Understanding deep convective organization: simple stochastic approaches and new metrics to bridge the gaps

Deep convective clouds can be observed in a variety of organizational states, from spatially random distributions to more coherent structures spanning a wide range of spatial scales. One puzzling mode of organization found in idealized numerical studies is the so called convective self-aggregation, in which the clouds spontaneously transition from a random distribution in space to a regime where they are clustered. This phenomenon can have important implications for tropical climate and its sensitivity, but the problems are that the models do not agree on their representation of it and there is also a lack of consensus on how to best quantify organization in both modeling and observational studies. To shed light on the discrepancies among models, we introduced a much simpler stochastic reaction-diffusion model of tropical convection, which, in spite of its minimal complexity, is still adequate to reproduce the behavior of full-physics systems and captures the transition to aggregation at parameter values that are a reasonable approximation of the present-day tropical atmosphere. The simplicity of the model allowed us to derive a dimensionless parameter, referred to as the aggregation number, whose value robustly indicates whether a given experimental configuration would undergo aggregation or not at all. The aggregation number incorporates the model key parameters, namely, a tropospheric radiative overturning timescale, the efficiency of horizontal moisture transport and the strength of the convection-vapor feedback, as well as the domain size and the horizontal resolution, in an attempt to explain these latter sensitivities detected in modeling studies. We suggest that this quantity can help understand the differences between full-physics models of the atmosphere. Regarding the quantification of the organization level of cloud field scenes, to provide a better assessment a new index has been developed that solves many of the drawbacks and weaknesses of existing methodologies. The index categorizes the organization in an absolute sense, is robust to the details of the calculation algorithm and is linear in spatial scale for most used cases, allowing a quantification of the organization level over and also beyond the beta-mesoscale. These advantages make it suitable for use in model intercomparison projects and in the analysis of a wide range of observation products.