We use the Grenier-Bretherton moist turbulence parameterization. This features an explicit entrainment parameterization and includes feedbacks of condensation on buoyancy generation and entrainment. In the MM5, this is coupled to a new shallow cumulus convection scheme which includes representations of penetrative entrainment at the tops of overshooting cumuli, of buoyancy sorting of turbulently mixed air parcels, and of the strong feedback between cloudbase convective inhibition and cumulus mass flux.
To test the parameterizations, we did multiyear global climatological simulations with a T42 horizontal resolution, 30-vertical-level version of CCM3. We also ran three-month regional simulations of the NE and SE Pacific with MM5 using 60 km horizontal resolution and 40 vertical levels, forced at the lateral boundaries by time-varying ECMWF reanalysis from June-Aug 1987. The standard configuration of CCM3 greatly underestimates surface observer estimated boundary layer cloud amount and ERBE-derived albedo in the eastern subtropical oceans and overestimates them in the trade cumulus regimes. In contrast, the standard MM5 PBL schemes overestimate the cloud amount and albedo everywhere.
With the new turbulence parameterization, the general location, cloud thickness, and vertical structure of the stratocumulus regimes in the eastern subtropical oceans are better simulated in both models. At the coarse (250 km) horizontal resolution of the CCM, cloud is underpredicted in the coastal zone, but this is greatly improved at 60 km resolution. Without the shallow cumulus scheme, there is excess cloud and albedo in the trade cumulus regimes. Drizzle microphysics in both models make the simulated cloud thickness sensitive to the assumed cloud droplet concentrations in the clouds, with implications for the indirect effect of aerosols on cloud albedo.
Supplementary URL: