Using the general circulation model of the Etablissement d'Etudes et de Recherches M¿t¿orologiques, a comprehensive simulation is made of the springtime Antarctic ozone depletion. Ozone is treated as an interactive variable calculated by means of a continuity equation which takes account of advection and photochemical production and loss. The ozone concentration is also used to compute the heating and cooling rates due to the absorption of solar ultraviolet radiation, and the infrared emission in the stratosphere. The daytime ozone decrease due to the ''perturbed'' chlorine chemistry found at high southern latitudes is introduced as an extra loss in the ozone continuity equation, with a rate dependent on altitude and temperature and adjusted to fit the ozone losses observed in September 1987 over several stations in Antarctica. Results of the perturbed simulation show a very good agreement with the ozone measurements made during spring 1987. The ozone decrease starts in late August inside the south polar vortex and reaches a minimum of 150 Dobson units in mid-October. The relative ozone hole experiment than in the unperturbed experiment, and the final warming in November is delayed by about 2 weeks. This simulation also shows the development of a high-latitude anomalous circulation, with a warming of the upper stratosphere resulting mainly from dynamical heating.

All these features have been observed in spring 1987. In addition, a substantial ozone decrease is found at mid-latitudes in a thin stratospheric layer located between the 390 and the 470 K &thgr; surfaces. It is attributed to the cooperative effects of chemical depletion at the edge of the vortex in situations where an upward meridional flow is forced by vertically propagating tropospheric motions (the ''chemical eddies'') and of the irreversible air mixing between the outer boundary of the vortex and the mid-latitudes. Above the 470 K &thgr; surface air within the vortex is more contained, and the vision of the vortex acting as a chemical containment vessel holds; but below that level the vessel is apparently leaking! A significant residual ozone decrease is found at the end of the model integration, 7 months after the final warming and the vortex breakdown. If there is a significant residual ozone decrease in the atmosphere, the ozone trends predicted by photochemical models which do not take into account the high-latitude perturbed chemistry are clearly inadequate. The ozone decrease anticipated for the next decades could be worse than expected. Finally, it is concluded that further model simulations at higher horizontal resolution, possibly with a better representation of the heterogeneous chemistry, will be needed to evaluate with more confidence the magnitude of the mid-latitudinal ozone depletion induced by the ozone hole formation. ¿ American Geophysical Union 1990