1. Introduction
The direct evaluation of buoyancy fluxes is a difficult issue in many oceanographic situations (e.g., in frontal eddies and currents), because it requires knowing the vertical currents, which are generally small (except, e.g., in convection phenomena, where these can be 2–8 cm s−1; Lilly et al. 1999), often of a few millimeters per second, and hence unresolved by most measurements. Nonetheless, observational studies like that of Schott and Johns (1987) in the Somali Current have shown that vertical current measurement from acoustic Doppler current profilers (ADCPs) can be potentially useful for the study of phenomena whose vertical velocities barely exceed a few millimeters per second. Indeed, ADCPs have been successfully used in other oceanographic applications involving vertical current measurements, such as studies of turbulence (Lu and Lueck 1999a, b; Stacey et al. 1999a, b; Stacey 2003), internal wave band eddy fluxes (van Haren et al. 1994; Gemmrich and van Haren 2002; van Haren et al. 2005), tidally induced vertical velocity (Marsden et al. 1994), convection and deep mixing (Schott et al. 1993), dense shelf water formation (Shcherbina et al. 2004), and others. The purpose of this paper is to show observational evidence of the vertical velocity and its associated vertical eddy heat flux in the central Gulf of Mexico, which is somehow associated with the activity of the anticyclonic eddies, shed from the Loop Current, prevalent in the upper levels. Apparently, the only study that reports direct ADCP measurements of vertical velocity within mesoscale (anticyclonic) eddies is that of van Haren et al. (2006), who show vertical velocities measured between 1900- and 2400-m depths in the Algerian Basin in the Mediterranean Sea, with values as large as ∼−3 cm s−1 during the passage of anticyclones that are O(100 km) in diameter.
As will be discussed in the following sections, our vertical velocity measurements show high-frequency (periods <2 days) signals attributable to the presence of tides, internal waves, and even biological activity. And as will also be discussed below, given the difficulty of removing this latter (nonphysical) signal in the high-frequency band and because this band contributes only about 15% to the mean heat flux, we focus more on the lower-frequency (periods >2 days) processes, which are mainly associated with the activity of the anticyclonic eddies prevalent in the surface layer of the Gulf of Mexico. This eddy activity is recorded in the low-passed time series of velocity and temperature fluctuations as vertically coherent events related to the passage of mesoscale anticyclonic eddies shed from the Loop Current.
The rest of the paper is organized as follows: section 2 describes the data and their preprocessing, as well as the quality control of the data useful for flux estimates. Section 3 describes the methods used for the analysis of the series of velocity and temperature, and the altimetry maps. Section 4 shows the results on estimating the vertical heat flux and its relation with the activity of anticyclonic eddies (shed from the Loop Current) at upper levels of the water column. In section 5, the implications of the vertical eddy fluxes are discussed. Section 6 summarizes the main results.
2. Data
a. Setup and ancillary data
Two upward-looking 75-kHz Teledyne RDI Long Ranger ADCPs at 361- and 1227-m depths (hereafter LR1 and LR2; see Table 1 for instrumental specifications) and four Aanderaa RCM11 current meters distributed along the water column (538-, 741-, 1449-, and 1955-m depths) were deployed on one mooring at 25°05.2′N, 90°30.0′W (Fig. 1) from May 2003 to August 2004, providing 16 months of velocity (3D in the ADCPs, and 2D in the current meters) and temperature measurements across the first 2000 m of the water column in the central Gulf of Mexico. The current meters provided hourly records, whereas the ADCPs provided records every 30 min, whence the ADCP data were averaged into hourly data.
Instrument depth variations (evidenced by pressure variations in the instrument’s records) were also available from the ADCPs, which, as will be explained below, were useful for normalizing the temperature data (recorded in the ADCPs and the current meters) to nominal depths. In addition to the mooring data, two hydrographic casts made in the vicinity of the mooring location, just before the deployment and just after the recovery of the mooring, provided temperature profiles with depth by lowering a Sea-Bird SBE911plus CTD (Fig. 2). The mean profile, resulting from averaging the temperature profiles from the two hydrographic casts, was used as an estimation of a “climatological” reference profile for the central Gulf of Mexico. This mean profile is consistent with that obtained by interpolating from the high-resolution (¼°) temperature analysis of the world’s oceans, version 2, of the National Oceanic and Atmospheric Administration (NOAA)/National Oceanographic Data Center (NODC) (www.nodc.noaa.gov/OC5/WOA01/qd_ts01.html) to the mooring position.
Maps of sea level anomalies for the Gulf of Mexico were useful for detection and tracking of anticyclonic eddies shed from the Loop Current, whose influence over our measurements is the focus of this study. These maps were obtained from the multiple-satellite [Jason-1, Envisat, Geosat follow-on (GFO), and Ocean Topography Experiment (TOPEX)/Poseidon] merged data available in the Archiving, Validation, and Interpretation of Satellite Oceanographic data (AVISO) Web site (see http://www.aviso.oceanobs.com), for the period from May 2003 to August 2004. These data have a spatial resolution of ⅓° and a temporal resolution of 3.5 days.
b. Data processing
The vertical excursions recorded by the ADCP pressure sensors allowed the correction of the vertical velocity records (measured at the ADCPs only) by adding the time rate of change of the instruments’ vertical displacements. This correction was done for completeness in the data correction but was not indeed necessary, because it was about two orders of magnitude smaller than the typical observed vertical velocity of the flow, so that it does not change the results in any significant way.
The hourly data series were low-pass filtered with a Lanczos filter of 48-h cutoff and decimated to 6-hourly values. The series were filtered because the vertical velocity records showed corruption (unrealistic magnitudes) by a nonphysical diurnal signal resulting possibly from zooplanktonic vertical migration. This motion has a magnitude of a few centimeters per second and occurs primarily at diurnal periods, as reported in several studies in which vertical currents were contaminated by this biological signal (see, e.g., Schott and Johns 1987) or, conversely, in which ADCPs were successfully used to measure zooplanktonic abundance and vertical migration (e.g., Flagg and Smith 1989; Tarling et al. 1998; Zhu et al. 2000; van Haren 2007). It is then uncertain how much of the variability within the spectral band <2 days is really caused by the physical signal of the flow, and this band contributes only about 15% to the mean values of heat flux discussed in section 4b.
