....

A tracer release experiment was initiated in the deep Brazil Basin in February 1996 to study mixing, stirring and transport in the abyssal ocean. Approximately 110 kg of sulfur hexafluoride was released on an isopycnal about 4000 meters depth near 21°40' S, 18°25' W. This location is over the system of ridge spurs and canyons that run zonally towards the crest of the Mid Atlantic Ridge (MAR). The spurs attain depths of about 4400 meters in the vicinity of the tracer release and the canyon valleys reach about 5000 meters depth. Spurs and canyons both shoal to the east towards the MAR crest (located at 12 W at these latitudes) where individual bathymetric peaks extend to about 2000 m depth.

The tracer was surveyed in 1996 within 2 weeks of its release, and a number of fine- and microstructure profiles were taken with the High Resolution Profiler (HRP). The results of these measurements have recently been reported by Polzin, et al., Science, 276, pp. 93-96, 4 April 1997. Using HRP, we found weak turbulent dissipation at all depths (supporting a diapycnal diffusivity of approximately 0.l $cm2/s$) in the western half of the Brazil Basin where the bottom is smooth. In contrast, greatly elevated mixing rates were observed over the rough topography of the MAR. The diapycnal diffusivity increases towards the bottom and appears to increase towards the crest of the MAR. We estimated diapycnal diffusivity values of about 0.5 $cm2/s$ for the depth and region of the tracer cloud. From the microstructure observations, much greater values were deduced (> 1 $cm2/s$) closer to the bottom.

We have just returned from a survey cruise, 13 March - 18 April, 1997 on R/V Seward Johnson, where we sampled the distribution of SF₆ while conducting extensive finestructure and microstructure profiling work. The cruise took place approximately 14 months after the release of the tracer. Floats that were released with the tracer and surfaced prior to our cruise suggested that the patch would be approximately isotropic, with a radius of about 210 km, and centered within 25 km of the release point. This, together with the estimate of K_z in the patch, led us to expect that only the eastern edge of the patch (where the injection isopycnal approached the depths of the ridge spurs) would have started to feel the effects of intense mixing characterized by a diffusivity greater than 1 $cm2/s$. However, to our pleasant surprise, we found the tracer dispersed a great deal more than expected. The average diffusivity over the 14 months on the injection isopycnal estimated from water samples collected near the release point appears to have been at least 1.5 $cm2/s$. Much larger diffusivities appear to have prevailed between the level of the release and the bottom. This was manifested by the tracer having mixed all the way to the bottom of the canyons. From the more extensive microstructure observations from this year's cruise we estimate diapycnal diffusivities in the canyons of 10 $cm2/s$ or larger. Crude estimates indicate that virtually all of the tracer was found.

In addition to pronounced downward dispersion, the tracer was also apparently carried eastward by flows within the canyons of the MAR. Tongues of tracer were sampled within the canyons that extended well east of tracer patch edge at the injection depth. Tracer was found as far east as 13 W in the canyon over which the tracer was released. Based on the tracer distribution, this up-canyon flow may have a speed speed on the order of 1 $cm/s$. Large up-canyon flow is also inferred from a heat budget for the canyon waters in terms of the estimated downward turbulent heat flux from microstructure data. Thus, the spatial pattern of mixing, and the hypothesis of flow toward the ridge in the canyons put forth in the Science paper seem to be confirmed. The levels of dissipation and of mixing over the rough topography, averaged over the 14 months appear to be even higher than reported in the Science paper. This is borne out by the HRP measurements made during the recent cruise as well as by the tracer results, and so the two techniques seem again to be consistent with one another.

The lateral dispersion of the tracer was consistent with the float data, with 2 important exceptions. The first is the movement of deep tracer up the canyons already mentioned. The second is that the center of mass of the tracer moved about 130 km to the southwest over 14 months, while the center of mass of the floats moved only 16 km or so to the southwest. This may indicate a vertical shear, since the mean depth of the floats was about 300 meters less than the tracer. Some of this movement may also be diffusive, but biased south by a ridge spur just north of the release site that may have acted as a topographic barrier (although its height was 400 to 800 meters below the depth of the release). We should obtain more information on the persistence of this movement, and also further tests of the distribution of the mixing and the up-canyon flows in another tracer survey scheduled for next March/April. 3-D modelling, with accurate bathymetry will be necessary to extract most of the useful information out of these surveys.

Lastly, repeated finescale velocity profiles separated in time by ~6 hours confirm the suggestion that tidal motions contribute significantly to the small-scale shear field (that in turn sustains the enhanced turbulent dissipation). These motions appear to be generated as the barotropic tide flows over bathymetric features having characteristic wavelengths smaller than 10 km. The subsequent propogation and breaking of the internal tide allows the enhancement of turbulent dissipation well away from the bottom boundary. We are attempting to model the evolution of the internal tide as a balance between vertical propogation and the tendency for internal wave/wave interactions to transfer energy to smaller vertical scales. The goal of the work is to predeict the vertical profile of turbulent dissipation given the amplitude of the barotrpic tide and a horizontal spectrum of bottom topography. Preliminary results are encouraging.