Self-aggregation conceptualized by cold pool organization

In radiative-convective equilibrium (RCE) simulations, self-aggregation is the spontaneous emergence of one or several long-lasting convective clusters from an apparently homogenous atmosphere (Wing, 2019). This phenomenon may implicate the formation of tropical cyclones (Wing et al., 2016; Muller et al., 2018) and large-scale events such as the Madden-Julian Oscillation (Arnold et al., 2015; Satoh et al., 2016; Khairoutdinov et al., 2018). However, it remains poorly understood how cold pools (CPs) contribute to self-aggregation. Using a suite of cloud-resolving numerical simulations, we link the life-cycle and the spatial organization of CPs to the evolution of self-aggregation. By tracking CPs, we determine the maximal CP radius R_max ≈ 20 km and show that cloud-free regions exceeding such radii always grow indefinitely. Besides, we identify a minimum CP radius R_min ≈ 8 km below which CPs are too cold, hence negatively buoyant, to initialize new convective cells. Finally, we suggest a simple mathematical framework that describes a mechanism, where cloud-free areas are likely to form when CPs have small R_max, whereas large R_max hampers cavity formation. Our findings imply that interactions between CPs crucially control the dynamics of self-aggregation, and known feedbacks may only be required in stabilizing the final, fully-aggregated state.