Turbulence, condensation, and liquid water transport in numerically simulated nonprecipitating stratocumulus clouds

Condensation and turbulent liquid water transport in stratocumulus clouds involve complicated interactions between turbulence dynamics and cloud microphysical processes, and play essential roles in defining the cloud structure. This work aims at understanding this dynamical-microphysical interaction and providing information necessary for parameterizations of the ensemble mean condensation rate and turbulent fluxes of liquid water variables in a coupled turbulence-microphysics model. The approach is to simulate nonprecipitating stratocumulus clouds with a coupled large eddy simulation and an explicit bin-microphysical model, and then perform a budget analysis for four liquid water variables: mean liquid water content, turbulent liquid water flux, mean cloud droplet number concentration, and the number density flux. The results show that the turbulence contribution to the mean condensation rate comes from covariance of the integral cloud droplet radius and supersaturation, which enhances condensation in turbulent updrafts and reduces evaporation in the downdrafts. Turbulent liquid water flux results from a close balance between turbulence dynamics and microphysical processes. Consequently, the flux can be parameterized in terms of the common diffusive downgradient formulation, fluxes of conservative thermodynamic variables, the turbulence mixing timescale, and the condensation timescale, which is determined by the droplet spectrum. The results also suggest that the condensation timescale regulates the turbulence fields, as does the number concentration, because it affects the condensation fluctuation, which is highly correlated with the turbulence vertical motion. A saturation adjustment cloud model, which diagnoses liquid water content at its equilibrium level, instantly condenses (evaporates) all available water vapor (liquid water) surplus. Consequently, there is likely to be a systematic difference between the turbulence field resolved with this type of model and that with a supersaturation-based cloud scheme for which a finite condensation timescale applies.