Bayesian Retrievals of Vertically Resolved Cloud Particle Size Distribution Properties

Estimates of cloud particle size distribution properties are important for both Earth's radiative budget and hydrologic cycle. Retrievals of vertically resolved liquid cloud properties are complicated by the fact that there are generally more unknowns than observations available. Observational under-constraint is even more of an issue in mixed phase and ice clouds, for which there are a myriad of possible ice crystal shapes and densities, and even in situ observations are prone to large systematic errors.

We present results from a Bayesian Markov chain Monte Carlo (MCMC) retrieval algorithm for shallow liquid-only clouds over the ocean for three cases: one that appears to contain primarily precipitation, one that contains primarily non-precipitating cloud, and a third that appears to be a mixture of the two. We examine the relative constraint provided by passive visible and near-infrared reflectance and by vertically resolved radar reflectivity measurements. Use of a MCMC algorithm allows us to rigorously determine the information content of the observations, as well as the role of measurement uncertainty and prior information.