Evolution and Lifecycles of Cloud Field Configurations, Not Just of "Clouds"

Sure, clouds (opaque contiguous areas) are easy to tally with image segmentation and track in time (albeit nonuniquely), defining the familiar "life cycle" of the session's call.

Here I argue that life and evolution metaphors would be better applied to cloud field configuration, as indicated by the entire horizontal pattern in satellite imagery, not just by tracking pseudo-"objects" defined by pixels touching. Quantifying possible field configurations, in order to estimate the unlikelihood (improbability) of observed configurations, provides a probabilistic approach to inferring the convective age of an airmass. Simulations show that shallow convection fields evolve steadily over a configuration 'life cycle' or age of at least tens of hours, far longer than individual cloud objects. Deep convective cloud field configurations can also indicate long airmass life spans or ages. Satellite data and information measures are used to quantify configuration evolution (controlling for forced mesoscales). A theoretical framework to infer convection's time dependent efficiency from the growth of configuration improbability is offered. This framework has implications for convective adjustment (time dependent approaches to time dependent target states). It also suggests the possibility of machine-learned pattern-to-profile retrievals, to wring data assimilation value out of the under-utilized information in fine-grained texture of optical imagery.