Investigating Turbulence in Artic Mixed-Phase Stratocumulus Clouds with Large Eddy Simulations

Description

Turbulence within Arctic mixed-phase clouds is known to be one of the important components of the processes needed to maintain these types of clouds for periods of days. The cloud of interest was observed on 26 April 2008 during the Indirect and Semi-Direct Aerosol Campaign (ISDAC). In this study, a large eddy simulation of a springtime Arctic mixed-phase stratocumulus cloud is performed. Following the Arctic mixed-phase cloud's turbulent assessment, two subgrid-scale turbulence parameterizations are compared. The first scheme is a 1.5-order turbulent kinetic energy (1.5-TKE) parameterization that has been previously applied to boundary layer cloud simulations. The second scheme, Cloud Layers Unified By Binormals (CLUBB), takes a different approach to turbulent closure. CLUBB has been tested at regional-model and global model grid spacings, but never before in mixed-phase stratocumulus clouds at cloud-resolving scales. Several experiments were performed, using grid spacings characteristic of large eddy simulation, cloud-resolving, and regional model simulations. It was found that the 1.5-TKE scheme produced a simulation more consistent with the observed cloud liquid water mass mixing ratio and number concentration at a cloud-resolving horizontal resolution of 100 m. However, both schemes under-predicted the observed ice number concentration and over-predicted the observed ice water mass mixing ratio. CLUBB produced smaller liquid water values due to weaker predicted turbulent kinetic energy in the vertical column, which in turn led to a weaker cloud-top temperature inversion and an inability to maintain a persistent stratocumulus cloud. Increasing CLUBB's maximum mixing length scale by a factor of four led to an increase in the liquid water mass mixing ratio and cloud fraction. In testing the behavior of CLUBB at horizontal resolutions more common to regional models, it was found that CLUBB was more consistent in maintaining a steady-state boundary-layer cloud than the 1.5-TKE scheme over a wider variety of resolutions. CLUBB's assumed double Gaussian representation of the grid box means was also tested, and it was found this assumption holds for mixed-phase clouds.