Evaluation of Clouds and Their Radiative Effects Simulated by the NCAR Community Atmospheric Model Against Satellite Observations

Cloud climatology and the cloud radiative forcing at the top of the atmosphere (TOA) simulated by the NCAR Community Atmospheric Model (CAM2) are compared with satellite observations of cloud amount from the International Satellite Cloud Climatology Project (ISCCP) and cloud forcing data from the Earth Radiation Budget Experiment (ERBE). The comparison is facilitated by using an ISCCP simulator in the model as a run-time diagnostic package. The results show that in both winter and summer seasons, the model substantially underestimated total cloud amount in the storm tracks and in the subtropical dry regions of the two hemispheres, and it overestimated total cloud amount in the tropical convection centers. The model, however, simulates reasonable cloud radiative forcing at the TOA at different latitudes. The differences of cloud vertical structures and their optical properties are analyzed between the model and the data for three regions selected to represent the storm tracks: the convective Tropics and the subtropical subsidence regions. Major cloud biases are identified as follows: the model overestimated high thin cirrus, high-top optically thick clouds, and low-top optically thick clouds, while it significantly underestimated middle- and low-top clouds with intermediate and small optical thickness. These multiple cloud biases compensate for each other to produce reasonable cloud forcing in the following way: for the longwave cloud forcing, excessive high clouds compensate for significantly deficient middle and low clouds; for the shortwave cloud forcing, excessive optically thick clouds offset significantly deficient optically intermediate and thin clouds. Possible causes of model biases are discussed.