Trends in cloud cover are not well established prior to the early 1950s [ Warren et al., 1986, 1988], but it seems clear that cloud cover has changed during the twentieth century in at least parts of North America [ Plantico et al., 1990]. If the imperfect records from all regions of the United States are averaged together, deserts with rain forests, average summer cloud cover has increased by as much as one tenth. Comparable changes may have occurred in other parts of the world [ Henderson-Sellers, 1986a, b, 1989; P. A. Jones and Henderson-Sellers, 1992]. The cause of the increase in cloud cover is not established, although it could plausibly result from the effects of pollution and sulfate aerosol on concentrations of cloud condensation nuclei [ Bolin and Charlson, 1976; Twomey, 1977; Charlson et al., 1987, 1990; Kim and Cess, 1993]. This increased cloud cover can be expected to have caused significant radiative cooling, at least regionally. It may also have caused the observed reduction in the amplitude of diurnal temperature variations [ Karl et al., 1993]. Is there evidence of the effect of changing clouds in the historical record of global average temperature?
Figure 3 compares the history of global average temperature
[ P. D. Jones et al., 1986, 1991; compare Hansen and Lebedeff,
1987, 1988; Folland et al., 1990; Vinnikov et al., 1990;
Boden et al., 1991, p.
509-529] with the history of carbon dioxide concentration for the
last century and a half [ Neftel et al., 1985; Friedli et al., 1986;
Keeling et al., 1989; Boden et al., 1991, p. 8-15]. The reliability of
the temperature record and the consistency of different sources of data
have been discussed in the cited publications. There is little room
for doubt that temperature has increased by about 0.5
C during this
time. The increase has been erratic, appearing as periods of
nearly constant temperature separated by periods of rapid increase.
Carbon dioxide, in contrast, has increased smoothly, having the
appearance of exponential growth. The qualitative difference between
these two histories constitutes a public relations problem for those of
us concerned about the greenhouse effect and climate change. We can
all think of many reasons for temperature to respond unevenly to a
gradual increase in greenhouse forcing, but to the skeptical,
these explanations may sound like special pleading. Much work will
be needed to explain observed climate change, including documentation of
the histories of all of the climate forcing terms and the use of
these forcing histories in a hierarchy of climate models to
successfully simulate the spatial and temporal distribution of
observed temperature change. I show now that the record suggests,
but certainly does not prove, that industrial aerosols may have
influenced global average temperature.
Figure 4 compares the smoothed record of global average temperature with the global rate of production of carbon dioxide by combustion of fossil fuel [ Keeling, 1973; Marland and Rotty, 1984; Boden et al., 1991, p. 379-389]. The fossil fuel rate is plotted on a logarithmic scale to reveal changes in the annual percentage rate of increase. The figure shows that the fossil fuel rate increased at a nearly constant percentage rate from the beginning of the record until the First World War. During this period, global average temperature showed no increase. There followed a period of slow growth in the rate of fossil fuel combustion extending through the Great Depression to the end of the Second World War. Temperature increased rapidly during this period. Renewed rapid growth in the rate of combustion of fossil fuel characterized the period from the mid forties to the mid seventies, when global average temperature again showed little persistent change. Since the oil price increases of the seventies, growth in the fossil fuel rate has been slow, and the increase of global average temperature has been rapid.
This behavior, at first sight surprising, is entirely consistent with a competition between warming by greenhouse gases and cooling by haze and aerosol-induced changes in cloud reflectivity [ Wigley, 1989; Charlson et al., 1991, 1992; Penner et al., 1992]. The atmospheric residence time of carbon dioxide, the principal greenhouse gas produced by human activities, is long compared with the time scale of industrial change [ Watson et al., 1990]. The carbon dioxide concentration and greenhouse forcing are approximately proportional to the cumulative production of carbon dioxide, the fossil fuel rate integrated over time. So greenhouse forcing has increased smoothly and steadily as carbon dioxide has accumulated in the atmosphere. But the average aerosol remains aloft for weeks or less. The aerosol cooling is approximately proportional to the current rate of production, not the cumulative rate. The rate of fossil fuel burning may be taken as an approximate index of the rate of aerosol production because it is an indicator of overall economic and industrial activity and because much of the aerosol originates in smokestacks and tailpipes.
The integral of an exponential is an exponential, so aerosol
cooling can balance greenhouse warming when the fossil fuel rate
is increasing exponentially, but a constant fuel rate results in
constant cooling while carbon dioxide continues to accumulate in
the atmosphere and greenhouse warming continues to increase [ Charlson
et al., 1991; Wigley, 1991]. Figure 5 illustrates how well the
main features of the history of global average temperature are reproduced
by a theory that simply equates temperature increase to the
difference between cumulative carbon dioxide release and a term
proportional to the current rate of release. In constructing this
figure, greenhouse warming is assumed to be proportional to the
cumulative fossil fuel consumption, with a 1990 value of 2.4
W/m
. Aerosol forcing, which includes clear air and
cloud-aerosol effects, is assumed to be linear in the sulfate release
rate, with a value in 1990 equal to -1.3 W/m
, a value within the
range of published estimates. The climate sensitivity assumed to
fit calculation to observation is 1.27 deg/(W/m
), a value at the
upper limit of published estimates [ Cubasch and Cess, 1990;
Hansen et al., 1984].
Climate scientists working on global change on the industrial time scale are well aware of the potential importance of clouds and are working observationally and theoretically to reduce the uncertainties that cloud processes introduce into our ``predictions'' [ Coakley and Cess, 1985; Coakley et al., 1987; Slingo, 1990; Turco et al., 1990; Falkowski et al., 1992; Langner et al., 1992; Engardt and Rodhe, 1993; Kiehl and Briegleb, 1993]. There is still disagreement about whether global cloud feedback is positive or negative in the case of greenhouse warming. The speculative suggestions I make in this paper are neither validated nor invalidated by current climate models. My main purpose here is to stimulate cloud-consciousness among paleoclimatologists, so I turn now to two illustrations of the possible importance of clouds on longer time scales.