Radiative Characteristics of the Canadian Climate Centre Second-Generation General Circulation Model

Several observational datasets were used to assess the quality of the radiative characteristics of the Canadian Climate Centre (CCC) second-generation GCM. The GCM data were obtained from the Atmospheric Model Intercomparison Project (AMIP) simulation. Data corresponding to the period January 1985 through December 1988 were examined since this period of the AMIP simulation overlaps with the Earth Radiation Budget Experiment (ERBE) and the International Satellite Cloud Climatology Project (ISCCP) datasets. Attention was given to mean January and July conditions. Optical properties of surfaces, clear skies, and cloudy skies were examined. Ocean albedos are too high in the Tropics and too low in the polar regions relative to surface observations and theoretical estimates. Compared to a satellite-derived dataset, however, they are slightly underestimated. Throughout much of the Sahara and Saudi Deserts surface albedos are too low, while for much of Western Australia they are too high. Excessive amounts of snow in Southeast Asia seem to have been sustained by a localized snow albedo feedback related to inappropriate snow albedo specification and a weak masking effect by vegetation. Neglect of freshwater lakes in the Canadian Shield leads to negative and positive albodo anomalies in winter and summer, respectively. Like many GCMS, the CCC model has too little atmospheric H20 vapor. This results in too much outgoing longwave radiation from clear skies, especially in the Tropics. Neglect of all trace gases except for C02 and weak H20 vapor absorption exacerbate this bias. Assessment of the radiative properties of clouds was done very generally at this stage due to lack of confidence in available observational data. Total and high cloud fractions were compared to ISCCP estimates. Warm tropical oceans appear to have too much high cloud. Evaluation of low cloud fraction is less straightforward but it is clear that due to lack of a shallow convection scheme and coarse vertical resolution, the GCM is almost devoid of low clouds over cool oceans. Cloud radiative forcing CRF from the GCM was compared to CRF obtained from ERBE data. Globally averaged, net CRF is in excellent accord with observations but shortwave and longwave CRFs are too strong. Zonal averages, however, reveal biases in which clouds act to cool the Tropics too much and cool the high latitudes too little during summer, yet they warm polar regions too much during winter. Regional examination shows that these biases are confined largely to oceans. Tropical oceans have excessive shortwave CRF despite good total cloud amounts. This may be due to neglect of cloud geometry effects on solar radiative transfer.