A Bimodal Diagnostic Cloud Fraction Parameterization. Part II: Evaluation and Resolution Sensitivity

Abstract A wide range of approaches exists to account for subgrid cloud variability in regional simulations of the atmosphere. This paper addresses the following questions: 1) Is there still benefit in representing subgrid variability of cloud in convection-permitting simulations? 2) What is the sensitivity to the cloud fraction parameterization complexity? 3) Are current cloud fraction parameterizations scale-aware across convection-permitting resolutions? These questions are addressed for regional simulations of a 6-week observation campaign in the U.S. southern Great Plains. Particular attention is given to a new diagnostic cloud fraction scheme with a bimodal subgrid saturation-departure PDF, described in Part I. The model evaluation is performed using ground-based remote sensing synergies, satellite-based retrievals, and surface observations. It is shown that not using a cloud fraction parameterization results in underestimated cloud frequency and water content, even for stratocumulus. The use of a cloud fraction parameterization does not guarantee improved cloud property simulations, however. Diagnostic and prognostic cloud schemes with a symmetric subgrid saturation-departure PDF underestimate cloud fraction and cloud optical thickness, and hence overestimate surface shortwave radiation. These schemes require empirical bias-correction techniques to improve the cloud cover. The new cloud fraction parameterization, introduced in Part I, improves cloud cover, liquid water content, cloud-base height, optical thickness, and surface radiation compared to schemes reliant on a symmetric PDF. Furthermore, cloud parameterizations using turbulence-based, rather than prescribed constant subgrid variances, are shown to be more scale-aware across convection-permitting resolutions.