Main

Previous calculations of vertical mixing based on the stratification of the deep ocean1,2 assumed that a flux of 25 or 30 sverdrups (Sv) of water, made up of both deep and bottom waters, is injected at depths of about 4,000 metres and mixed upwards to depths near 1,000 m by turbulent mixing. Both reports conclude that the spatially averaged diapycnal (cross-density surface) mixing coefficient is 10−4 m2 s−1.

However, observations of turbulence3 and dye diffusion4 in the deep ocean indicate that there exists a background diapycnal diffusivity of only 10−5 m2 s−1, although much larger values are found in localized regions near rough topography5. The background value is consistent with mixing due to the background internal wave field, and the larger values are consistent with extra internal waves due to the interaction of currents with topography.

But it is not obvious that the latter is enough to raise diffusivity by an order of magnitude when averaged over the whole ocean. The extra power required to do this is also large6. If the efficiency is 20%, which is normally considered a maximum for the final stage of breaking internal waves, then the power required is 2.1 terawatts. This is just possible, given current estimates of the energy input from the wind and tides, but this figure does not allow for losses at other stages in the conversion process.

A contrasting view of the thermohaline circulation has come from low-resolution7 and high-resolution8 computer model studies of the ocean circulation. These show that between 9 and 12 Sv of deep water is brought to the surface by Ekman suction in the Southern Ocean. This is driven northwards in the surface Ekman layer and is reduced in density primarily by surface freshening. The model results also emphasize earlier observations9 that in the primary regions of bottom-water formation around Antarctica, the near-surface water masses have the same density as North Atlantic deep water. It is therefore not necessary for the bottom water to be mixed through the whole depth of the water column, only up to the level of the deep waters.

Using this new view of the thermohaline circulation, we need only consider the vertical mixing of the main deep-water mass, North Atlantic Deep Water, whose flux is estimated to lie between 14 and 17 Sv (ref. 10). Taking the larger of these two values and the smaller of the two model-based estimates of upwelling leaves a maximum of 8 Sv to be mixed vertically within the ocean. The 10−5 m2 s−1 background term can upwell 3 Sv, leaving 5 Sv to be upwelled by localized regions of intense mixing. If this view is correct, then the vertical mixing coefficient, averaged over the whole ocean, is less than 3 × 10−5 m2 s−1 and, assuming 20% efficiency, the total amount of extra energy required is less than 0.6 terawatts.

The revised values are consistent with existing observations of mixing within the ocean. They also emphasize again the importance of the Southern Ocean and imply that although further research is needed on the localized mixing in the deep ocean, such mixing does not control the thermohaline circulation.