Buying Clouds in Bulk? Strengths and Limitations of Bulk Microphysics Parameterizations of Warm Phase Processes

Bulk and bin microphysics schemes are the two most popular and commonly used approaches to representing cloud processes in cloud-resolving models. The primary difference between the two schemes lies in their representation of particle size distributions. Bulk schemes assume a probability distribution function (PDF) for each particle category and predict bulk quantities (e.g. total mass, total number concentration) of the PDF. Bin schemes on the other hand explicitly predict the mass and/or number of particles in discrete size bins and, in so doing, explicitly predict the shape of the size distributions. Myriad studies have been conducted to compare simulations using bin and bulk schemes to determine if bin schemes can more accurately represent clouds in models. The issue that inevitably arises is that any particular bin scheme and particular bulk scheme differ not only in their representation of the size distribution, but also in the assumptions (such as fall speed-diameter relationships) used in the parameterization of each microphysical process. Therefore, it is impossible to determine whether differences arise between two schemes due to the representation of the size distribution, or due simply to the different assumptions about individual processes.

Here we present a methodology to isolate differences in simulations using the bin and bulk scheme approaches that arise solely due to the simple vs. complex representation of the size distributions. This methodology can be applied to any model type, from 0D box models through to full 3D cloud-resolving models. It can also be easily adapted to test any combination of predicted moments for the bulk approach. We will present results in which we investigate the simulation of warm phase microphysical processes including condensation, evaporation, and collision coalescence. We find that condensation and evaporation can be well simulated with a bulk approach when the shape parameter is known. Larger differences arise between the two approaches for collision coalescence, even under the most ideal conditions. Some combinations of predicted moments used by bulk schemes clearly outperform others and these will be discussed. The results give unprecedented insight into the strengths and limitations of a “perfect” bulk scheme for representing warm phase cloud processes, and they can be used to guide the future improvement and development of these schemes.