Finally, all the series were smoothed with a 7-day running mean for better visualization. All the averages, covariances, and confidence intervals shown in the following sections were calculated before this smoothing, but only smoothed series will be shown hereafter.
c. Data quality
ADCPs perform four tests on velocity data, which are based on the comparison of the quality indicators of the records (correlation, echo intensity, error velocity, and percentage of good data) to the internal thresholds, defined during the instrument’s programming. Records with less than 64 correlation counts, 50 echo intensity counts, 25% good, or those whose vertical velocities exceed 2000 mm s−1 are excluded. Details on the data quality tests and the internal thresholds can be found in RD Instruments (1998), and further technical documentation is available on the Teledyne RDI Web site (www.rdinstruments.com).
The correlation, echo intensity, and percentage of good data in our ADCP records were well above, and the vertical velocity well below, the excluding thresholds mentioned above (Fig. 4). Indeed, their percentage of good velocity estimates (PG4), that is, the percentage of the velocity estimates per ensemble calculated by using the four ADCP’s beams (see RD Instruments 1998), is high in both instruments, with PG4 > 90%. Also, both the echo intensity and the correlation are above 80 in LR1 and above 95 in LR2. Nonetheless, only those data from bins 2 to 31 (from 333- to 43-m depth) in LR1 and from bins 7 to 48 (from 1148- to 738-m depth) in LR2 were retained to estimate eddy heat flux; the discarded bins showed either unrealistic data or poor correlation (r < 0.5) among the vertical velocity signals. In the uppermost LR1 records, the lack of coherence is probably a consequence of a sidelobe contamination from surface reflection (Schott 1986; Schott and Johns 1987; Flagg and Smith 1989). In old-fashioned narrow-band ADCPs, in the records close to the instruments’ heads (usually the first 1–3 bins), poor coherence has been reported as a result of a “residual transducer ringing” (Flagg and Smith 1989) or of a bias that the signal-processing filter is not centered on the spectrum of the Doppler-shifted signal (Chereskin et al. 1989). Also in narrow-band ADCPs, Schott (1988) discarded any significant interference by the mooring’s cable. A thorough analysis is required to determine the causes of this kind of bias in modern broadband ADCP Long Rangers (which do not even use any tracking signal-processing filter), such as those used in this study.
Another possible limitation in ADCP measurements can be large instrument tilts, because they can change the accuracy of both the compass and the tilt sensor, thus causing biased velocity records. Teledyne RDI recommends that the tilt not exceed ±15°. Fortunately, the tilt of both LR1 and LR2 never exceeded 5°; indeed, the tilt was generally small: 1.7 ± 0.1° in LR1 and 1.2 ± 0.6° in LR2. Moreover, no correlation was found between records of tilt and vertical velocity.
The left-hand side in Fig. 5 shows the time series of vertical velocity w at both LR1 and LR2, and the right-hand side shows the so-called error velocity e. The latter e is a standard output of the ADCPs, but the recorded values have to be divided by a factor of 4 cos θ (where θ = 20° is the beam inclination with respect to the vertical axis) to be comparable with w and thus to match the definition of the true velocity components (van Haren et al. 1994; van Haren 2000; Gemmrich and van Haren 2002; van Haren et al. 2005), because it is normalized so that its rms is the same as that for the horizontal velocity components (RD Instruments 1998; see also section 3.1 of Ott 2005). This error velocity e is the difference in beam vertical currents between beam pairs and is indicative of the current inhomogeneity across the beam spread (and/or a failure of one or more beams), so that its data can be used to verify the level of significance of w with respect to instrumental noise (van Haren et al. 2005). In the series shown in Fig. 5, w is larger than e and is also rather vertically coherent; indeed, the correlation coefficient r between the series of a bin and the series of the next bin (rejecting the corrupted data mentioned above) is significant throughout: 0.87 ± 0.05 in LR1 and 0.82 ± 0.05 in LR2. Nonetheless, the magnitude of w (∼2–3 mm s−1) is about one order of magnitude greater than that typically predicted by theoretical models of mesoscale vertical motion (e.g., Shearman et al. 1999; Viúdez and Dritschel 2003), even though Pallàs Sanz and Viúdez (2005) report in one of their maps a maximum value ∼−0.7 mm s−1, of the same order of magnitude as our measurements. The values of w are, however, smaller than those close to −30 mm s−1 observed by van Haren et al. (2006) at 1900–2400-m depth in the Algerian Basin during the passage of mesoscale anticyclonic eddies.
Although the accuracy of the 30-min ensemble data is ∼5 mm s−1 (see Table 1), here we use the 48-h-averaged (filtered) data, which are ∼factor 10 (
Finally, according to Ott (2005), the vertical velocity can be greatly biased by errors in the measured tilt angles of the instruments (as large as ∼5 mm s−1 for tilt errors varying from −2.5° to +1.5°). This can be inspected by plotting the change in angle δϕ (independently at each depth; pitch and roll separately) needed to minimize rms w (i.e., putting most of the energy into the horizontal velocity components) and the magnitude of this minimization. In our series, the minimized rms w does not change significantly (∼10−2 mm s−1) with respect to δϕ (with minimization-optimal values of |δϕ| < 0.6°), so we can be fairly confident in our results (M. Ott 2006, personal communication).
3. Methods
a. Estimations of vertical heat flux
When averages are calculated in Eqs. (5) and (7), it is important to establish a criterion of significance of such averages. As our observations are dominated by low-frequency motions lasting more than one month (see Fig. 3), our 16-month-long observation series are too short for a proper covariance analysis. Alternatively, here we use a simple bootstrap method for the estimation of the significance limits (for further information, see, e.g., Gentle 2002). We define the 96% confidence interval, whose limits are the 2nd and 98th percentiles of a sorted set of 1000 bootstrapped means (because these estimated means show a normal distribution and 96% of these are contained between these two percentiles). Function bootstrp of Matlab 6.5 was used for these calculations.
b. An eddy-centered coordinate system
As will be discussed in the following sections, significant signals of vertical velocity and temperature, and the corresponding covariances, in our observations are associated with the presence of mesoscale anticyclonic eddies over our mooring (identified by means of altimetry maps). And interestingly, there is an apparent horizontal dependence of the magnitude and sign of the vertical velocity (and its associated heat transport) within the eddies.
