1. Introduction
The upper-layer circulation of the Gulf of Mexico (GoM) is dominated by the Loop Current (LC) and the warm-core anticyclonic Loop Current eddies (LCEs) that detach from the LC. LCEs are coherent upper-layer structures confined mainly above 1000-m depth that propagate westward because of the beta effect (Nof 1981) at 2–5 km day−1 into the western Gulf of Mexico (WGoM) along three preferred paths (Vukovich 2007). Typically, LCEs are 200–300 km wide with maximum swirl speeds of 1.8–2.0 m s−1 (Oey et al. 2005; Sturges 2005). These structures dominate the mesoscale upper-layer circulation in the GoM and are thought to play an important role in the exchange of heat, momentum, and tracers between the eastern and western basins as well as within the WGoM, where most of LCEs strongly interact with the topography (e.g., Vidal et al. 1999; Vukovich 2007).
Less well known and understood is the interaction between LCEs and lower-layer circulation in the WGoM. There are few observational studies of the WGoM deep circulation (Kolodziejczyk et al. 2011; Kolodziejczyk et al. 2012; Hamilton et al. 2016) and of LCEs interacting with other eddies and the topography (Hamilton et al. 1999; DeHaan and Sturges 2005; Kolodziejczyk et al. 2012). Kolodziejczyk et al. (2011) analyzed the mean and subinertial currents in a layer from 1000-m depth to the bottom in the Bay of Campeche (BoC). The authors found mean currents in agreement with a cyclonic circulation at 1000-m depth and below 1000 m a flow field consistent with topographic Rossby waves (TRWs). The TRWs in the western BoC have periods between 5 and 60 days, horizontal wavelengths of 90–140 km, and vertical trapping scales up to 700 m (Kolodziejczyk et al. 2011).
Recent observations of the deep circulation in the LC system indicate the presence of deep eddies that are linked to the dynamics of the upper coherent structures (Donohue et al. 2016). The presence of deep eddies, specifically deep dipoles, and their westward translation in conjunction with the LCEs after a shedding event has been observed numerically (Hurlburt and Thompson 1982; Welsh and Inoue 2000; Romanou et al. 2004; Oey 2008) and explained using potential vorticity (PV) arguments in a simplified two-layer ocean by the water column compression ahead the LCE trajectory and stretching behind (Cushman-Roisin et al. 1990). Welsh and Inoue (2000) made a detailed numerical examination of the three-dimensional circulation in the Gulf of Mexico and found anticyclone–cyclone pairs beneath the LCEs as they separate and migrate westward. These deep anticyclone–cyclone pairs are “guided by the bottom bathymetry” (Welsh and Inoue 2000, p. 16 951) as they travel westward attached to the upper LCE. The authors also found a faster decay of the deep anticyclonic structure, which transits slightly ahead of the LCE, than the cyclonic part of the deep dipole just behind. The presence of deep anticyclone–cyclone pairs in conjunction with the surface eddy is schematically presented in Fig. 1, and the direct observation of this structure is the main subject of the present work. In Fig. 1a, the coupled eddies, translating to the west, are schematically represented in two layers, surface and deep. The system is composed of three structures: one anticyclonic in the upper layer (LCE) and two eddies in the deeper layer with an anticyclone leading and a cyclone trailing the upper-layer eddy. Figure 1b shows the meridional velocity field associated with this coupled eddy structure where the centers or cores of the deep dipole are below the rim of the LCE.
Schematic representation of the coupling system between surface and deep layers discussed in the present work shown in (a) horizontal and (b) layered views.
Citation: Journal of Physical Oceanography 48, 1; 10.1175/JPO-D-17-0018.1
Observations have suggested the presence of deep eddies in tandem with the mesoscale eddies dominantly residing in the upper layers. Interactions among eddies (LCEs as well as other cyclones and anticyclones) were investigated by Kolodziejczyk et al. (2012), who found intense currents lasting for 10–30 days with an enhanced barotropic component between counterrotating flows in the WGoM. Two deep highly instrumented moorings have been installed in the preferential path of LCEs in the western GoM, and several structures went through them. The analysis of those data, presented here, will show the existence of deep eddy signals consistent with the schematics of Figs. 1a and 1b. Using a more extensive mooring array, the intensification of the mean deep cyclonic circulation driven by the persistence and dynamics of the LCEs in the WGoM will be discussed.
