Monday, 7 July 2014
Cloud and precipitation processes occur on scales much smaller than can be resolved by most current global climate models (GCMs), and the simplification is often made that these properties are horizontally homogeneous on the scale of model gridboxes. This simplification leads to significant biases in both the calculated radiative fluxes and in simulated satellite cloud diagnostics calculated using the CFMIP Observational Simulator Package (COSP), indicating that the full variability in hydrometeor properties cannot be ignored in GCMs and doing so leads to a misrepresentation of clouds and their radiative effects. We use output from the Superparameterized Community Atmosphere Model (SP-CAM), which embeds a two dimensional cloud resolving model into each gridbox of a traditional global climate model, to provide global fields with resolved subgrid-scale variability. We test the sensitivity to the subgrid-scale variability by replacing the horizontally varying hydrometeor properties with gridbox layer means and running COSP and a radiative transfer code on both the full fields and on the horizontally averaged fields. There are significant differences in the radiative fluxes and in the cloud diagnostics between the calculations done on the full and the averaged fields, including an increase in clouds with high simulated radar reflectivity factor with a corresponding decrease in those with low radar reflectivity factor, and an increase in MISR, ISCCP, and MODIS-simulated optically thick clouds with a corresponding decrease in optically thin clouds. The sensitivity of these diagnostics shows that a realistic treatment of clouds in large-scale models cannot neglect subgrid-scale variability in hydrometeors and must include parameterization of their variability on the scales that clouds actually occur at that is consistent between the cloud physics, radiative transfer, and cloud diagnostics schemes.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner