Recently, a new computational approach that couples cloud-scale dynamics with large-scale dynamics in global climate models (GCMs) has been proposed and implemented. In this approach, called a Multiscale Modeling Framework (MMF) or superparameterization, all the cloud-related parameterizations are removed from a traditional GCM and, in each GCM grid column, replaced with a 2D or a small 3D cloud resolving model (CRM). We propose to compare MMF (and parent GCM) simulations of the vertical profiles of cloud condensate and precipitation against CloudSat cloud-radar observations and retrievals. The proposed comparison and analysis of observations and model output will be conducted in two ways; one based on traditional aggregation of observations and model output on seasonal to annual scales, and one based on an atmospheric classification approach. We plan to use both CloudSat retrievals and reflectivity observations directly. CloudSat reflectivity observations can be compared to model output using a CloudSat-satellite-simulator. By this we mean that we will calculate from the high-resolution CRM, run as part of the MMF, the reflectivity field that would be observed by CloudSat and compare the statistical properties of the model-simulated vertical reflectivity distribution to the observed distribution. We believe the proposed combination of instrument-simulator and classification techniques will prove capable of uncovering model shortcomings and provide insights to developing new ideas and ultimately, a better model. Regardless of how the model output and data are aggregated (traditional or classification), and whether we use an instrument simulator or satellite retrieved parameters in the comparison, or both, we will use robust statistical tests to evaluate our results. Ultimately our intent is not only to evaluate the current model, but to use this framework to evaluate a number of currently planned and future model improvements.