Marine Stratocumulus during VOCALS: Comparing Microphysical Observations to Large-Eddy Simulation Results

Large-eddy simulation (LES) is a tool capable of resolving cloud-scale processes and has been used extensively for study of the stratocumulus-topped boundary layer. Understanding the strengths and deficiencies of LES is crucial if we are to use it effectively. The ability of LES to accurately represent detailed microphysics has been sparsely investigated (Khairoutdinov and Kogan, 1999), and we seek to advance knowledge in this area. Here we study how well LES coupled with an explicit binned resolving model can simulate daytime observations of stratocumulus dynamics and microphysics during VOCALS. Our observations were acquired from the CIRPAS Twin Otter on October 19th, 2008 centered around 20 S, 72 W. During this day a well-mixed, non-drizzling stratus-topped boundary layer of ~300m thickness was observed. The cloud top height, thermodynamic profile, and wind profile all remained relatively stationary throughout the observation period. Potential temperature and moisture content jumps were 15.2 K and -6.55 g/kg, respectively. The Passive Cavity Aerosol Spectrometer Probe (PCASP) showed aerosol concentrations elevated (~600 cm^-3) from what is expected for clean maritime conditions. The Twin Otter was outfitted an airborne phase Doppler Interferometer (PDI) providing detailed microphysical information about the cloud layer. The PDI data show a monomodal drop size distribution that exhibits little change in shape with change in liquid water content (LWC), consistent with extreme inhomogeneous mixing of air parcels. For our numerical model we employ the Regional Atmospheric Modeling System (RAMS, Cotton et al., 2001) in LES mode. So that the results of our LES best matched the detailed microphysical data from the PDI, we coupled a binned microphysical model to our LES (Feingold et al., 1996; Tzivion et al., 1987). Our LES cloud top height after model spin-up then matches the observations while model cloud base is 25 m than observations. Comparisons between probability distribution functions of LWC at matched heights in the observations and the LES results show reasonable agreement. The strength of simulated boundary-layer circulations is substantially weaker than what the observations suggest. These weak circulations are associated with a somewhat decoupled cloud layer in the LES which was not evident in our daytime observations. We suggest that this decoupling could be related to modeled overentrainment of free tropospheric air. Our model cloud top increases by 10 m over one hour of simulation, while no cloud top height increase was observed. For large LWCs (0.3 g/kg or greater) LES predicts drop size distribution remarkably well. For lower LWCs, the LES shows a substantial tail to smaller drop sizes not present in the observations. We attribute this discrepancy to the assumption of purely homogeneous mixing in the LES model. The LES also predicts a substantial number of small cloud droplets (~ 2 micron diameter) that are not observed by the PDI.