CMO Dye Studies - Sept '96 Cruise
The second cruise of the CMO dye studies was performed aboard the R/V Oceanus (leg 287) from September 4-18, 1996. Data were collected from CTD casts, the dye injection and sampling systems, shipboard ADCP, shipboard meteorological instruments, and drogues.
Two dye experiments were performed, the first using rhodamine and the second using fluorescein. During this cruise, and also the 1997 cruise, dissipation rates of temperature variance and turbulent kinetic energy were also measured by Neil Oakey of Bedford Institute of Oceanography (Oakey and Greenan; 1998). These microstructure observations are discussed breifly in the context of the dye studies by Sundermeyer (1998), and in a forthcoming paper by Ledwell et al.
A summary of the injection and sampling surveys for the two 1996 dye-release experiments is shown to the right. Insets to the figures show mean vertical profiles of dye as measured during the respective experiments.
A cruise report for this leg, written by J. Ledwell (chief scientist) is also available.
Three hydrographic transects were made during the 1996 cruise, the first on September 5, the second on September 8, and the third between September 17-18. The first transect was the most complete of the three, and consisted of 23 stations extending from approximately the 40 m to the 420 m isobath Contour plots (right) show the (a) temperature, (b) salinity, and (c) potential density from this section. The approximate location of the dye injections relative to the hydrography is shown as bold ellipses. Station locations are shown in plan view (d) (small circles) along with the CMO mooring locations (large circles).
As in the 1995 hydrographic section, the shelf-slope front is seen to intersect the bottom at around the 100 m isobath and extend toward the surface farther off-shore. The cold pool is also visible in the temperature record, again just inshore of the shelf-slope front. Unlike the 1995 section, however, this section shows no warm saline water intruding from offshore in the surface layers. In fact, on average, the shelf water was as much as 5oC cooler and 2-3 PSU fresher during this section compared to the one from 1995. Some of the differences in water properties are undoubtedly due to the passage of hurricane Edouard between days 245-246, 1996, which likely caused significant mixing over the shelf.
As in 1995, stratification remained fairly constant over the course of both experiments. However, for the first of the two 1996 experiments, the average buoyancy frequency at the level of the dye release were lower, only 5-6 cph compared to about 12 cph in 1995. For the second of the 1996 experiments, the buoyancy frequency was again about 12 cph.
Experiment 1, 1996
The injection for the 1996 rhodamine dye experiment was performed on September 7 while the ship was steaming southward at 1 kt. The injection cast began with a CTD profile to 50 m in an overall water depth of 70 m. The injection sled was then brought to the target density surface, sigmatheta = 24.063 kg m-3, or 40 m depth, and a total of 100 kg of dye released over a period of one hour. The standard deviation in density for this injection was 0.0035 kg m-3, which due to the low stratification in this part of the water column corresponded to a standard deviation in pressure of 2.5 dbar.
CTD data collected from the injection sled indicate that the depth of the target density surface steadily increased during the injection, from 30 m when pumping began to over 45 m an hour later. It is not clear to what extent this change in depth is the result of a temporal or spatial trend, i.e., whether it is the result of internal waves or some geostrophically balanced flow. (If interpreted as a strictly spatial trend, the corresponding tilt in isopycnals would be in the opposite sense as seen in the above hydrographic survey. Despite these variations in the depth of the target surface, however, the injection data show that the temperature and salinity remain fairly constant during the injection. This is also seen in T-S plots of these data (not shown), which show minimal scatter along the target isopycnal (see Sundermeyer, 1998).
The first survey of the 1996 rhodamine dye experiment was performed between 7 and 20 hours after injection and yielded ten transects with significant dye concentrations. Six meridional and two zonal transects spanned the whole patch, and two provided incomplete sections of the dye. A mass budget indicated that virtually all of the dye was found during this survey.
Plan-view maps of the vertical integral of the dye (see stick plot) show that the patch was also slightly elongated in the southwest/northeast direction; however, estimates of the major and minor axes indicate only a slight ellipticity (see Sundermeyer, 1998).
As in the 1995 experiment, the dye appeared well-homogenized in both the horizontal and vertical directions, although again some smaller-scale variability in concentration can be seen in individual transects. Significant vertical tilting was again evident in both the meridional and zonal transects.
The second survey for the 1996 rhodamine dye experiment was performed between 38 and 53 hours after injection. Due to problems with the real-time data display on-board the ship, this survey provided only eight transects through the dye patch. Of these, only one, transect M05 (see figure), appears to have spanned the breadth of the dye patch. Indeed, a mass budget suggests that only 45% of the dye was accounted for in this survey.
In contrast to the fairly smooth dye distributions seen in the 1995 experiment and in the first survey of this experiment, transects from this survey show significant patchiness in the tracer distribution, both zonally and meridionally. The non-Gaussian distributions seen in these vertical sections show multiple distinct maxima in the tracer concentration in both zonal and meridional transects.
The final survey of the 1996 rhodamine dye experiment was performed between 67 and 85 hours after injection, and provided nine transects through the dye patch. This was the most complete survey performed in this experiment owing to the regular spacing between transects and because it clearly delimited the patch boundaries. It is therefore unclear why a mass budget suggests that only 38% of the dye was found during this survey.
The meridional transects of this survey show significant homogenization of the patch compared to the previous survey, although patchiness is still evident. Some vertical tilting of the tracer patch is also seen in these transects, although it is not nearly as pronounced as that observed in the 1995 experiment. A northwest/southeast elongation of the patch is also evident in this survey.
