Developing and Testing a Novel Stochastic Ice Microphysics Parameterization for Cloud and Climate Models Using ARM Field Campaign Data

Principal Investigator(s):
Hugh Morrison, National Center for Atmospheric Research

Co-Investigator(s):
Greg McFarquhar, University of Illinois
Adam Varble, University of Utah
Wojciech Grabowski, National Center for Atmospheric Research
Ed Zipser, University of Utah
Junshik Um, University of Illinois

Ice microphysical processes have a critical impact on weather and climate by altering cloud radiative forcing, latent heating, precipitation, and storm dynamics, yet their representation in models is highly uncertain. This uncertainty is due in large part to the wide range of ice particle characteristics (e.g., concentrations, masses, shapes, densities, fall speeds, aspect ratios, scattering properties) that are controlled by parameters usually specified as constants in models. Previous studies have shown considerable sensitivity of model solutions to changes in these uncertain parameters, especially for deep convection. However, studies have generally tested the effects of parameter perturbations by holding them fixed during each simulation. Thus, the effects of parameter variability within simulations have generally been ignored. On the other hand, observations show large variability of these parameters over scales smaller than typical model domain sizes, and significant correlation in the variability of parameters. The effects of this parameter variability on simulated cloud and precipitation properties are essentially unknown.

We will use Department of Energy Atmospheric Radiation Measurement Program (ARM) field campaign observations to characterize this parameter variability, and develop an observationally constrained stochastic microphysical framework that can account for it. This approach represents a significant shift in the overall design and implementation of microphysics schemes, and can provide an avenue for further development of stochastic parameterizations in later work. It builds upon recent work developing stochastic physics schemes that perturb convection parameters. However, in contrast to these schemes, our approach is constrained by observations, and the stochasticity is therefore directly linked to observed parameter variability. Stochastic microphysics provides a physically based way to include variability in high-resolution “convection-permitting” weather and climate models run without a convection scheme, while also being relevant to coarser resolution large-scale models.

The stochastic framework will be implemented within two bulk microphysics schemes available in the Weather Research and Forecasting model (WRF). Impacts on convective and ice cloud processes will be tested using well-observed cases developed from ARM observations. The main goals are to: 1) quantify impacts of including ice microphysical parameter variability on deep convective cloud and precipitation processes, compared to using fixed parameter values, 2) characterize the spread of ensemble solutions generated from different realizations applying the stochastic scheme, with practical implications for ensemble weather and climate prediction, 3) identify which specific parameters cause the greatest impacts on cloud and precipitation properties when varied, thereby motivating a focus on measuring of these parameters in future field campaigns, and 4) assess potential improvement and causes for improvement of simulated cloud properties using stochastic ice parameters through comparison of model output with multi-frequency radar observations.

This is a collaborative project between three institutions: National Center for Atmospheric Research (NCAR), University of Illinois, and University of Utah. Each institution brings unique expertise needed to complete this work since this proposal covers many themes. Illinois will lead the analysis of in-situ observations to characterize parameter variability. NCAR will lead development of the stochastic approach and its implementation into WRF, and Utah will lead analysis of model simulations and comparison with radar observations. These components are strongly interdependent and will be integrated to complete this project.