Spatial Dispersion and Statistical Description of Organized Cumulus Cloud Ensembles in Radiative Convective Equilibrium

Abstract

Abstract Descriptions of cloud ensembles in radiative convective equilibrium (RCE) rely mostly on the assumption that clouds are randomly distributed in space, a hypothesis that obviously fails in presence of strong organization. In this work, idealized RCE simulations at horizontal grid spacings ranging from 2 km to 125 m are analyzed, displaying a transition between complete randomness at coarse resolution to a high degree of cold pool driven organization at high resolution. The objective is then twofold: (a) characterizing the spatial cloud patterns focusing on correlation scales and association with covariates; (b) providing a statistical description of organized cloud ensembles to generalize existing theories based on complete randomness. It is demonstrated that organized cloud ensembles may be treated as heterogeneous point processes where individual clouds do not interact substantially with their neighbors. In other words, clouds may be considered as randomly distributed if we restrict the space to regions of high cloud density. In addition, it is shown that in highly organized situations local cloud counts may be modeled using negative binomial distributions, while cloud mass fluxes follow lognormal distributions. The total mass flux within regions of varying sizes can then be predicted by compounding the two aforementioned distributions. The total mass flux variability at scales ≲5 km is dominated by the heavy‐tails of the mass flux distributions, whereas at scales ≳5 km, it is entirely controlled by the spatial cloud count variability (the spatial cloud pattern). These results have important implications for the parameterization of organized convection in quasi‐equilibrium.

Keywords

• cloud resolving modeling
• convection organization
• statistical modeling