The Cooperative Institute for Meteorological Satellite Studies (CIMSS) at the University of Wisconsin-Madison is heavily involved in GOES-R measurement simulation and demonstration activities. As part of this work, a very large-scale, high-resolution WRF model simulation was recently performed on a high-performance supercomputer at the National Center for Supercomputing Applications. The simulation contained a single 5950 x 5420 grid point domain covering most of the Meteosat viewing area (80º W to 80º E; -58º S to 58º N) with 3-km horizontal resolution. Approximately 1.5 TB of memory and 92,000 CPU hours were required to complete the 30-hr simulation.

The availability of this unique dataset provides a valuable opportunity to evaluate the ability of a particular physical parameterization suite to accurately simulate the atmospheric state for a diverse range of cloud types and conditions. Thus far, synthetic SEVIRI infrared brightness temperatures and CloudSat cloud reflectivities have been computed for this case. Preliminary comparisons between the real and simulated observations are very encouraging. For instance, probability distributions constructed for each SEVIRI channel show a remarkable level of agreement with mean brightness temperature biases generally less than 1 K over the entire model domain. Although limited in geographical coverage, CloudSat observations were used to examine the vertical structure of the simulated cloud fields. Overall, the simulation realistically depicted the vertical structure of many cloud types, particularly those associated with mid-latitude disturbances, but performed poorly in other regions, such as those containing shallow marine stratocumulus clouds.