To explore this notion, we change the point of view from the three eddies passing over the observation point to an eddy with observation points in its interior, in an eddy-centered coordinate system. For such a purpose, we chose the altimetry maps coincident with the periods in which, as will be discussed in section 4c, three anticyclonic eddies shed from the Loop Current were over or near the mooring. Maps were then interpolated to coincide with the mooring-observation times (6-hourly). The next step was locating the centroid of the corresponding eddy and then plotting the variables observed by the mooring as functions of the distance between the eddy’s centroid and the mooring (r) in an eddy-centered coordinate system. This method can be helpful to elucidate the horizontal property distribution within the eddies.
Figure 6 shows the observation points in the eddy-centered system. Evidently, our sampling is not extensive enough to properly resolve the interior structures of the eddies. Only a few regions of the eddy are sampled: about five points are within the 50 km closest to the eddy’s center, and the southwest quadrant is not sampled. Nonetheless, this method is helpful to elucidate the eddies’ interior distribution of vertical velocity and temperature transport.
4. Results
a. Events of enhanced vertical heat transport
Here we analyze the vertical heat flux and show that it is related to the mesoscale eddies shed from the Loop Current, present in the upper levels. Figure 7 shows the time series of w′T ′ at different levels; these series correspond to Eq. (5) without averaging the covariant products of velocity and temperature. Three significant peaks are highlighted in the series, which are rather vertically coherent and last about one and a half months. Below 700 m, the events during the periods September and November–December 2003 are observed as upward heat transports (positive values), in contrast with the event during February–March 2004, which is essentially downward (negative values). These peaks coincide with features visible in the altimetry series (Fig. 8), as three consecutive events of negative vorticity and significant values of the second invariant of the surface velocity gradient, which is indicative of mesoscale eddies (see Isern-Fontanet et al. 2003). These eddies were shed from the Loop Current.
Figures 9 –11 show the time sequence of altimetry maps corresponding to the eddies that affected our observations. During the observation period, three eddies formed from the Loop Current and a fourth one was nearly detached by the end of this period. The first one was a small anticyclone formed in early August 2003 in the southeastern edge of a large anticyclone nearly detached from the Loop Current, east-southeast of the mooring location. These small anticlonic eddies are typically observed to be shed from the Loop Current, as well as from the larger warm-core rings, through their interactions with peripheral cyclones (Schmitz 2005; Leben 2005). As will be shown below, this minor eddy (hereafter MEd) had significant effects on our observations. The other eddy-formation events correspond to two of the so-called Loop Current eddies, “Sargassum” and “Titanic” (hereafter SEd and TEd), as they are named by the Eddy Watch program led by Horizon Marine, Inc. (see http://www.horizonmarine.com). SEd is a larger anticyclone from which MEd detached. The fourth eddy (that nearly detached by the end of the observation period), called “Ulysses,” shed from the Loop Current in mid-September 2004, about 3 weeks after the series ended, but it had little effect on the observation series, causing a peak toward the end of the series in Fig. 7. The details of the shedding of these eddies can be seen in eddies 15–17 in Plate 4 and Table 3 of Leben (2005).
The signals attributed to the eddies are observable as deep as 2000 m, as shown in the velocity series in Fig. 12. The velocities recorded by the ADCPs and those calculated from the altimetry are in good agreement, especially those of the first 333-m depths. The anticyclonic veerings of the vectors, associated with the eddies mentioned above, are common features in the series. Below 400-m depth, however, the velocity signals are associated with those near the surface, but there are significant differences among them.
Figures 13 –14 show the sections of depth versus time of the velocity and temperature during the influence of the eddies on the mooring. Above 400 m, the horizontal velocity is coherent, but below that depth, there are sign reversals of the velocity (Figs. 13a,b). This might be indicative of vorticity anomalies of opposite sign to the eddies’ vorticity in the fluid beneath them, or of an inclination of the fluid columns, or both. Also, during the presence of an eddy, coherent signals in the vertical velocity are observed, especially below 700-m depth (Fig. 13c).
b. Diapycnal flux versus adiabatic flux
It is now important to estimate the net contribution of the eddy-induced heat flux events on the mean state. Figure 15 shows the vertical profiles of the mean heat flux
Given the mean heat flux
Several authors (e.g., Gent and McWilliams 1990; Gent et al. 1995; Cessi and Fantini 2004; Henning and Vallis 2004; Cessi et al. 2006) have stated that below the surface mixed layer in the ocean, the flow is essentially adiabatic and consequently the mixing of material properties by mesoscale eddies occurs mostly along surfaces of constant density or isopycnal surfaces. On the other hand, the numerical results of Canuto and Dubovikov (2006) show that the mesoscale diapycnal flux must be of the same order of magnitude as the remaining terms of the mesoscale energy budgets, including the isopycnal flux, so that the diapycnal flux cannot be neglected. However, in all these studies the eddy-induced net exchanges result from averaging in time (the series length) and space (the whole basin), in contrast with our results from observations in a single point at the middle of the Gulf Basin. Then, our single-point observations should not be compared with the works mentioned above, but they must be thought of as observations of single eddy events like the basin-averaged ones in those works.
