One way to diagnose climate models is to see how well they forecast the weather. Although not a new concept, the thrust of this work is to run a climate model in NWP mode and to study the evolution of the forecast in order to characterize the behavior of the physical parameterizations. Length of the forecasts are kept short (~5days) to insure that the interactions between the physical processes will not have time to obscure the specific model deficiencies in parameterizations. We initialized the atmospheric component of the NCAR Community Climate System Model (CCSM) with reanalysis fields from NCEP/NCAR and the ECMWF and then compared the forecasts in regions where there is sufficient data to enhance our confidence in the observations. In addition to the reanalysis, we use the Atmospheric Radiation Measurement (ARM) sites to provide even more accurate observations for certain variables. The intent of this study is to move beyond documenting model errors and determine the causes.

This work was performed under the auspices of the U.S. Department of Energy by the University of California, Lawrence Livermore National Laboratory under contract No. W-7405-Eng-48.