Mean vertical profiles of dye from each of these surveys are shown to the right. For the first survey, the mean vertical variance was sigmaz2 = 12.9 m2, which corresponds to a mean vertical extent of 4*sigmaz = 14.4 m. By the final survey, the variance had grown to sigmaz2 = 22.8 m2, or a vertical extent of 4*sigmaz = 19.1 m. Using a Fickian diffusion model, a best-fit to the data thus implies a vertical diffusivity of Kz = 1-3 x 10-5 m2 s-1.
Horizontal diffusivities were again estimated in two ways. First, the overall dispersion Ktot was estimated based on the vertically-integrated tracer. Second, the irreversible component of diffusivity, Kirrev, was estimated based on the tracer along the target density surface.
Based on these two measures, the horizontal diffusivities inferred from the above two surveys were Ktot=12.7 (5.7 to 24.2) m2 s-1 and Kirrev=4.6 (3.0 to 6.5) m2 s-1.
Experiment 2, 1996
The second dye experiment of the 1996 cruise was conducted between September 13-17. Fluorescein dye was released in a series of closely spaced streaks, and subsequently sampled during three surveys. The injection for this experiment was not as clean as the previous one due to clogging problems with the injection system. After three failed injection attempts, each lasting less than 10 min, the system was recovered and modified to flow more freely. The ship was then re-positioned, and the injection completed in two additional pumping sequences lasting 15 min and 35 min, respectively. The total inject time was about 1 hour spread over a 4 hour time period.
The target density for this injection was sigma_theta = 24.03 kg m-3, which corresponded to a depth of 50 m. During subsequent surveys of the patch, the target surface was at 46 m depth. Again, 100 kg of dye were released along the target surface, with a standard deviation of 0.0067 kg m-3, or 0.19 dbar. Accounting for the separation of the injection streaks, the overall length of the injection was between 0.4-1.0 km based on a Lagrangian-corrected map of the injection.
A T-S diagram of the injection CTD data (not shown; see Sundermeyer, 1998) shows localized clusters of data points associated with the different periods when the dye was being pumped. These clusters imply a change in water mass properties from one portion of the injection streak to another. Further analysis of this change reveals a trend towards cooler, fresher water to the northeast. This corresponded to a shallowing of the target density surface toward the end of the injection than at the beginning. However, it is unclear whether this tendency is the result of spatial or temporal variability.
As in the previous experiments, stratification remained fairly constant. Buoyancy frequencies ranged from 8 to 15~cph, with typical values of 12~cph at the level of the dye. Finally, due to the threat of hurricane Hortense, three holey sock drogues were released during the second survey rather than during the injection.
The first survey of the 1996 fluorescein dye experiment was performed between 10 and 20 hours after the start of the injection Ten transects oriented in the northwest/southeast direction and one short transect in the southwest/northeast direction yielded significant concentrations of dye. The patch was well delimited by this survey and was somewhat elongated in the north/south direction. A mass budget shows that all of the dye was accounted for.
Although in some transects the dye patch appears homogeneous and even Gaussian in shape, other transects show dye distributions which are quite patchy and disjointed. These distributions lie in the southwest part of the patch where the injection itself was more inhomogeneous. Consequently, it is more likely that patchy distribution stems from the sporadic release of dye rather than from some stirring mechanism which acted on the patch after the injection. As in earlier experiments, significant vertical tilting of the dye patch is evident throughout this survey.
The second survey for the 1996 fluorescein dye study was performed between 47 and 68 hours after the start of the injection, and yielded three zonal and four meridional transects containing dye. A mass budget indicates that all of the dye was found during this survey.
As in the first survey, homogeneous and patchy dye distributions are visible, depending on the transect examined. It is unclear whether this patchiness is again a remnant of the injection discontinuity or if it is due to some physical process acting on the dye.
The third and final survey of the 1996 fluorescein experiment was performed between 95 and 114 hours after the start of the injection. This was the most complete survey of this experiment, consisting of twenty-one transects that delimited the dye patch very well. A mass budget shows that 70% of the patch was found.
Numerous transects of this survey show significant homogenization of the tracer patch, both in the horizontal and vertical directions. However, some transects show two or three distinct disconnected patches. Overall, the majority of the dye was located at the southwest end of the patch, while a long thin streamer of dye extended over 10~km to the northeast. As discussed in Sundermeyer (1998) this extreme elongation of the patch suggests the presence of a significant horizontal strain or shear acting on the patch.
Mean vertical profiles for each survey show a clear evolution of the patch over the course of the experiment. As in the other experiments, an upper bound on the vertical diffusivity was estimated by assuming a delta function initial condition for the dye. However, in this case the mean vertical profile from the third survey showed an asymmetry such that the upper half of the profile had a smaller variance than the lower. Using this profile as a final condition and assuming a Fickian diffusion model, the upper bound for the vertical diffusivity in the upper portion of the patch is Kz = 0.2 x 10-5 m2 s-1, while in the lower portion the upper bound is Kz = 0.9 x 10-5 m2 s-1.
As in the previous experiments, horizontal diffusivities were estimated in two ways. First, the overall dispersion Ktot was estimated based on the vertically-integrated tracer. Second, the irreversible component of diffusivity, Kirrev, was estimated based on the tracer along the target density surface.
Based on these two measures, the horizontal diffusivities inferred from the above two surveys were Ktot=2.5 (0.7 to 9.8) m2 s-1 and Kirrev=0.5 (0.1 to 1.1) m2 s-1.