Recently, a new computational approach that couples cloud-scale dynamics with larger scale dynamics in global climate models (GCMs) has been developed. In this approach, called a Multiscale Modeling Framework (MMF) or superparameteri-zation, all the cloud-related parameterizations are removed from a traditional GCM and, in each GCM grid cell, replaced with a 2D or a small 3D cloud resolving model (CRM) [Khairoutdinov and Randall, 2001; Randall et al., 2003]. A CRM explicitly calculates cloud properties from physical equations at a scale consistent with cloud dynamics, thereby removing the need for the most problematic GCM cloud parameterizations. This approach has the potential to produce a more accurate representation of cloud and ideally an improved climate simulation.

High resolution CRM output from an initial global MMF simulation (of 500 days) along with output from the parent (parameterized) GCM has been compared against data obtained from the U.S. Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) Program sites [Ovtchinnikov et al., 2004]. Results of this comparison showed that the MMF simulation produced improvement in the representation of clouds and precipitation in the tropical Pacific but not in the Southern Great Plains (SGP) of the United States.

In this article, we extend the previous studies by Khairoutdinov and Randall [2001] and Ovtchinnikov et al. [2004] in comparing MMF and standard CAM climate model output with top of atmosphere fluxes retrieved from the NASA Clouds and Earth's Radiant Energy System (CERES) instrument [Wielicki et al. 1996] and cloud properties retrieved by the International Satellite Cloud Climatology Project (ISCCP) [Rossow and Schiffer 1999]. We use the ISCCP D1 data (in which all cloudy pixels in a grid-box are placed in one of 42 cloud optical thickness-cloud top pressure types) and compare this data to model output using an ISCCP-simulator following the approach of Webb et al. [2001].