice age, but the cloud forcing is so large that a
small increase may suffice.
The feedback I have proposed is speculative. It is a large leap from dust in ice cores to cloud-condensation nuclei to the distribution of clouds in latitude and altitude, but the seasonal and latitudinal dependence of cloud forcing is observed, not assumed, and the correlation of high dust concentrations and cold climates is observed also. Paleoclimatologists should consider the causes and consequences of these high dust concentrations [ Rosenzweig and Hillel, 1993].