The paper is organized as follows: Section 2 provides a description of the data. In section 3, the coupling between upper and deep circulations is investigated using (i) horizontal velocity measured by deep moorings located along one of the preferred LCEs’ path and (ii) normalized vertical vorticity computed from objectively interpolated velocity maps from an extended set of moorings deployed in the WGoM. A general discussion of the results and main conclusions are presented in section 4.
2. Data
a. Moorings
The principal dataset for this study comes from a set of 30 heavily instrumented moorings deployed from October 2008 to October 2013 in five different periods in the WGoM (see Fig. 2a). The moorings were distributed following the 130-, 500-, 2000-, and 3500-m isobaths gathered in nine perpendicular lines to the coast. A subset of two deep moorings (3500 m) were deployed during a shorter period of time (May 2010–August 2011), forming a triangle with mooring LMP3500 (Fig. 2a) to allow direct estimation of vorticity. The array consists of four moorings deployed at 3500-m depth: nine at 2000 m, nine at 500 m, and eight along the 130-m isobath. The experiment was initially designed to support the oil industry, guaranteeing at the same time that the scientific objective of understanding the circulation in the region would be fulfilled. For both the original and triangle experiments, the moorings sampled horizontal velocity throughout the water column. The deep moorings were equipped with three acoustic Doppler current profilers (ADCPs) covering from the near surface (30–50 m) to about 1200-m depth and a fourth one at 20 m above the bottom to survey the bottom boundary layer. Below 1200-m depth, single-point current meters were installed at nominal depths (see Fig. 2b for timeline of measurements at two different depths and Table 1 for details on the instrumentation mounted on the mooring lines and mean velocity results). Because of strong currents, vertical excursions of the instruments occur; nonetheless, the records of pressure sensors are used to determine the actual depth of the measurements. The maximum vertical excursion registered was 130 m during Jumbo’s crossing (May 2013). The mean measuring depth of the shallowest bin in these two deep moorings was ~50 m below the surface. We will focus on the velocity data measured by the LMP3500 (8) and ARE3500 (12) moorings that were instrumented with a 1) WorkHorse 300-kHz (WH300) ADCP looking up, 2) LongRanger 75-kHz (LR75) ADCP looking down installed at ~150-m depth, 3) LongRanger looking down from ~750 m below the surface, 4) WorkHorse 600-kHz looking down from 20 m above the bottom, and 5) six single-level current meters distributed from ~1300 to ~3000 m below the surface. All ADCPs provided data every hour by averaging ensembles of 32 to 120 pings depending on the instrument.
(a) CANEK group experimental design where 30 moorings (1–30) were deployed for 5 years along the 130-, 500-, 2000-, and 3500-m isobaths making up 4 (9) parallel (perpendicular) lines to the coast. Two more moorings (31–32) were deployed for a shorter period of time making up a triangle within one of the original design moorings (8). (b) Timelines of velocity data for 100- (black) and 2000-m (red) depth during October 2008 and October 2013 period.
Citation: Journal of Physical Oceanography 48, 1; 10.1175/JPO-D-17-0018.1
General mooring description in terms of depth of deployment and ADCP depth installation. Labels WH and LR mean Teledyne RDI’s WorkHorse and LongRanger ADCP, respectively. Labels up and down indicate the orientation of the ADCP beams through the water column. Current meters used are Anderaa models RCM11 and Seaguard, and Nortek models A2L and A3L. Length period of measurement and mean velocity field by components are also shown for 100- and 2000-m depths.
b. Remote sensing
Daily 1/4° × 1/4° gridded absolute dynamic topography (ADT) and absolute geostrophic velocity provided by AVISO are used to describe upper-ocean mesoscale circulation ubiquitous in the WGoM. AVISO data are also used to identify the position, trajectory, and size of the LCEs that reach the WGoM during the study period. The LCEs are identified (center and eddy edge) using a geometrical eddy detection method described in Chaigneau et al. (2008). The size is defined by the diameter of a circular vortex with the same area as the observed eddy. The LCEs identification has been made qualitatively on the ADT maps and quantitatively on the ADT anomalies with the knowledge of the lifetime of each LCE obtained online (from www.horizonmarine.com).
