A method for characterizing the spatial organization of convection in deep convective systems’ cloud shield

Deep convective systems (DCSs) play a major role in the radiative budget and the water cycle over the tropics, as they are responsible for a significant part of the tropical precipitation and represents the major contributors to extreme rain rates. The spatial arrangement of deep convection within the convective system’s cloud shield exerts a strong influence on the morphology of the systems shield yet difficult to quantify objectively.

A new method is introduced that aims to evaluate this spatial arrangement of convective areas in the cloud shield. The method is based on 2D autocorrelation metrics and a stochastic approach to generate randomly organized scenes. A bootstrap technique permits to compare each scene with respect to these stochastic distributions. The technique is applied on a large satellite-based dataset and a non-supervised classification of spatial arrangement is performed. The classification reveals well separated classes corresponding to well identified organization of convection. The method is further applied onto idealized km scale simulations and is shown to hold also for the model. A comparison of the results of our approach with existing metrics will also be shown at the conference to highlight the added value of the present effort.