A two-dimensional climate model which links the northern hemisphere atmosphere, ocean mixed layer, sea ice, and continents has been asynchronously coupled to a model of the three main northern ice sheets and their underlying bedrock. The coupled model has been used to test the influence of several factors, including snow surface albedo over the ice sheets, in producing plausible ice age simulations using astronomically derived insolation and CO2 data from the Vostok ice core. The impact of potentially important processes, such as the water vapor transport, clouds, and deep sea circulation, was not investigated in this study. After several sensitivity experiments designed to identify the main mechanisms governing surface temperature and ice accumulations, the model is first run with ice sheet feedback by forcing it only with the astronomical insolation over the past 122 kyr. Large variations of ice volume are simulated between 122 and 55 kyr B.P., with a rapid latitudinal extension of the North American and Eurasian ice sheets starting at 120 kyr B.P. The simulated last glacial maximum is at 19 kyr B.P. The model is able to simulate the deglaciation as well. The simulated evolution of the three northern ice sheets is generally in phase with geological reconstructions. The major discrepancy between the simulation and paleoclimate reconstructions lies in the underestimation of temperature variations (linked with an underestimation of the ice sheet extent and an excess in the prescribed CO2 concentration).

Sensitivity experiments show that ablation is more important to the ice sheet response than snow precipitation variations. In the model a key mechanism in the deglaciation after the last glacial maximum appears to be the ''aging'' of snow, which decreases its albedo. The other factors which play an important role are, in decreasing level of importance, the ice sheet altitude, insolation, taiga cover, and ice sheet extent. A final set of experiments addresses the effects of CO2 on the simulated climate of the last glacial maximum and on a new long term experiment covering the last 122 kyr. This last experiment is made by forcing the model with both insolation and CO2 variations. This additional forcing improves the temperature and ice volume results. Despite the limitations inherent to the present modeling approach, the sensitivity experiments performed can provide insight into the relative importance of possible mechanisms responsible for the building and melting of huge ice sheets during the last glacial-interglacial cycle. ¿ American Geophysical Union 1992