Based on data obtained from the Tropical Ocean Global Atmosphere - Coupled Ocean-Atmosphere Response Experiment, Sud et al. (1999) demonstrated that deep convection in the tropics acts as a thermostat to keep sea surface temperature (SST) vacillating between approximately 28 and 30°C.  Their analysis suggests that as SSTs reach 28-29°C, the cloud-base airmass is charged with the moist static energy needed for clouds to reach the upper troposphere, at which point the cloud cover reduces the amount of solar radiation received at the surface of the sea, while cool and dry downdrafts promote ocean surface cooling by increasing sensible and latent heat fluxes there.  This "thermostat-like control," as Sud et al. describe it, tends "to ventilate the tropical ocean efficiently and help contain the SST between 28-30°C."  The phenomenon would also be expected to prevent SSTs from rising any higher in response to enhanced CO2-induced radiative forcing.

Lindzen et al. (2001) used upper-level cloudiness data obtained from the Japanese Geostationary Meteorological Satellite and SST data obtained from the National Centers for Environmental Prediction to derive a strong inverse relationship between upper-level cloud area and the mean SST of cloudy regions of the eastern part of the western Pacific (30°S-30°N; 130°E-170°W), such that the area of cirrus cloud coverage normalized by a measure of the area of cumulus coverage decreases about 22% per degree C increase in the SST of the cloudy region.  In describing this phenomenon, Lindzen et al. say "the cloudy-moist region appears to act as an infrared adaptive iris that opens up and closes down the regions free of upper-level clouds, which more effectively permit infrared cooling, in such a manner as to resist changes in tropical surface temperature."  More recently, however, Hartmann and Michelsen (2002) have produced evidence suggesting that the correlation derived by Lindzen et al. results from variations in subtropical clouds that are not physically connected to deep convection near the equator, stating that it is thus "unreasonable to interpret these changes as evidence that deep tropical convective anvils contract in response to SST increases."  As a result, this particular feedback mechanism remains controversial.

In a more straightforward study, Croke et al. (1999) used land-based observations of cloud cover for three regions of the United States (coastal southwest, coastal northeast, and southern plains) to demonstrate that, over the period 1900-1987, cloud cover had a high correlation with global air temperature, with mean cloud cover rising from an initial value of 35% to a final value of 47% as the mean global air temperature rose by 0.5°C.  This phenomenon would again tend to counteract the effects of any impetus for warming.

In another study, Herman et al. (2001) used Total Ozone Mapping Spectrometer 380-nm reflectivity data to determine changes in radiation reflected back to space over the period 1979 to 1992, finding that "when the 11.3-year solar-cycle and ENSO effects are removed from the time series, the zonally averaged annual linear-fit trends show that there have been increases in reflectivity (cloudiness) poleward of 40°N and 30°S, with some smaller but significant changes occurring in the equatorial and lower middle latitudes."  The overall long-term effect was an increase in radiation reflected back to space of 2.8 Wm^-2 per decade, which represents a large cloud-induced cooling influence.

Last of all, Rosenfeld (2000) used satellite data obtained from the Tropical Rainfall Measuring Mission to look for terrestrial analogues of the cloud trails that form in the wakes of ships at sea as a consequence of their emissions of particulates that redistribute cloud-water into larger numbers of smaller droplets that do not rain out of the atmosphere as readily as they would in the absence of this phenomenon. Visualizations produced from the mission data clearly revealed the existence of enhanced cloud trails downwind of urban and industrial complexes in Turkey, Canada and Australia, to which Rosenfeld gave the name pollution tracks in view of their similarity to ship tracks.  Rosenfeld also demonstrated that the clouds comprising these pollution tracks were composed of droplets of reduced size that did indeed suppress precipitation by inhibiting further coalescence and ice precipitation formation.  As Toon (2000) noted in a commentary on this study, these smaller droplets will not "rain out" as quickly and will therefore last longer and cover more of the earth, both of which effects tend to cool the globe.

In summation, as the earth warms, the atmosphere has a tendency to become more cloudy and exert a natural brake upon the rising temperature.  Also, many of man's aerosol-producing activities tend to do the same thing.  Hence, there appear to be a number of different ways in which cloud-mediated processes help the planet to "keep its cool," relatively speaking.

References
Croke, M.S., Cess, R.D. and Hameed, S.  1999.  Regional cloud cover change associated with global climate change: Case studies for three regions of the United States.  Journal of Climate 12: 2128-2134.

Hartmann, D.L. and Michelsen, M.L.  2002.  No evidence for IRIS.  Bulletin of the American Meteorological Society 83: 249-254.

Herman, J.R., Larko, D., Celarier, E. and Ziemke, J.  2001.  Changes in the Earth's UV reflectivity from the surface, clouds, and aerosols.  Journal of Geophysical Research 106: 5353-5368.

Lindzen, R.S., Chou, M.-D. and Hou, A.Y.  2001.  Does the earth have an adaptive infrared iris?  Bulletin of the American Meteorological Society 82: 417-432.

Rosenfeld, D.  2000.  Suppression of rain and snow by urban and industrial air pollution.  Science 287: 1793-1796.

Sud, Y.C., Walker, G.K. and Lau, K.-M.  1999.  Mechanisms regulating sea-surface temperatures and deep convection in the tropics.  Geophysical Research Letters 26: 1019-1022.

Toon, O.W.  2000.  How pollution suppresses rain.  Science 287: 1763-1765.