Atmospheric convection grows differently over land and ocean. We address land-ocean contrast in convective organization using idealized cloud-resolving model simulations in a non-rotating radiative-convective equilibrium framework. In our ocean forced simulations, we mimic ocean and land forcing, exploiting their difference in heat capacities by fixing and oscillating sea surface temperature (SST), respectively, keeping temporal mean SST the same and also having a spatially homogeneous SST at a given time. Examining the speed of transition to an aggregated state we found that, an oscillating SST forces a faster onset of an aggregated state. Spatial heterogeneities of long-wave cooling are known to favor aggregation. With oscillating SST, we find that spatial anomalies of outgoing long-wave radiation are stronger, thus favoring aggregation. During the warm SST phase, long-wave cooling is enhanced in dry regions compared to neighboring moist convective regions. Stronger long-wave cooling allows stronger subsidence which allows low-level circulation to more efficiently transport moisture and energy up-gradient, driving convection to aggregate faster. We also note a sensitivity of our experimental setup to initial conditions, more so at warmer SST. This stochastic behavior, we suspect, might be critical in reconciling the differences of opinion regarding the response of convection aggregation to oscillating SST forcing.

Recently, \citeA{Biagioli2023} used a simple stochastic model to derive a dimensionless parameter to predict convective self aggregation (SA) development, which was based on the derivation of the maximum free convective distance ($d_{clr}$) expected in the pre-aggregated, random state. Our goal is to test and further investigate this hypothesis, namely that $d_{clr}$ can predict SA occurrence, using an ensemble of twenty-four distinct combinations of horizontal mixing, planetary boundary layer (PBL), and microphysical parameterizations. We conclude that the key impact of parameterization schemes on SA is through their control of the number of convective cores and their relative spacing, $d_{clr}$, which itself is impacted by cold-pool (CP) properties and mean updraft core size. SA is more likely when the convective core count is small, while CPs modify convective spacing via suppression in their interiors and triggering by gust-front convergence and collisions. Each parameterization scheme emphasizes a different mechanism. Subgrid-scale horizontal turbulent mixing mainly affects SA through the determination of convective core size and thus spacing. The sensitivity to the microphysics is mainly through rain evaporation and the subsequent impact on CPs, while perturbations to the ice cloud microphysics have a limited effect. Non-local PBL mixing schemes promote SA primarily by increasing convective inhibition through inversion entrainment and altering low cloud amounts, leading to fewer convective cores and larger $d_{clr}$.

Some of the classical models of tropical cyclone intensification predict tropical cyclones to intensify up to a steady intensity, which depends on surface fluxes only, without any relevant role played by convective motions in the troposphere, typically assumed to have a moist adiabatic lapse rate. Simulations performed using the non-hydrostatic, high-resolution model SAM in idealized settings (rotating radiative-convective equilibrium on a doubly-periodic domain) show early intensification consistent with these theoretical expectations, but different intensity evolution, with the cyclone undergoing an oscillation in wind speed. This oscillation can be linked to feedbacks between the cyclone intensity and air buoyancy: convective heating, radiative heating, and mixing with warm low stratospheric air warm the mid and upper troposphere of the cyclone stabilizing the air column and thus reducing its intensity. After the intensity decay phase, mid and upper tropospheric cooling, mostly through cold advection from the surroundings, cooled by radiation, rebuilds CAPE and triggers a new intensification. These idealized simulations thus highlight the potentially important interactions between a tropical cyclone, its environment and radiation.