Aggregated Convection and the Regulation of Tropical Climate

Simulations with cloud-permitting models run under horizontally homogeneous forcing in doubly periodic domains show that under certain circumstances, the near random spatial distribution of convective clouds collapses into one or a few superclusters. This appears to require, among other things, a strong enough feedback between convectively induced surface wind gusts and surface enthalpy flux; this in turn requires a high enough specified value of the sea surface temperature. These simulations show that the phase transition to the self-aggregated state is accompanied by a dramatic drying of the atmosphere above the boundary layer. This implies that were the sea surface calculated rather than specified, self-aggregation could lead to a decrease of the SST owing to the dramatic reduction of water vapor, an important greenhouse gas. At the same time, smaller SST is not conducive to aggregation, so it is possible that the convection would disaggregate as the SST falls. This suggests that the actual model state might be attracted to the phase transition to aggregation, an example of self-organized criticality. We test this idea first with a toy model, with very simple convective and radiative physics. This shows self-aggregation, provided there is sufficient feedback between convection and surface fluxes, and the system is indeed attracted to the phase transition and thus to a particular surface temperature. We will also report on preliminary efforts to demonstrate self-organized criticality in a cloud-permitting, full-physics model in which the surface temperature is calculated from the surface energy budget. We will discuss possible implications of these findings for tropical climate sensitivity.