The diapycnal or isopycnal character of our observed fluxes remains to be clarified. It may seem reasonable to suppose that the observed vertical flux (Fig. 7) is essentially the vertical component of a flux occurring only along, and not across, isopycnal surfaces. Figure 7 shows the series of the adiabatic flux from Eq. (7), superimposed to the series of the direct estimation from Eq. (5). It can be observed that below 700 m both estimations show an important qualitative similarity and that both the events associated with the Loop Current Eddies, and their averages (Fig. 15), have the same sign. This similarity between the flux estimates suggests that part of the observed flux must be directed along the isopycnals. The heat flux induced by the mesoscale eddies must carry heat from the intermediate levels to the surface diabatic layer, whose thickness is changed by mixing, thus also changing the depth of the thermocline (e.g., Cessi and Fantini 2004; Henning and Vallis 2004; Cessi et al. 2006). However, at 333 m the two estimates are anticorrelated, which suggests that within the upper layer the vertical circulation differs from the isopycnal flux. With respect to this, it is important to keep in mind that only one data point is available in the upper 600 m; nonetheless, given the coherence of the vertical velocity signals recorded by LR1 and the fact that the temperature above 300 m should be similar (but greater) to that recorded at 333 m, it is expected that the heat flux pattern observed at 333 m continues toward the surface. In addition, the estimated adiabatic flux is an order of magnitude smaller than the observed flux (Fig. 15), which is in apparent disagreement with the adiabatic character of the mesoscale eddies.
c. Spatial variation of vertical velocity and heat transport
As shown in section 4a, the main qualitative difference among the signals caused by each eddy on the mooring is that the first two eddies induce signals of vertical velocity and heat transport of a certain sign, whereas the third eddy induces signals of the opposite sign (Figs. 7, 13 –14). From an altimetry perspective, the difference among the events is apparently that the MEd and SEd centroids were closer to the mooring (see Figs. 9, 10), whereas in the case of TEd only its periphery passed by the mooring (see Fig. 11). This suggests that the direction of the observed vertical heat transport depends on where within the eddy we are measuring: above 400 m, in the eddy’s centroid, the heat transport is negative (downward), whereas in the eddy periphery, it is positive (upward); below 700 m, the situation is reversed.
To explore the possible horizontal dependence of the vertical velocity (and its associated heat transport) within the eddies, we apply the method described in section 3b. Thus, in Figs. 16 –18 the values of w′, T ′, and w′T ′ at two different depths are plotted as functions of the radial distance r. These data correspond to values measured during the periods in which the eddies were over or near the mooring, that is, the periods shown in Figs. 13 –14.
Despite the large data scatter, the plots suggest some structural characteristics. SEd is the best example of this (Fig. 17), showing a fairly clear structure: T ′ decreases as r increases, as expected, and close to the eddy’s center (r < 75 km), the values of w′ and w′ T ′ are essentially negative at 333 m and positive at 738 m, and switch sign around r = 75 km at both levels; these patterns remain above 333 m and below 738 m (not shown). TEd, on the other hand, was not sampled close enough to the center to better compare it with the other eddies (Fig. 18), but the points suggest that the structure is similar to that shown by SEd, at least beyond r = 50 km. Even if these characteristic structures exist in MEd, the data scatter makes it difficult to observe (Fig. 16).
Then, the method works reasonably well for SEd, but it presents important deficiencies for MEd and TEd. This can be observed also in Fig. 19, where η and |(ug, υg)| show the expected patterns for SEd but important deviations from these for MEd and TEd. As mentioned in section 3b, the method presents deficiencies during the periods of strong deformations of the eddies, and both MEd and TEd presented important elongation and weakening/strengthening periods (see Figs. 16, 18).
The data scatter can be caused not only by noise in the data or by failures of the method but also by irregularities or perturbations within the eddies, most of them with spatial scales <⅓°, the resolution of the altimetry maps. In the eddies’ periphery there may be disturbances that introduce variations in the vertical velocity. Indeed, in other anticyclones with characteristics similar to those of the Loop Current, such as the Algerian Eddies, chlorophyll data and particulate matter distributions reported by Taupier-Letage et al. (2003) suggest different regimes of vertical motion associated with instability along the eddies’ peripheries.
Thus, the property distributions within the eddies are not axially symmetric (see Fig. 17), but they present significant differences between one quadrant and another, especially the velocity. So the eddies must possess more complicated interior structures. Our sampling coverage within the eddies is too small (in space and time) for properly resolving them (see Fig. 6). Nonetheless, our result is consistent with indirect results present in the literature; this will be discussed in the following section.
5. Discussion
We have addressed, from ADCP measurements, the existence of significant vertical velocities and enhanced vertical heat transport in the intermediate levels of the central Gulf of Mexico, which is driven by the mesoscale anticyclonic eddies, shed from the Loop Current, prevalent in the upper layers. During the passage of these eddies over the observation point, vertically coherent signals are observed in the series of vertical velocity, which results in significant covariant products with the temperature.
From the existent literature, what it is known about the distribution of vertical velocity in mesoscale eddies is mainly provided by theoretical studies, such as that of Viúdez and Dritschel (2003), who focus on a surface baroclinic eddy. Their results show vertical and horizontal quadrupolar structures with vertical symmetry and symmetry with respect to the eddy’s center (see Figs. 7, 8 of these authors). Nonetheless, apparently such a structure is not observed in our Fig. 13c, except perhaps toward the end of the series, in late February and early March 2004. In addition, the Viúdez and Dritschel’s (2003) results show that the w lobes are closely related with filaments present around the eddies. This notion is in agreement with Lapeyre and Klein (2006), whose results from their surface quasigeostrophic model show that the submesoscale filaments produce significant vertical velocities, by stirring processes outside the eddies; however, these filament-induced vertical motions have a significant impact on the net vertical flux of tracers into the surface layer, but they have a weak effect on the flux of density. In addition, results from an ocean adaptation of the generalized Q-vector ω equation by Giordani et al. (2006), which are focused on submesoscale dynamics (fronts and eddies), show that the strongest vertical currents are frequently associated with strong horizontal currents.
On the other hand, in contrast with the theoretical results mentioned above, indirect results from the analyses of biological and biochemical tracers, like that of McGillicuddy et al. (1998), have suggested that anticyclonic eddies are associated with downwelling in their cores. This notion is supported by the biochemical analysis for the Sargasso Sea by Sweeney et al. (2003) and the chlorophyll-image analysis for the Southern Ocean by Kahru et al. (2007), and these studies also suggest the presence of upwelling in the anticyclonic eddies’ peripheries. Similarly, observations (hydrography, chlorophyll, and nutrient sections) across three mesoscale eddies in the Algerian Basin suggest that mesoscale anticyclones possess a toroidal circulation (Isern-Fontanet et al. 2004; Taupier-Letage et al. 2003); as shown in section 4c, given our limited sampling, we cannot surely determine the interior structures of the eddies, but our results are somewhat in agreement with these authors. Downwelling is observed near the eddies’ centers at the upper levels, and opposite flows toward the peripheries. The variability in the eddies’ peripheries, on the other hand, can be due to the interactions of filaments with the ambient field, as suggested by the theoretical works mentioned above. Also interesting, and not addressed in the works mentioned above, is the reversed pattern in the intermediate levels. Further analysis is necessary to determine the tridimensional structures of the eddies.