3. Coupling between upper and deep circulation
a. Deep eddies in the western Gulf of Mexico
In this section, we describe the coupling between upper and deep circulation in the WGoM. Figure 3 shows the study area (WGoM) and the mean geostrophic circulation (gray arrows) between October 2008 and October 2013 obtained from AVISO. Typically, the surface geostrophic circulation is in good agreement with the low-pass filtered (cutoff period of 7 days) currents at 100-m depth in the interior basin and less so near the coast measured with the moorings (Fig. 3). The black dots indicate the location of the moorings, and the black arrows are the mean ADCP current at 100-m depth measured for the same period (for detailed information, see Table 1). The color lines represent the trajectories of the seven LCEs that reached the WGoM during the study period. The moorings LMP3500 (black square) and ARE3500 (black circle) used to infer the vertical coupling between upper and deep circulation are strategically located within the most frequent LCEs’ paths and provide extended vertical data coverage.
The 5-yr time-averaged (October 2008–October 2013) horizontal velocity at 100-m depth (black arrows) and at the surface (gray arrows) obtained from mooring and satellite data (AVISO; http://www.aviso.altimetry.fr), respectively. The solid lines represent the trajectories of seven Loop Current eddies that reached the central western Gulf of Mexico during this 5-yr period. Symbols 
Citation: Journal of Physical Oceanography 48, 1; 10.1175/JPO-D-17-0018.1
Both satellite and mooring data, show that the upper-ocean mean flow is dominated by two large regions of oppositely signed circulation: one anticyclonic over the central WGoM centered approximately at ~(24°N, 95°W) and another cyclonic over the BoC, centered on the western side of the Bay of Campeche ~(20°N, 95°W). The existence of two regions of oppositely signed vertical vorticity leads to a westward mean surface flow centered around 21°N. On the continental shelf this westward flow splits into two branches: one to the north and the other to the south, both flowing along the bathymetry. The presence and location of LCEs modulate the latitude of the westward jet (Elliott 1982; Hamilton et al. 1999; Oey et al. 2005; Hamilton and Badan 2009) and the BoC cyclone (Vidal et al. 1992; Vázquez De La Cerda et al. 2005; Pérez-Brunius et al. 2013).
The 5-yr time series of the surface geostrophic meridional velocity (gray line) and ADCP meridional velocity υ at the LMP3500 and ARE3500 moorings at two different depths, 100 (black thick line) and 2000 m (black thin line), are presented in Fig. 4. Mooring data are filtered using a Lanczos filter with a 7-day low-pass cutoff period. The filtered ADCP meridional velocity at 100-m depth agrees very well with the surface geostrophic velocity obtained from AVISO. Both time series are characterized by strong meridional current events (υ ± 0.4 m s−1). Nine of these events are identified as LCEs crossing the mooring LMP3500. The time interval straddling a LCE crossing event over the mooring LMP3500 is defined as τmaxVN − 30, τmaxVS + 30, that is, 30 days before the time of maximum northward current above 0.3 m s−1 produced by the passage of the western edge of an LCE over the mooring and 30 days after the occurrence of the maximum southward current with speeds >0.3 m s−1 associated with the eastern edge of the LCE passing over the moorings. The 30 days account for the possible time lag between the deep and surface current responses to eddy translation through the mooring (see next section). An event lasts ~80–90 days on average (gray shaded regions in Fig. 4) and the time interval between the two absolute velocity maxima is ≥30 days to account for the transit time of the LCEs over the mooring. This transit time depends on the translation speed and size (diameter) of the LCEs. The names, sizes, and brief history of the LCEs identified during the study period are obtained from AVISO data analysis and Horizon Marine’s databases (http://www.horizonmarine.com) and shown in Table 2.
Time series of the meridional component of the horizontal velocity υ at the LMP3500 (23.38°N, 95.14°W and ARE3500 (22.50°N, 95.02°W) moorings. The gray solid line represents the geostrophic meridional velocity at the surface obtained from AVISO (http://www.aviso.altimetry.fr); black thick and thin solid lines are the meridional velocity measured at 100- and 2000-m depth by the current meters, respectively. The hourly velocity data are filtered and smoothed using a Lanczos low-pass filter (7-day cutoff). The gray shaded periods indicate the duration of the crossing events of the nine Loop Current eddies over the mooring (names are included). Vertical arrows and thin solid lines in period events Cameron, Darwin, and Ekman (3) are shown in order to highlight the coupling process.
