The one-dimensinal radiative-convective equillibrium model with an explicit hydrologic cycle introduced in part 1 is used to study the senstivity of the model's atmosphere to large changes in the solar forcing, when various cumulus convection parameterizations are used. As shown by Simpson <1927>, by Komabayasi <1967>, and by Ingersoll <1969> when the concentration of the absorbing gas in the atmosphere is temperature dependent, equilibrium is impossible for values of the solar forcing larger than a critical value. This result is referred to as a runaway greenhouse. The cumulus convection parameterization schemes currently in use in global climate models (GCMs) employ different assumptions about moistening. This causes the critical solar forcing above which a runaway greenhouse occurs to be very sensitive to the cumulus convection scheme employed. Furthermore, using the microphysically based cumulus convection scheme proposed by Emanuel <1991>, we show that the sensitivity of the equilibrium temperature to changes in the solar forcing depends crucially on the microphysics of cumulus convection.

For fixed cloud conditions, the critical solar forcing for a runaway greenhouse to occur is between approximately 1.22 and 1.498 times the global mean value for the Earth, and for clear sky conditions, it is a few percent lower. The runaway greenhouse in the experiments with the mass flux schemes generally occurs more rapidly than in the experiments with the adjustment schemes. In addition, the inability of the hard convectie adjustment scheme to produce an efficient vertical transport of moisture, together with the saturation requirement for convection to occur, leads to the breakdown of the radiative-convection equilibria when other processes are not available to provide the necessary vertical transport of water vapor. ¿American Geophysical Union 1994