Convective self-aggregation feedbacks in near-global cloud-resolving simulations of an aquaplanet

Positive feedbacks between precipitable water, reduced radiative cooling and enhanced surface fluxes promote convective self-aggregation in limited-area cloud-resolving model (CRM) simulations over uniform sea-surface temperature (SST). Near-global aquaplanet simulations with 4 km horizontal grid spacing and no cumulus or boundary layer parameterization are used to test the importance of these feedbacks to realistically organized tropical convection. A 20,480 x 10,240 km equatorially centered channel with latitudinally varying SST is used. Realistic midlatitude and tropical cloud structures develop. The natural zonal variability of humidity and convection are studied in a 30 day control simulation. The temporal growth of a small white-noise humidity perturbation and intrinsic predictability implications are explored. Atmospheric column budgets of moist-static energy (MSE) quantify its covariation with precipitation, surface heat flux, and radiative energy loss. Zonal Fourier analysis partitions these budgets by length scale. Radiative feedbacks on MSE natural variability and perturbation growth are found to be positive, broadly similar across scales, and comparable to limited-area CRMs, capable of e-folding a column MSE perturbation in 6-14 days. Surface fluxes are highest in synoptic-scale dry intrusions, inhibiting aggregation by damping tropical MSE perturbations. Sub-4-day MSE variations are due mainly to advection. Both tropics and midlatitudes have large-scale intrinsic predictability horizons of 15-30 days. An identical simulation but with 20 km grid spacing has more mesoscale variability and low cloud.