Citation: Journal of Physical Oceanography 48, 1; 10.1175/JPO-D-17-0018.1
Names, eastern and western diameter (km), and history of the Loop Current eddies in the Gulf of Mexico obtained from AVISO data analysis for the period October 2008 through October 2013 and supported online (by http://www.horizonmarine.com). The eastern and western diameters are calculated for the time when the LCE sheds from the Loop Current and the western rim reached the mooring position, respectively. (Loop Current lifetime is directly from http://www.horizonmarine.com.)
It is evident that the WGoM surface circulation is strongly influenced by the arrival of LCEs since more than 55% of the total observation period currents at LMP3500 can be connected to an LCE crossing the mooring. A closer look at Fig. 4 suggests that surface and deep meridional currents are related. Typically, the surface intensification of the northward current (~0.4 m s−1) associated with the western edge of the LCEs is out of phase with the northward current intensification at 2000-m depth (~0.1 m s−1). This occurs for almost all crossing LCEs, being more evident for Cameron, Darwin, and Ekman (3) cases (see Fig. 4). By contrast, surface and deep southward current maxima are mostly in phase producing a vertically coherent southward signal of the flow similar to that reported in Kolodziejczyk et al. (2012). However, in several events the signal is not clear because of the way the LCE translates across the moorings.
Figure 5 shows a stick-plot diagram of currents at different depths for the moorings LMP3500 and ARE3500. The arrival of the surface-intensified LCEs at the moorings are usually associated with what appears to be a deep dipolar structure (the dashed box highlights the process with vertical arrows indicating preferential meridional currents at 2000-m depth for the first three events at LMP3500 mooring). Notice, however, that the signature of a well-defined deep dipole below an upper LCE is not always clear. The deep dipole axis (region in between both deep structures) results in a current directed southward, and it is approximately located beneath the core of the LCE. Hence, deep anticyclonic (cyclonic) circulation precedes (follows) the arrival of the LCE to the mooring. The same process can be observed in both moorings. However, because of the cross trajectories of the LCEs through the mooring position, sometimes there are considerable differences. One of the differences consists in the magnitude of the southward (northward) velocity, being consistently higher (lower) for mooring ARE3500.
Stick diagram of currents for selected depths between surface and 2000-m depth measured at the moorings LMP3500 and ARE3500 between October 2008 and October 2013. The gray sticks are currents at the surface obtained from AVISO (http://www.aviso.altimetry.fr). The dashed boxes and vertical arrows at the upper panel (ARE3500 mooring) for events Cameron, Darwin, and Ekman (3) highlight the deep dipole signal at 2000-m depth.
Citation: Journal of Physical Oceanography 48, 1; 10.1175/JPO-D-17-0018.1
Figures 6 and 7 show the time evolution of the meridional velocity as a function of depth for the first two crossing events (LCEs Cameron and Darwin, respectively) over moorings LMP3500 and ARE3500. For clarity, four snapshots of the geostrophic flow field from AVISO together with the velocity vectors at 100- and 2000-m depth are included in these figures. There are significant differences between these two events: LCE Cameron (Fig. 6) is slower, has a more westward trajectory (Fig. 3), and shallower/weaker surface-trapped velocity cores (only intensified in the upper 250 m) in comparison with LCE Darwin (Fig. 7), which is smaller, faster, and has a more southwestward trajectory with a deeper vertical extent of the surface velocity cores (~500 m). Cameron was slightly less intense than Darwin with maximum near-surface speeds of 0.4 and 0.5 m s−1, respectively (Fig. 4). A northward current in the deep layers of ~0.1 m s−1 occurs ~10–20 days before the northward intensification of the current in the upper layers associated with the arrival of the western edge of the LCEs (Figs. 6, 7). Following this deep northward jet, the meridional current reverses and flows southward (|υ| ~ 0.1 m s−1), suggesting the presence of a deep anticyclone. The center of this deep anticyclone (identified by the zero isotach in Figs. 6 and 7 on 23 December 2008 and 20 July 2009, respectively) occurs approximately beneath the near-surface maximum velocity core of the western rim of the LCEs.
