Anyone who looks at the sky can see that clouds have a big effect on weather and climate. But global computer models cannot deal explicitly with individual clouds, let alone the water droplets in them. Phenomena that mix the ocean are likewise important but too small to be resolved in models of ocean circulation. These 'subgridscale' processes need to be mapped, in both the atmosphere and the ocean. They also need to be understood so that their effects can be 'parametrized' — that is, represented in terms of larger-scale features that are dealt with explicitly in the models. Only in this way can we correctly allow for changes in such processes and their effects in a changing world. As they describe on page 513, Gregg and colleagues1 have produced an exciting advance in ideas about how both mapping and parametrization in the ocean might be accomplished.

The overall oceanographic context for their observations is as follows. Below a uniform surface layer that is typically some tens of metres thick, and directly stirred by wind and surface cooling, the ocean is stratified according to density, with lighter water above denser water. This gives it stability, which from time to time is disrupted by bursts of turbulence and mixing. The mixing helps to redistribute ocean properties, influencing circulation and heat transport, and hence climate2. Mixing also redistributes nutrients and dissolved gases, thus affecting biological productivity in the oceans.

Most of the turbulence and mixing is caused by an 'overturning instability', which has a vertical scale that varies from a fraction of a metre to many metres and is associated with the vertical gradient, or shear, of transient horizontal currents. These currents are a manifestation of internal waves (Fig. 1) — fluctuating motions that resemble waves at the ocean surface in relying on restoring buoyancy forces associated with a vertical change in density, but having much longer periods (typically tens of minutes to many hours). Some of the internal-wave energy originates as internal tides3, which are waves of tidal period generated by the main tidal currents flowing over bumps on the sea floor. Another source is the 'inertial' waves that are generated as the ocean responds to currents driven at the sea surface by the winds of fast-moving storms4. These waves have a period, related to the Coriolis force resulting from Earth's rotation, of 12 hours at the poles and much longer at low latitudes.

Figure 1: Internal waves and scales of ocean mixing.
figure 1

Waves produced at the sea surface and sea floor propagate into the ocean interior, generating small-scale mixing processes that affect circulation, heat transport and nutrient distribution (and so biological productivity). In confirming theory, two conclusions emerge from the observations of Gregg and colleagues1. First, that the vertical gradient of the horizontal currents with vertical scales of tens of metres is a good guide to overturning at a scale of a metre or less and turbulent mixing at the centimetre scale. Second, that the connection is strongly dependent on latitude. Note that the scale of this graphic is distorted to show processes at scales of metres and less in an ocean that is about 4,000 m deep.

Full size image

As the internal waves propagate into the ocean interior, some energy spreads into waves with other periods and generally smaller vertical scales, causing a greater vertical shear and hence more likelihood of wave-breaking, turbulence and mixing (Fig. 1). This cascade to smaller scales is reminiscent of what occurs in turbulent motions in unstratified water, with big eddies tearing one another apart and giving rise to ever smaller eddies, and it is also described by nonlinearities in the governing equations. The processes are, however, subtle and not fully understood. Nonetheless, a plausible model5,6 suggests that the cascade is not only more vigorous in places where internal waves are more energetic, but is also a function of latitude, with a slower cascade as one approaches the Equator.

It is this prediction that is so nicely confirmed by Gregg et al.1. They have simultaneously measured the energy level of internal waves (as largely represented by the shear of currents on a vertical scale of the order of ten metres) and the turbulent mixing at scales of a centimetre or less. They reinforce their earlier confirmation7, from a limited range of latitudes, that turbulence depends on the internal-wave energy, but they now have enough data from low latitudes to show that, as predicted, a given energy level at low latitudes causes much less mixing.

This confirmation of theory is as if a cloud physicist had shown that it is no longer necessary to study individual droplets, but only the larger-scale features of clouds. In the ocean we may no longer need to undertake the technically demanding measurement of the turbulence itself, but can predict it from motions at much larger scales. To be sure, the theoretical underpinnings of the formula are not completely robust, and it is known to break down in some locations8. We need more theoretical studies6 and numerical simulations9 of the energy cascade of internal waves. But it does seem that we have a good provisional basis for indirectly mapping the global distribution of ocean mixing, and of any seasonal or longer-term variations.

Such mapping would still require vertical profiles of horizontal current. This is a more sophisticated measurement than is routinely made. Alternatively, improvements in measuring and analysing profiles of temperature and salinity alone, from ships or profiling floats10, could provide a simpler approach based on analysis of, say, the overturning instability itself11.

Even with a confirmed parametrization of mixing, predictive models of ocean circulation will still need information on the internal-wave field, probably requiring a model to run in parallel. In a windier world, for example, internal waves might be more energetic but lose more of their energy closer to their source, changing the strength and spatial patterns of mixing rates. This kind of feedback on circulation and climate will have to be allowed for. Finally, although mixing from internal-wave-breaking is a key phenomenon in the deep ocean, particularly over rough topography, it may actually be less important for changing water properties in the top kilometre or so — there, some combination of air–sea interaction and stirring by eddies, with scales of tens or hundreds of kilometres12, may dominate.

There is still much to do before we can claim to have simple operational techniques for monitoring ocean mixing rates, and a full understanding of the mechanisms and effects of mixing. But the connections between motions with scales from millimetres to thousands of kilometres are becoming clearer, and we are closer to having the parametrization of small-scale processes that is needed if models are to have predictive capability.