A GCSS intercomparison of single-column model simulations of a nocturnal drizzling stratocumulus-topped boundary layer

During the last year, the GEWEX Cloud System Study (GCSS) Boundary Layer Cloud Working Group has been conducting an intercomparison of simulations of a nocturnal drizzling stratocumulus-topped boundary layer using single column models (SCMs) and large-eddy simulations (LESs). The case is slightly idealized from observations taken during the second research flight (RF02) of the DYCOMS-II field experiment off the coast of Baja California in July 2001.

This presentation focuses on results from the 8 participating SCMs. These include column versions of four leading AGCMs (UKMO, MPI, GFDL, NCAR), the ECMWF weather forecast model, regional models from the Netherlands and Japan, and two higher-order closure models (Lappen and Larson) based on assumed PDF shapes that allow simulation of both stratocumulus and shallow cumulus convection. Six-hour simulations using (when feasible) both standard and high-resolution versions of each model were performed. High-resolution simulations had vertical resolution of 5 m and a timestep of 10 s. When possible, the cloud droplet or CCN concentration assumed in each SCM was matched to the observational estimate.

As in past GCSS stratocumulus intercomparisons, the liquid water path (LWP) simulated by the different SCMs and LESs quickly diverged from each other (and in most cases, from the observations) due to differences in the relationship of the simulated entrainment rate to the PBL structure. To still obtain a useful comparison of the SCM microphysics, we compared the time-evolution of the cloud base and surface drizzle rates vs. LWP across the different models. The cloud base drizzle rate tests parameterizations of precipitation production (which in reality is complicated by the PBL turbulence, usually largely ignored in such parameterizations), while differences between the cloud base and surface drizzle rate test the evaporation parameterization.

The SCMs scattered considerably around empirical relationships between horizontal-mean cloud base precipitation and LWP derived from DYCOMS-II and EPIC2001. This scatter was predictably reduced after participants were allowed to resubmit results after seeing the observations. Most SCMs underestimated the fraction of cloud base drizzle that evaporated before reaching the surface. This may reflect an assumed raindrop size that is too large for stratocumulus drizzle.