Unrealistic parameterizations are the Achilles' heel of climate models. When a single-column model (SCM), which consists of one isolated column of a global atmospheric model, is forced with observational estimates of horizontal advection terms, the parameterizations within the SCM produce time-dependent fields which can be compared directly with measurements. In the case of cloud microphysical schemes, these fields include cloud altitude, cloud amount, liquid and ice content, particle size spectra, and radiative fluxes at the surface and the top of the atmosphere. Comparisons with data from the Atmospheric Radiation Measurement (ARM) Program show conclusively that prognostic cloud algorithms with detailed microphysics are far more realistic than simpler diagnostic approaches. Long-term comparisons of quantities strongly modulated by clouds, such as monthly mean downwelling surface shortwave radiation, clearly demonstrate the superiority of parameterizations based on comprehensive treatments of cloud microphysics and radiative interactions. These results also demonstrate the critical need for more and better in situ observations of cloud microphysical variables.

We have used an SCM to examine the sensitivity of fundamental quantities such as atmospheric radiative heating rates and surface and top-of-atmosphere radiative fluxes to various parameterizations of clouds and cloud microphysics. The single-column model was run at the ARM Southern Great Plains, Tropical Western Pacific, and North Slope of Alaska sites using forcing data derived from operational numerical weather prediction. Our results indicate that atmospheric radiative fluxes are sensitive to the scheme used to specify the ice particle effective radius by up to 30 W m-2 on a daily time scale and up to 5 W m-2 on a seasonal time scale. We also found that the inclusion of ice particle fallout can have a significant effect on the amount and location of high cirrus clouds. On a seasonal time scale, atmospheric fluxes were sensitive to the inclusion of ice particle fallout by 8 W m-2. An unexpected finding was that the variance of the modeled ice particle effective radius at a given level is considerably smaller than that suggested by ARM cloud radar measurements. Our results indicate that this theoretical underestimate of the ice particle effective radius variance can have effects on modeled radiative fluxes comparable in magnitude to those cited above for sensitivity to the mean values of ice particle effective radius.