Uncertainties and Approximations in Bulk Cloud Microphysics Schemes & How to Address Them with BOSS, a Probabilistic Microphysical Parameterization Framework for Weather and Climate Models.

We discuss limitations in traditional bulk approaches to parameterizing cloud microphysics; these limitations are associated with various approximations, assumptions, and fundamental physical uncertainties. For example, most schemes make fixed assumptions about the form of the drop-size distribution, and typically assume separate cloud and rain categories separated by a fixed size threshold. Additionally, some microphysical processes such as collisional rain drop breakup are highly uncertain even at the scale of individual drops, and are thus targets for future laboratory research. All of these problems are exacerbated by the relative inflexibility of most parameterization schemes, which hampers systematic calibration and uncertainty quantification. We present work on development of the Bayesian Observationally-constrained Statistical-physical Scheme (BOSS), a novel probabilistic parameterization framework. BOSS combines existing, though limited, process level microphysical knowledge with flexible process rate formulations and parameters trained to observations and data from detailed process models through Bayesian inference. The approach is flexible and can be used to examine structural choices in bulk schemes, particularly the functional form of process rates and the number of predicted microphysical variables. This allows for systematic quantification of both parametric and structural uncertainty, which is not possible using traditional schemes. Results using an MCMC sampler with BOSS to constrain process rates and parameters with synthetic “observations' generated by a detailed bin microphysical model will be presented. We will show versions of BOSS with varying structure, including comparison of traditional two-category BOSS (cloud and rain as separate categories) to a new single-category BOSS, where rain and cloud droplets are treated together.