Imprint of the convective parameterization and sea-surface temperature on large-scale convective self-aggregation

Radiative-convective equilibrium simulations with the general circulation model ECHAM6 are used to explore to what extent the dependence of large-scale convective self-aggregation on sea-surface temperature (SST) is driven by the convective parameterization. Within the convective parameterization, we concentrate on the entrainment parameter and show that large-scale convective self-aggregation is independent of SST when the entrainment rate for deep convection is set to zero or when the convective parameterization is removed from the model. In the former case, convection always aggregates very weakly, whereas in the latter case, convection always aggregates very strongly. With a nontrivial representation of convective entrainment, large-scale convective self-aggregation depends nonmonotonically on SST. For SSTs below 295 K, convection is more aggregated the smaller the SST because large-scale moisture convergence is relatively small, constraining convective activity to regions with high wind-induced surface moisture fluxes. For SSTs above 295 K, convection is more aggregated the higher the SST because entrainment is most efficient in decreasing updraft buoyancy at high SSTs, amplifying the moisture-convection feedback. When halving the entrainment rate, convection is less efficient in reducing updraft buoyancy, and convection is less aggregated, in particular at high SSTs. Despite most early work on self-aggregation highlighted the role of nonconvective processes, we conclude that convective self-aggregation and the global climate state are sensitive to the convective parameterization.