Preliminary results from a 16-year simulation of the OSU coupled atmosphere-ocean general circulation model (GCM) are evaluated by comparison with observations and other contemporary coupled GCM calculations. The results are also compared with those from control runs of the individual atmosphere and ocean GCMs to show some of the unique characteristics of the air-sea— ice coupling. The CGCM simulation was started with initial conditions obtained from the individual ocean and atmosphere model runs which were made with prescribed boundary conditions appropriate for each model. After a few months of the coupled run several conspicuous sea-surface temperature (SST) errors developed in the eastern tropical oceans and in the western oceans of the Northern Hemisphere. There is also an early indication of a steady SST warming and sea ice melting in the southern oceans at high latitudes. It took only a few months of the integration for the tropical SST errors to reach maximum values of 5°-6°C, while the middle and high latitude errors grew steadily during the entire period of the 16-year integration. Both the annual mean and seasonal variation of the simulated Antarctic sea-ice area are much smaller than the observed. Interestingly, all existing coupled model results, regardless of the degree of model sophistication, exhibit more or less the same error characteristics as found in the present simulation. In an effort to isolate the causes of such model errors a heat budget analysis of the upper ocean has been made using the simulated data. The results indicate that the relatively rapid initial warming in the eastern tropical oceans is due to excessively large downward insolation during summer. The less-than-observed cloud amount simulated in these regions is the probable cause of this insolation error. The SST errors in the western oceans are mainly due to the excessively large latent and sensible heat fluxes simulated there during winter. The fact that the regional patterns of these particular SST errors coincide with those of the simulated Aleutian and Icelandic lows indicates a significant contribution to the error by the erroneous atmospheric surface wind. Because of the complicated feedback nature of the air—sea—ice interaction, the SST and sea-ice errors in the southern high latitude oceans are difficult to diagnose. Nevertheless, at least the initial phase of the error development is found to be due to the excessively large insolation simulated during summer. In addition, the cooling by the Ekman transport simulated in these oceans is much weaker than in the ocean control simulation, and also contributed to the gradual SST increase and sea-ice melting around Antarctica.
ASJC Scopus subject areas