The representation of cloud systems and cloud-radiation interaction is considered to be one of major uncertainties in GCMs. Most GCM parameterizations of cloud-radiation interactions assume that clouds are horizontally homogeneous and model results strongly depend on cloud-overlap assumptions and optical property approximations. To reduce the uncertainties, we attempt to utilize the month-long cloud-resolving model (CRM) simulation to examine the radiative effects of cloud inhomogeneity and geometric association.

While GCMs (and hence SCMs) require convection and cloud parameterizations, the CRM explicitly resolves convection and mesoscale organization, where cloud microphysical processes and cloud-radiation interactions directly respond to the cloud-scale dynamics. This produces vertical and horizontal distributions of cloud liquid and ice that interact much more realistically with radiation than SCMs and GCMs, where the radiative effect is calculated from a single volume of ``effective'' cloud. Consequently, the CRM simulation can get TOA and surface net shortwave fluxes to agree simultaneously with observations from TOGA COARE, whereas GCMs and SCMs generally fail to do so. The result demonstrates that the radiative effect of subgrid cloud variability is one of major factors responsible for the difference between the CRM and SCM radiative fluxes. Incorporation of such variability using the mosaic treatment will provide a cost-effective solution to reduce the SCM, and hence, the GCM, biases.