It is important to note that as discussed by Sweeney et al. (2003), these vertical motions can depend on the temporal evolution of the eddies, that is, their life cycle. During the eddies’ formation and intensification phases, the density surfaces depress and upwelling establishes; on the other hand, as the eddies spin down, the density surfaces relax back to their mean positions and the upwelling decays. Also, eddies can increase and decrease intensity multiple times during their life cycle (as observed in Figs. 9 –11) due to interactions with the ambient eddy field. Interestingly, a good example of an interaction of a Loop Current Eddy with a smaller eddy is that which occurred between MEd and SEd in late October and early November 2003: once shed, SEd drifted westward roughly following the 1500-m isobath up to where it apparently “felt” the presence of MEd and started drifting southward to eventually interact (merger) with Med; both eddies merged into a single eddy (SEd absorbing MEd). This process probably favored a spinup of SEd, and hence of its vertical currents.
In terms of vertical heat transport within an anticyclonic eddy and according to what was observed in Figs. 16 –18, the heat transported in the eddy’s core may be essentially compensated by the heat transported in its periphery, by means of a vertical circulation cell. The result would be that the net vertical heat transport is barely a small fraction of the peaks ∼10−3°C m s−1 shown in Fig. 7, perhaps comparable to the magnitude of the adiabatic flux. This may suggest that a residual heat, not recirculated within the eddy, should escape from it and flow outward, mainly along the sloping isopycnals. Perhaps, from the vertical convergence of heat that occurs around 500–600-m depth, a small fraction of adiabatically transported heat may reach the upper levels and also the surface diabatic layer, where mixing dominates, feeding a diapycnal exchange at those levels. This may result in a net upward heat transport.
Nonetheless, the contribution of the internal waves (not present in our low-passed series) to the heat budget should not be discarded; the distribution of chemical tracers across a mesoscale anticyclone (with a horizontal scale of ∼150 km) in the South China Sea (Li and Pohlmann 2002) reveals strong vertical disturbances of the isolines at the eddy’s edge, which suggests enhanced vertical mixing and upwelling, probably associated with internal wave activity generated at the eddy’s wall. This is consistent with the theoretical results of Zhai et al. (2005), who showed that the vertical propagation of near-inertial energy is somewhat enhanced by the presence of mesoscale eddies, emphasizing the important role played by anticyclones that drain this energy quickly to the deep ocean.
The origin of the vertical velocity signals present in our observations remains to be clarified. Numerical results for a subtropical warm-core eddy by Lee and Niiler (1998) show that the vertical velocity generated by the wind–eddy interaction depends on the relative angle between the wind stress and the existing surface current. They found that under the current flowing against the wind, downwelling develops, while upwelling forms under a current in the same direction as the wind; and the depth of this secondary circulation cell is over 300 m. This is supported by results from the interaction of a uniform-assumed surface wind over observed eddies (with altimetry and shipboard observations), which tend to sustain anticyclones and dampen cyclones (McGillicuddy et al. 2007). However, in our data we do not find enough evidence that this mechanism might be responsible of the radial distributions shown in Fig. 16.
On the other hand, an intriguing fact is the existence of such an opposite behavior of the heat flux during the presence of the eddies. Curiously, this is consistent with the mean heat budget proposed by Rivas et al. (2005) for the deep Gulf of Mexico, which requires an upward heat flux from the intermediate levels (700–1300 m) to the upper levels (above 700 m) to balance the excess heat gained in the intermediate levels by exchanges with the Caribbean Sea through the Yucatan Channel. Interestingly, near the sign reversal of
Then, an upward mean heat flux like that required by Rivas et al. (2005) may be given by an along-isopycnal flux as described above, but it also might be induced by baroclinic instability of the eddies. Such a process requires an extraction of mean potential energy to drive the kinetic energy of the disturbances (e.g., Kundu and Cohen 2002), by means of a buoyancy flux, which, given a linear dependence between the density and the temperature, is directly proportional to the temperature flux
As already mentioned, there were no ADCP measurements along the “transitional” levels close to the 6.85°C isotherm depth, so that a revised experimental setup is necessary. It should include another ADCP measuring across the space between the LR1 and LR2 coverages, working with a frequency different to that of LR1 and LR2, to avoid any sound interference with them. Also, it would be suitable to use more thermistors at the depth levels of the vertical currents (within the ADCPs’ ranges), especially near the surface. Naturally, having more than one mooring would be useful to better resolve the eddies’ velocity structure and would also allow to estimate time-dependent horizontal derivatives, instead of using the thermal wind approach to estimate horizontal derivatives of the temperature. It would be important for the separation between moorings to be adequate to resolve the mesoscales—an intramooring separation of ∼30–50 km may be a proper choice.
A topic deserving of further research is the comparison between the observed vertical velocity field and that resulting from some method that combines a theoretical model and observations. A good example is the study by Pallàs Sanz and Viúdez (2005), who diagnose the mesoscale vertical velocity by solving a generalized omega equation using observed density and horizontal velocity data from three consecutive surveys in the Alboran Sea. Measurements of a Loop Current Eddy by an adequate experimental setup including ADCPs and current meters (like that suggested above), complemented by several hydrographic crossings of CTD/lowered ADCP (with an adequate resolution to resolve the eddy’s interior structure), would diagnose confidently the threedimensional structure of these mesoscale eddies.
6. Conclusions
Our observational results show evidence of significant vertical velocity and vertical heat transport in the upper and intermediate levels of the central Gulf of Mexico, associated with the presence of mesoscale anticyclonic eddies shed from the Loop Current, over the observation point. A simple analysis shows negative vertical current and vertical heat transport near the eddies’ centers and essentially positive ones in the peripheries; below 700 m, the pattern is indeed the opposite. Although further analysis is necessary, this is somewhat consistent with indirect results from distributions of biochemical tracers reported in the literature. Also, the variability observed in the eddies’ peripheries could be associated with stirring processes due to interaction of filaments with the ambient eddy field, as suggested by theoretical studies.
