Constraining and Estimating Microphysical Parameterization Uncertainty using Polarimetric Radar Observations and the Bayesian Observationally-constrained Statistical-physical Scheme (BOSS), a Novel Probabilistic Microphysics Framework

Microphysical parameterization schemes often suffer from errors and uncertainties, partly resulting from inadequately constrained parameters, and partly owing to structural errors and approximations inherent in their formulation. We propose to estimate and constrain both of these sources of uncertainty using a novel microphysical parameterization framework, the Bayesian Observationally-constrained Statistical-physical Scheme (BOSS), which eschews a fixed form for its microphysical process rates, avoids specification of a functional form of the drop-size distribution, and makes few of the ad hoc assumptions common in other, more traditional, bulk microphysics schemes. This scheme is coupled with a moment-based polarimetric radar forward-operator capable of estimating polarimetric radar observations and uncertainty therein from any combination of scheme prognostic moments. Development of the forward operator requires leveraging hundreds of millions of drop size distributions from disdrometers and detailed bin simulations. We constrain BOSS using a diverse set of observational regimes, resulting in a microphysics scheme that is capable of both predicting observed rain microphysics, and estimating microphysical process uncertainty -- a feature critically relevant to probabilistic forecast ensembles.