An Experiment on the Sensitivity of A Global Circulation Model

The results of a series of six 60-day integrations of the Mintz-Arakawa two-level, global-circulation model, designed to test the hypothesis that sizable changes in boundary conditions could be detected in the solutions even in the presence of uncertainties in the initial conditions. The variables considered are zonally averaged and time averaged values from the last 30 days of each 60-day run; the significance of the changes is tested by a classical analysis-of-variance technique. The model was run with a standard set of initial conditions and with two independent sets of random temperature variations superposed on the initial temperature field. With the ice of the Arctic Ocean replaced with water at the freezing point, the model was run once with the other standard initial conditions and once each with two sets of temperature variations. Changes in boundary conditions in the model are shown to produce detectable changes in the general circulation.