Somewhere around 600 m there is a vertical convergence of heat, and estimations of adiabatic heat flux suggest that part of this converged heat must escape outward from the eddy and flow along the sloping isopycnals up to the surface layers. This is in good agreement with previous results, which have shown that the Gulf of Mexico must export heat from the intermediate levels (700–1300 m) to the upper levels (above 700 m) to compensate for the excess heat gained by exchange through the Yucatan Channel. However, further observations are required to determine whether these hypotheses are valid.
Acknowledgments
We thank the reviewers for their critical comments and suggestions to an earlier version of this manuscript. This study was funded by MMS Contract 1435–01–02-CT-85309 and by CONACyT through its block funding of CICESE as well as through a postgraduate scholarship to DR. DR has also been supported by the National Science Foundation (NSF) Science and Technology Center for Coastal Margin Observation and Prediction (CMOP), NFS award 0424602, during the last stages of the preparation of this manuscript. We gratefully acknowledge the encouragement of Alexis Lugo-Fernández, and the fruitful discussion with Prof. William Young. The altimeter products were produced by Ssalto/Duacs and distributed by AVISO, with support from Cnes.
REFERENCES
Benilov, E. S., 2005: On the stability of oceanic vortices: A solution to the problem? Dyn. Atmos. Oceans, 40 , 133–149.
Canuto, V. M., and M. S. Dubovikov, 2006: Dynamical model of mesoscales in z-coordinates. Ocean Modell., 11 , 123–166.
Carton, X. J., and J. C. Mc Williams, 1996: Nonlinear oscillatory evolution of a baroclinically unstable geostrophic vortex. Dyn. Atmos. Oceans, 24 , 207–214.
Cessi, P., and M. Fantini, 2004: The eddy-driven thermocline. J. Phys. Oceanogr., 34 , 2642–2658.
Cessi, P., W. R. Young, and J. A. Polton, 2006: Control of large-scale heat transport by small-scale mixing. J. Phys. Oceanogr., 36 , 1877–1894.
Chereskin, T. K., E. Firing, and J. A. Gast, 1989: Identifying and screening filter skew and noise bias in Acoustic Doppler Current Profiler measurements. J. Atmos. Oceanic Technol., 6 , 1040–1054.
Flagg, C. N., and S. L. Smith, 1989: On the use of the Acoustic Doppler Current Profiler to measure zooplankton abundance. Deep-Sea Res., 36 , 455–474.
Gemmrich, J. R., and H. van Haren, 2002: Internal wave band eddy fluxes above a continental slope. J. Mar. Res., 60 , 227–253.
Gent, P. R., and J. C. McWilliams, 1990: Isopycnal mixing in ocean circulation models. J. Phys. Oceanogr., 20 , 150–155.
Gent, P. R., J. Willebrand, T. J. McDougall, and J. C. McWilliams, 1995: Parameterizing eddy-induced tracer transports in ocean circulation models. J. Phys. Oceanogr., 25 , 463–474.
Gentle, J. E., 2002: Elements of Computational Statistics. Springer-Verlag, 420 pp.
Giordani, H., L. Prieur, and G. Caniaux, 2006: Advance insights into sources of vertical velocity in the ocean. Ocean Dyn., 56 , 513–524.
Helfrich, K. R., and U. Send, 1988: Finite-amplitude evolution of two-layer geostrophic vortices. J. Fluid Mech., 197 , 331–348.
Henning, C. C., and G. K. Vallis, 2004: The effects of mesoscale eddies on the main subtropical thermocline. J. Phys. Oceanogr., 34 , 2428–2443.
Isern-Fontanet, J., E. García-Ladona, and J. Font, 2003: Identification of marine eddies from altimetric maps. J. Atmos. Oceanic Technol., 20 , 772–778.
Isern-Fontanet, J., J. Font, E. García-Ladona, M. Emelianov, C. Millot, and I. Taupier-Letage, 2004: Spatial structure of anticyclonic eddies in the Algerian basin (Mediterranean Sea) analyzed using the Okubo-Weiss parameter. Deep-Sea Res. II, 51 , 3009–3028.
Johns, W. E., T. J. Shay, J. M. Bane, and D. R. Watts, 1995: Gulf Stream structure, transport, and recirculation near 68°W. J. Geophys. Res., 100 , 817–838.
Kahru, M., B. G. Mitchell, S. T. Gille, C. D. Hewes, and O. Holm-Hansen, 2007: Eddies enhance biological production in the Weddell-Scotia Confluence of the Southern Ocean. Geophys. Res. Lett., 34 .L14603, doi:10.1029/2007GL030430.
Kundu, P. K., and I. M. Cohen, 2002: Fluid Mechanics. Academic Press, 730 pp.
Lapeyre, G., and P. Klein, 2006: Impact of the small-scale elongated filaments on the oceanic vertical pump. J. Mar. Res., 64 , 835–851.
Leben, R. R., 2005: Altimeter-derived Loop Current metrics. Circulation in the Gulf of Mexico: Observations and Models, Geophys. Monogr., Vol. 161, Amer. Geophys. Union, 181–201.
Lee, D-K., and P. P. Niiler, 1998: The inertial chimney: The near-inertial energy drainage from the ocean surface to the deep layer. J. Geophys. Res., 103 , 7579–7591.
Lee, H-C., and G. M. Mellor, 2003: Numerical simulation of the Gulf Stream System: the Loop Current and the deep circulation. J. Geophys. Res., 108 .3043, doi:10.1029/2001JC001074.
Li, L., and T. Pohlmann, 2002: The South China Sea warm-core ring 94S and its influence on the distribution of chemical tracers. Ocean Dyn., 52 , 116–122.
Lilly, J. M., P. B. Rhines, M. Visbeck, R. Davis, J. R. N. Lazier, F. Schott, and D. Farmer, 1999: Observing deep convection in the Labrador Sea during winter 1994/95. J. Phys. Oceanogr., 29 , 2065–2098.
