The performance of the CCSR/NIES AGCM, which was recently developed by the Center for Climate System Research, University of Tokyo and the National Institute for Environmental Studies is examined mainly from the aspect of clouds and the hydrological cycle. This model is newly developed and has not participated in the AMIP program. The model uses the spectral transformation method in the horizontal and grid differentiation on sigma coordinates in the vertical. The physical parameterization includes a sophisticated radiation scheme using the 2-stream k-distribution method, a simplified Arakawa-Schubert cumulus scheme, a prognostic cloud water scheme, and turbulence closure scheme with cloud effect.
Results of 10-year integrations with T21 and T42 versions of the model are compared with ERBE and ISCCP products. Both versions have 20 vertical levels. Zonal mean values of cloud radiative forcing as well as OLR and planetary albedo compare well with satellite-derived values. However, there are several conspicuous discrepancies in longitudinal distribution. Their revision through adjustment of parameterization schemes is the subject of this presentation. The discrepancies include low OLR and high albedo within the subtropical anticyclonic regions over the ocean. The low-level cloud is too dense in these regions. On the other hand, the low-level cloud amount over the land is found to be too small. One of other defects of the model simulation is a severe dry bias of the tropical lower troposphere. Some of these aspects could be alleviated by adjusting treatments of clouds and sub-grid scale turbulence.