Various cloud parameterizations in large-scale models attempt to predict the amount of subgrid scale variability in cloud properties to address the significant non-linear effects of radiation and precipitation. Statistical cloud schemes provide an attractive framework for self-consistently predicting the variability in radiation and microphysics by using the width and asymmetry of the distribution of cloud properties.

Data from the Atmospheric Radiation Measurement (ARM) program are used to assess variability in boundary layer cloud properties for several cases of stratocumulus observed at the Oklahoma ARM site during the March 2000 Intensive Observing Period. Cloud boundaries, liquid water content, and liquid water path are retrieved from the millimeter wavelength cloud radar, laser ceilometer, and the microwave radiometer. Aircraft data provide complementary information on the variability of liquid water content and total water amount. Satellite observations provide information on the horizontal variability in cloud optical thickness, and this is used to assess how much of the variability observed by the ground-based sensors is due to advected spatial variability and how much is due to change in cloud properties over time. These data are used to test predictions of variability from a statistical cloud scheme with prognostic variance that is being tested in the GFDL global atmospheric model.