The feasibility of reductions of tropical tropospheric water vapor as a cause of global cooling, such as occured during Quaternary glacial periods, is examined. A simple coupled atmosphere-ocean model is used which divides the tropics into a region of persistent deep convection and a subtropical region with no deep convection. The regions are coupled via a radiatively driven Hadley Cell and a wind-driven meridional overturning cell in the ocean. The model includes reasonably detailed treatments of radiation and the convective boundary layer (CBL).
The amount of tropical cooling depends on the height of the tropospheric drying and on the extent to which cloud water in the CBL is converted into rainwater. If the CBL clouds do not precipitate then the SST is about equally sensitive to drying at the tropopause, immediately above the inversion and within the CBL. This results differs from that expected on the basis of radiation physics alone which suggests drying above the CBL would be most effective. This is because changes in CBL depth amplify the impact of drying at the tropopause but oppose the impact of drying immediately above the CBL.
When the CBL clouds are allowed to precipitate, variations in CBL depth are small, and the tropical SST becomes most sensitive to drying immediately above the CBL. Reducing the relative humidity of the entire troposphere above the subcloud layer by about 10-20% cools the tropical SST by close to 3K. However, changes in low level cloud fraction, appear capable of acting as a negative feedback that reduce this cooling to about 1.5K. It is hypothesized that a change in the tropical atmosphere and ocean circulation, perhaps forced by Milankovitch cycles, which causes the lower mid-troposphere to dry would be the most effective way to induce strong cooling of tropical climate.