Lu, Y., and R. G. Lueck, 1999a: Using a broadband ADCP in a tidal channel. Part I: Mean flow and shear. J. Atmos. Oceanic Technol., 16 , 1556–1567.
Lu, Y., and R. G. Lueck, 1999b: Using a broadband ADCP in a tidal channel. Part II: Turbulence. J. Atmos. Oceanic Technol., 16 , 1568–1579.
Marsden, R. F., R. Paquet, and R. G. Ingram, 1994: Currents under land-fast ice in the Canadian Arctic Archipelago. Part 1: Vertical velocities. J. Mar. Res., 52 , 1017–1036.
McGillicuddy Jr., D. J., and Coauthors, 1998: Influence of mesoscale eddies on new production in the Sargasso Sea. Nature, 394 , 263–266.
McGillicuddy Jr., D. J., and Coauthors, 2007: Eddy/wind interactions stimulate extraordinary mid-ocean plankton blooms. Science, 316 , 1021–1026.
Ott, M. W., 2005: The accuracy of acoustic vertical velocity measurements: Instrument biases and the effect of zooplankton migration. Cont. Shelf Res., 25 , 243–257.
Pallàs Sanz, E., and Á Viúdez, 2005: Diagnosing mesoscale vertical motion from horizontal velocity and density data. J. Phys. Oceanogr., 35 , 1744–1762.
RD Instruments, 1998: ADCP coordinate transformations: Formulas and calculations. RDI Tech. Manual P/N 951–6079–00, 26 pp.
Rivas, D., A. Badan, and J. Ochoa, 2005: The ventilation of the deep Gulf of Mexico. J. Phys. Oceanogr., 35 , 1763–1781.
Schmitz Jr., W. J., 2005: Cyclones and westward propagation in the shedding of anticyclonic rings from the Loop Current. Circulation in the Gulf of Mexico: Observations and Models, GeophysMonogr, 161 , , Amer. Geophys. Union. 241–261.
Schott, F., 1986: Medium-range vertical acoustic Doppler current profiling from submerged buoys. Deep-Sea Res., 33 , 1279–1292.
Schott, F., 1988: Effects of a thermistor string mounted between the acoustic beams of an Acoustic Doppler Current Profiler. J. Atmos. Oceanic Technol., 5 , 154–159.
Schott, F., and W. Johns, 1987: Half-year-long measurements with a buoy-mounted Acoustic Doppler Current Profiler in the Somali Current. J. Geophys. Res., 92 , 5169–5176.
Schott, F., M. Visbeck, and J. Fischer, 1993: Observations of vertical currents and convection in the central Greenland Sea during the winter of 1988–1989. J. Geophys. Res., 98 , 14401–14421.
Shcherbina, A. Y., L. D. Talley, and D. L. Rudnick, 2004: Dense water formation on the northwestern shelf of the Okhotsk Sea: 1. Direct observations of brine rejection. J. Geophys. Res., 109 .C09S08, doi:10.1029/2003JC002196.
Shearman, R. K., J. A. Barth, and P. M. Kosro, 1999: Diagnosis of the three-dimensional circulation associated with mesoscale motion in the California Current. J. Phys. Oceanogr., 29 , 651–670.
Stacey, M. T., 2003: Estimation of diffusive transport of turbulent kinetic energy from Acoustic Doppler Current Profiler data. J. Atmos. Oceanic Technol., 20 , 927–935.
Stacey, M. T., S. G. Monismith, and J. R. Burau, 1999a: Measurements of Reynolds stress profiles in unstratified tidal flow. J. Geophys. Res., 104 , 10933–10949.
Stacey, M. T., S. G. Monismith, and J. R. Burau, 1999b: Observations of turbulence in a partially stratified estuary. J. Phys. Oceanogr., 29 , 1950–1970.
Starr, V. P., 1968: Physics of Negative Viscosity Phenomena. McGraw-Hill, 256 pp.
Sweeney, E. N., D. J. McGillicuddy Jr., and K. O. Buesseler, 2003: Biogeochemical impacts due to mesoscale eddy activity in the Sargasso Sea as measured at the Bermuda Atlantic Time-series Study (BATS). Deep-Sea Res. II, 50 , 3017–3039.
Tarling, G. A., J. B. L. Matthews, R. Saborowski, and F. Buchholz, 1998: Vertical migratory behaviour of the euphausiid, Meganyctiphanes norvegica, and its dispersion in the Kattegat Channel. Hydrobiologia, 375/376 , 331–341.
Taupier-Letage, I., I. Puillat, C. Millot, and R. Raimbault, 2003: Biological response to mesoscale eddies in the Algerian Basin. J. Geophys. Res., 108 .3245, doi:10.1029/1999JC000117.
Thompson, A. F., S. T. Gille, J. A. MacKinnon, and J. Sprintall, 2007: Spatial and temporal patterns of small-scale mixing in Drake Passage. J. Phys. Oceanogr., 37 , 572–592.
van Haren, H., 2000: Comment on “An efficient method for determining the significance of covariance estimates”. J. Atmos. Oceanic Technol., 17 , 885–886.
van Haren, H., 2007: Monthly periodicity in acoustic reflections and vertical motions in the deep ocean. Geophys. Res. Lett., 34 .L12603, doi:10.1029/2007GL029947.
van Haren, H., N. Oakey, and C. Garrett, 1994: Measurements of internal wave band eddy fluxes above a sloping bottom. J. Mar. Res., 52 , 909–946.
van Haren, H., R. Groenewegen, M. Laan, and B. Koster, 2005: High sampling rate thermistor string observations at the slope of Great Meteor Seamount. Ocean Sci., 1 , 17–28.
van Haren, H., C. Millot, and I. Taupier-Letage, 2006: Fast deep sinking in Mediterranean eddies. Geophys. Res. Lett., 33 .L04606, doi:10.1029/2005GL025367.
Viúdez, Á, and D. G. Dritschel, 2003: Vertical velocity in mesoscale geophysical flows. J. Fluid Mech., 483 , 199–223.
