Cloud development from simulated CLOUDS stereo-images

Convective clouds play a fundamental role in affecting meteorological processes in so far they contribute to the redistribution of energy and water in the atmosphere. For resolving the fine structures of such clouds at scales where turbulence and microphysics are key processes, a spatial resolution of about a few tens of meters has to be achieved. The CLOUDS instrument of the C3IEL mission will measure at such a high spatial resolution. Space-borne visible cameras will simultaneously image, under multiple view angles, the same surface domain. In order to prove the concept here we present a method for determining geometry and velocities of the cloud top from synthetic successive multi-angle radiances. The latter are obtained through the radiative transfer model 3DMCPOL starting from physical cloud fields generated by a large-eddy simulation (LES) via the Méso-NH model. 3D point clouds are obtained from stereo pairs or triplets through s2p, a Python library already used to produce elevation maps from high resolution satellite imagery. From the reconstructed cloud geometry, cloud envelope development rates are then retrieved. These are compared against ground truth cloud top development as derived from the physical cloud model.