The uptake and redistribution of chlorofluorocarbons (CFCs) CFC-11 and CFC-12 are studied in a series of world ocean model experiments. In part 1 of this study the sensitivity of the simulated CFC distributions to the model parameterization of air--sea CFC fluxes is examined within a control experiment. The control experiment represents a low-resolution ocean model with global coverage and a proper seasonal cycling in surface thermohaline and wind stress conditions. The specification of a surface ocean CFC concentration that is instantaneously in saturated equilibrium with the atmosphere is found to flux too much CFC into the model. Signatures of CFC-11 are found to be grossly overestimated in regions of deep and bottom water formation, both in the surface mixed layer and at depth. The use of a classical air--sea gas exchange formula (even on with a simplified gas transfer velocity that is independent of wind speed) is seen to greatly improve the CFC simulations at depth. In addition, the model reproduces many of the observed trends in surface CFC concentrations; namely undersaturation in regions of deep convection overturn and near-surface upwelling and supersaturation in the summer mixed layer.

In further sensivity experiments, we consider the effect of sea ice cover in limiting air--sea gas exchange in polar waters. It is found that bottom water in the Arctic Ocean and around the Antarctic continent is significantly reduced in CFC content once regions covered with sea ice are limited to fractional air-sea gas exchange. This more physically meaningful framework is found to further reduce the spurious uptake of CFC-11 and CFC-12 found under a ''saturated surface'' assumption. In a final sensitivity experiment the gas exchange rate is parameterized using a complete wind speed and Schmidt number dependence. The wind speed dependent gas forcing increases the surface CFC equilibration rate under the subpolar westerlies. On the other hand, the polar and tropical oceans witness reduced CFC uptake under a wind speed dependent flux regime. Simulated ocean CFC concentrations are compared directly with observational data in certain key areas for deep and bottom water formation. It is found that a reasonable representation of oceanic CFC is achieved in the convected water column in the Weddell and Labrador Seas.

In contrast, deep waters that have left the convective area with the model ocean currents are found to be deficient in CFC-11 in the North Atlantic Ocean. This is because the model advective timescale for North Atlantic Deep Water (NADW) outflow across the equator is too long compared with observed ocean estimates. The long timescale is not due to unrealistically sluggish deep currents. Rather, the path of NADW outflow includes a loop eastward from the Labrador Sea into the Northeastern Atlantic Basin, effectively increasing the required outflow journey by around 4000 km. This ages the water mass by almost 10 years, thereby yielding significantly lower CFC concentrations in the NADW extension. In addition, the outflow signature spreads too far into the eastern North Atlantic, presumably because the advective process is too broad and the horizontal diffusion too strong at depth. Contrasting the North Atlantic, bottom water CFC ventilation in the Southern Ocean is found to be too strong, even when significant levels of surface undersaturation are simulated in polar waters. CFC-tagged waters flowing into the deep south Atlantic basin (from the Weddell Sea formation zone) are too enriched in CFC-11, even when the deep signatures adjacent to the Antarctic shell remain close to observations.

This suggests that the advective timescale for bottom water ventilation is too rapid in the Southern Ocean. In addition, too much convective overturn persists in the Southern Ocean at 55 ¿S--70 ¿S, with unrealistically deep CFC-11 penetration noted it particular longitudes. this is because not enough older (CFC-deprived) water recirculates and upwells into the Southern Ocean. for example, more upwelled circumpolar deep water in the Southern Ocean would weaken the CFC-11 concentrations by contributing to a lower CFC mixture and by suppressing the convective activity in the region. Botton and deep level CFC signatures are broad and diffuse compared with the real ocean. The broadness of the CFC imprint is due, in part, to the model resolution, which gives any convective event a spatial extent of at east 3.75¿ longitude by 4.5¿ latitude and a bottom level CFC signal thickness excess of 800 m. An important finding of our study is that the vertical convection of unstable waters acts as the efficieny tracer ventilator of the ocean system. this has significant implications for numerical studies of the world's climate, since the meridional overturning has traditionally been considered the reason for the ocean's moderating influence during global warming scenarios. Our study suggests that the vertical convection would play a much greater role over the typical timescale for anthropogenic climate change. ¿ American Geophysical Union 1994