A Stochastic Approach for Representing Ice Cloud Microphysical Processes in Models

Poster PDF

Description

The representation of ice microphysics in models is challenged by the range of ice particle shapes and sizes. Typical parameterization schemes use various hydrometeor categories defined by prescribed physical characteristics (e.g., density, fall speed) that broadly describe particle type (e.g., cloud ice, snow, graupel). These characteristics are traditionally specified using empirical parameters (e.g., mass (m)-dimension (D) relations, fall velocity V-D relations, single-scattering properties) derived from observations and held fixed in models. Although some studies have investigated the sensitivity of simulated fields to choice of coefficients, few studies have shown the impact of natural parameter variability. Here, a stochastic approach for representing such variability is described. A surface of equally realizable solutions in the phase space of coefficients used in model schemes, rather than fixed values, is used to represent uncertainty and co-variability. Two examples of the approach are shown. First, in situ size distributions (SDs) measured by cloud probes on the University of North Dakota (UND) Citation in the stratiform area behind a mesoscale convective system sampled 20 May, 2011 during the Mid-latitude Continental Convective Clouds Experiment are compared against those simulated using the Weather Research and Forecasting model with three different spectral bin microphysics schemes. Observed and simulated SDs, represented by ellipsoids in the phase space of gamma fit parameters used to describe SDs, are compared as functions of meteorological conditions allowing hypotheses on the role of varying microphysical processes in simulations to be tested. Second, coefficients used to define m=aD^b relations are determined as ellipses of equally realizable solutions in a-b phase space by minimizing the chi-squared difference between the reflectivity derived from the SDs and that measured by ground-based radar in the volume surrounding the UND Citation. The dependence of solution surfaces on environmental conditions establishes how meteorological conditions and spatial and temporal variability control the a-b coefficients. It is shown that fixed a-b coefficients in numerical modeling schemes do not adequately represent the variability of cloud conditions. Future efforts to determine how coefficient surfaces vary as a function of spatiotemporal scale and how they will be implemented in parameterization schemes are discussed.