A simple slab ocean of 50 m depth, which allows for seasonal ocean heat storage but no ocean heat transport, is coupled to a global spectral general circulation model with global domain, realistic geography, and computed clouds. Globally averaged, the annual mean surface air temperature increase computed over the last 3 years of an integration with a full annual cycle for 2¿CO2 compared to the control for 1¿CO2 is 3.5¿C. Zonal mean air temperature differences indicate stratospheric cooling and tropospheric warming as seen in other CO2 modeling studies. Greatest increases of surface air temperature in the 2¿CO2 case, compared to the control, occur near the sea ice margins. Retreat of sea ice in the 2¿CO2 case is associated with changes in the positions of the cloud maxima. Ice-free areas of ocean in the 2¿CO2 case, which are ice covered in the 1¿CO2 case, store relatively more heat during the summer season. Warmer surface air temperatures then occur in areas that are much colder in the control case because of the lack of the insulating effect of the sea ice, especially in winter. Increases of zonal mean precipitation are evident at most latitudes as a result of increases of available moisture evaporated from the warmer oceans. In the tropics this is associated with a strengthening of the mean meridional circulation and with intensification of the upper level zonal-component winds in the subtropics. Warming near the surface associated with the retreat of the ice line in the 2¿CO2 case slackens the meridional temperature gradient and results in weaker upper level zone-component winds in the mid-latitudes.

Three-year seasonal means of soil moisture show decreases in tropical and subtropical continental areas and increases at high latitudes, but at mid-latitudes the change depends on the season. An analysis of the statistical significance of the geographical distribution of 7-year seasonal means of suface air temperature and soil moisture differences is given for the 2¿CO2 case compared to the control. Areas of significant differences correspond to similar regions of large differences seen in the 3-year seasonal means. Certain regions experience summer drying seen in other studies, but zonal mean soil moisture differences show increases of soil moisture at mid and high latitudes of the northern hemisphere year-round, with a relative minimum of increase in late summer. These differences are attributed to large increases of soil moisture in late spring that persist into summer and cause a positive feedback with precipitation and low clouds. This inhibits continental warming and limits summer drying seen in the zonal mean as a result of the doubling of CO2. A comparison of the present experiment with the previous swamp model experiments is consistent with other studies in that the extent of sea ice in the control case, critically influences the climatic response to increased CO2 such that more extensive sea ice is associated with a larger response. The seasonal cycle along with ocean heat storage in the mixed layer model are shown to be important in producing a more realistic simulation of the present climate than does the swamp experiment and, presumably, a more credible response to increased CO2. However, in the context of recent studies the overextensive sea ice in the present mixed layer model suggests that the inclusion of ocean heat transport from a fully computed ocean model and a resulting sea ice distribution closer to the observed would possibly produce less of a response to increased CO2.