It is suggested that the high albedo of a desert contributes to a net radiative heat loss relative to its surroundings and that the resultant horizontal temperature gradients induce a frictionally controlled circulation which imports heat aloft and maintains thermal equilibrium through sinking motion and adiabatic compression. In the subtropics this sinking motion is superimposed on the descending branch of the mean Hadley circulation but is more intense. Thus the desert feeds back upon itself in an important manner.

If one takes into account the biosphere, this feedback mechanism could conceivably lead to instabilities or metastabilities in desert border regions. It is argued that a reduction of vegetation, with consequent increase in albedo, in the Sahel region at the southern margin of the Sahara would cause sinking motion, additional drying, and would therefore perpetuate the arid conditions. Numerical integrations with the general circulation model of NASA's Goddard Institute for Space Studies appear to substantiate this hypothesis. Increasing the albedo north of the ITCZ from 14% to 35% had the effect of shifting the ITCZ several degrees of latitude south and decreasing the rainfall in the Sahel about 40% during the rainy season.