Using an ice sheet model that is asynchronously coupled to an atmospheric general circulation model (GCM), we have computed the volume and areal extent of northern hemisphere ice sheets as a function of atmospheric carbon dioxide concentration. The ice sheet model includes an asthenosphere component for computing isostatic adjustment. Our basic strategy involves determining the net rate of snow accumulation in the GCM as a function of CO2 by seeding all northern hemisphere land points initially with 10 m of (water equivalent) snow and determining the resultant net snow balance over the course of a year. The net snow accumulations were then used as input to the ice sheet model, which in turn was used to compute the resultant ice sheet height and extent after 25 kyr. In order to judge model uncertainty, two different but reasonable snowmelt formulations were used in the GCM.

The basic results of the simulations are: (1) The ice sheet volume is strongly a function of CO2 concentration, with a more or less monotonic increases in volume as CO2 is decreased. (2) The computed ice sheet volumes are an order of magnitude less than those of (reconstructed) Pleistocene ice sheets. These results suggest that CO2 changes are an important but not dominant factor in accounting for the Pleistocene ice ages, though considerable future work is required to clarify this more fully. (3) While the choice of snowmelt formulation strongly affected the total volume of ice, the rate of change dV/d2> appears to be basically unaffected. (4) Through use of a computed ''glaciation sensitivity'' index we show that Greenland is by far the most likely to have an ice sheet: Tibet and Siberia were more likely to have an ice sheet than Canada or Fenno-Scandinavia: and Alaska and North Africa were least likely to have ice sheets (though each did occur for at least one CO2 value). These results represent the first step in the process of asynchronous coupling of a GCM and ice sheet model. These simulations do not account for such effects as the elevational feedback of the ice sheets on the overall climate or orbital insolation changes: only albedo effects are accounted for. In a future study we plan to continue the runs iteratively by using the GCM to determine new net snow accumulations consistent with the ice sheets computed in this present study and to investigate the possible effects of orbital insolation changes. ¿ American Geophysical Union 1992