Zhai, X., R. J. Richard, J. Greatbatch, and J. Zhao, 2005: Enhanced vertical propagation of storm-induced near-inertial energy in an eddying ocean channel model. Geophys. Res. Lett., 32 .L18602, doi:10.1029/2005GL023643.
Zhu, X-H., Y. Takasugi, M. Nagao, and E. Hashimoto, 2000: Diurnal cycle of sound scatterers and measurements of turbidity using ADCP in Beppu Bay. J. Oceanogr., 56 , 559–565.
(left) Location of the mooring in the central Gulf of Mexico (25°05.2′N, 90°30.0′W). (right) The vertical distribution of the sensors includes two 75-kHz Long Ranger ADCPs (LR1 and LR2) and four Aanderaa RCM11 current meters (Aa’s); the thick lines represent measurement coverage of the ADCPs. The bottom at the mooring position is 3590 m from the surface.
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Temperature profiles from two hydrographic casts made in the vicinity of the mooring location (see Fig. 1), just before the deployment (11 May 2003) and just after the recovery (27 Aug 2004) of the mooring. The thick line corresponds to the average of both profiles. Asterisks correspond to the averages of the observed temperature series. LR2’s domain is indicated; within this domain’s depth range and superimposed to the hydrographic profiles is the mean profile of the interpolated temperatures [see section 2b, below Eqs. (3)–(4)], barely distinguishable.
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Temporal evolution of the temperature (°C) at the six instruments from the central mooring, corrected for instrument depth variations. The superimposed thin line corresponds to uncorrected series. (c) The dotted line corresponds to the test applied to the interpolation method mentioned in section 2b, which presents a correlation coefficient r = 0.97 and rms = 0.08°C. Tick marks in the time axis denote the start of the month.
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Mean values of some data quality indicators for (a) LR1 and (b) LR2. (left scale) White circles correspond to echo intensity; triangles correspond to four-beam averaged correlation; error bars in both series indicate their std dev. (right scale) Black circles correspond to the percentage of good data per ensemble for velocity components estimated from four beams (PG4); the dotted lines correspond to the percentage of data estimated from either three or four beams.
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
(left series) Temporal evolution of vertical velocity w and (right series) error velocity e at LR1 and LR2 (separated by the horizontal dashed line). The vertical axis scale is located in the top left corner (mm s−1). The first number at the end of each series indicates the bin number; the second number indicates the corresponding depth (m). Tick marks in the time axis denote the start of the month.
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Observation points in an eddy-centered system. The data points correspond to the period of the eddies’ influence on the mooring. The three different marker types correspond to the three anticyclonic eddies described in section 4c: triangles correspond to MEd, crosses to SEd, and circles to TEd. Contours correspond to radial distance r ranging from 25 to 150 km (contour interval of 25 km).
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Temporal evolution of the vertical eddy temperature transport (in 10−5 m °C s−1) at eight different levels. Thick lines correspond to w′T ′ [see Eq. (5)]; thin lines correspond to (w′T ′)adiab [see Eq. (7)]. This latter series has been set ×10 for a better qualitative comparison between both series. Numbers on the left-hand side in each panel indicate the nominal depth of the series, those on the right-hand side indicate the mean over the period: the first ones correspond to w′T ′; the second ones (in parentheses) correspond to (w′T ′)adiab. Records with no ordinate axis are scaled to the upper next plot with an ordinate axis. Tick marks in the time axis denote the start of the month.
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Temporal evolution of (a) relative vorticity ω = (∂υg/∂x − ∂ug/∂y), (b) the second invariant of the surface velocity gradient Q = −(∂ug/∂x)2 − (∂υg/∂x) (∂ug/∂y) (see Isern-Fontanet et al. 2003), and (c) eddy kinetic energy Ke = (u2g + υ2g)/2, where (ug, υg) are the geostrophic velocity components calculated from altimetry data in the Gulf of Mexico, interpolated to the mooring location. According to Isern-Fontanet et al. (2003), eddy cores are characterized by Q > Q0, where Q0 is a “high,” positive arbitrary value; here, we choose the value Q0 = 0.54 σQ (with σQ = 2.98 × 10−11 s−2 the std dev of the series of Q maps), denoted by the horizontal thin line in (b). Tick marks in the time axis denote the start of the month.
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Time sequence of sea level anomaly maps (cm), at 3.5-day intervals, corresponding to MEd over the mooring location, denoted by a circle when the vertical velocity is positive at 333-m depth or by a cross when it is negative. The white line shows the Q0 = 0.54σQ (σQ = 2.98 × 10−11 s−2) contour, the same value of Fig. 8b. Notice that north-northeast of the mooring is SEd, which eventually affects it, as shown in Fig. 10.
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Same as in Fig. 9, but for SEd.
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Same as is Fig. 10, but for TEd at 7-day intervals.
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Temporal evolution of the stick vectors of the velocity anomaly (m s−1) at different levels in the central Gulf of Mexico. The series labeled by “surface” corresponds to geostrophic velocity calculated from altimetry, whose values were interpolated to the times of the mooring’s series. Records with no ordinate axis are scaled to the upper-next plot with an ordinate axis. Tick marks in the time axis denote the start of the month.
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Sections of depth vs time of the velocity anomaly components, during the period of influence of three mesoscale eddies on the mooring position. Tick 40 marks in the time axis denote the start of the month.
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Same as in Fig. 13, but for the temperature anomaly T ′ and temperature transports w′T ′ and (w′T ′)adiab [see Eqs. (5) and (7)].
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Profiles of vertical eddy temperature flux 
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Values of w′ (mm s−1), T ′ (°C), w′T ′ (10−4 m °C s−1), and |(u′, υ′)| (cm s−1) for MEd as functions of the radial distance r, at two different depths. The data points are same as those plotted in Fig. 6.
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Same as in Fig. 16, but for SEd.
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Same as in Fig. 17, but for TEd.
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Sea level anomaly η (in cm) and geostrophic speed |(ug, υg)| (cm s−1) as functions of the radial distance rfor MEd, SEd, and TEd. These variables were interpolated from the altimetry maps to the mooring position.
Citation: Journal of Physical Oceanography 38, 11; 10.1175/2008JPO3755.1
Some specifications of the ADCPs used in this study.