Diagnosis of Climate Model Simulations by Downscaling With a High-Resolution Regional Model

Climate models must parameterize subgrid-scale processes, and the diagnosis of the consequences of these simplifications and development of better parameterizations is an important part of improving climate simulations. Here we use a dynamical downscaling technique as a method of evaluating the performance of climate simulations from a coarse resolution general circulation model (GCM). For the eastern part of the United States, we first validate a regional model (Regional Atmospheric Modeling System - RAMS) by driving it with real atmospheric data (NCEP reanalysis) and compare the results to actual observations. RAMS is then nested within the NASA Goddard Institute for Space Studies GCM, using the same technique as when driven by reanalysis data. We compare the downscaled results to the GCM-scale simulations and to observations to identify systematic deficiencies in the GCM performance.