Time series of the meridional component of the horizontal velocity v as a function of depth at moorings (top) LMP3500 (23.38°N, 95.14°W) and (bottom) ARE3500 (22.50°N, 95.02°W) during the crossing event of LCE Cameron. The υ = 0 m s−1 isotach is shown to identify different structures. The data were filtered using a 7-day Lanczos low-pass filter. Between the time series snapshots are shown at selected times (as labeled in figure) of the geostrophic flow field from AVISO (gray arrows) together with the instantaneous velocity measured at 100 m at the mooring (black arrow) and at 2000 m (red arrows). The red dot indicates the center of the LCE at that time and the discontinuous red line joints the LCE center position between December 2008 and March 2009.
Citation: Journal of Physical Oceanography 48, 1; 10.1175/JPO-D-17-0018.1
As in Fig. 6, but for the Loop Current eddy Darwin crossing event.
Citation: Journal of Physical Oceanography 48, 1; 10.1175/JPO-D-17-0018.1
When the eastern edge of the LCE (southward flowing part) crosses the moorings, a vertically southward coherent signal in the meridional velocity component (|υ| > 0.1 m s−1) is evident. These vertically coherent jets have been reported previously in the region between surface-intensified anticyclonic–cyclonic structures (Kolodziejczyk et al. 2012). Following this vertically coherent signal, the deep current reverses its direction again to become a northward deep flow behind the rear part of the LCE. This change of sign of the deep currents is presumably the expression of a deep cyclone. The events just described are especially evident during the crossing of LCE Cameron over mooring LMP3500. The second reversal of the deep flow is more diffuse in the case of LCE Darwin. However, analysis of the full water column vector time series of all the crossing events shown in Fig. 5 reveals a similar pattern: a deep anticyclone preceding the western part of the LCEs and a deep cyclone or, at times, just a deep southward current behind the eastern part of the LCEs. Either scenario depends on how the LCE crossed the moorings. Only a few eddy cores passed over the moorings and most of them crossed peripherally.
To better identify the similarities and differences in all crossing events, time series of the meridional velocity at mooring LMP3500 are plotted separately in Fig. 8. Two different kinds of events result from the analysis: (i) crossing events lasting ~90 days (fast-propagating LCEs; events 2, 3, and 6) and (ii) crossing events lasting ~110 days (slow-propagating LCEs; events 1, 4, 5, 7, and 8). The trajectory and history of LCE Jumbo (event 9) was very different from the other LCEs (events 1 to 8). Jumbo was a very large coherent structure and persisted for more than 200 days near the moorings LMP3500 and ARE3500. The crossing events 3, 4, and 5 are associated with the passage of the same LCE Ekman over the mooring LMP3500. Ekman was initially a very large and fast-propagating LCE (crossing event 3) that later interacted with the continental upper slope and crossed mooring LMP3500 two more times (crossing events 4 and 5). Obviously there is a relation between the duration of the crossing events and the size and translation speed of the LCEs. The two largest LCEs (Jumbo and Ekman) were the slowest and thus the longest crossing events. The translation speed of the LCEs is discussed later in this section.
Isolated time series of the meridional component of the horizontal velocity υ for the nine crossing events of the Loop Current eddies over the mooring LMP3500 (see Fig. 3). Each independent crossing event time series is labeled and colored following Fig. 3. Crossing events 2, 3, and 6 (1, 4, 5, 7, and 8) lasting ~90 (~110) days are considered fast (slow). Crossing event 9 was very different from the others (1 to 8).
Citation: Journal of Physical Oceanography 48, 1; 10.1175/JPO-D-17-0018.1
The composites for the two types of crossing events, fast and slow, are shown in Fig. 9. For the slow crossing events (Fig. 9a), the time lag between the maximum upper- (surface and 100 m) and deep-ocean (2000 m) meridional velocity northward is ~10–20 days (deep flow leading as shown on the sketch in the upper panel). For the eastern rim crossing, the maximum deep meridional velocity southward shows two minima ~30 days apart, the second one being mostly in phase with the upper-ocean meridional southward maximum, and a coherent southward flow signal throughout the full water column can be perceived. This result is interpreted in terms of the deep dipole axis. For the fast crossing events (Fig. 9b), the time lag between the maximum upper- and deep-ocean meridional velocity northward is ~10–20 days. There are negligible deep-ocean currents when the upper northward currents are maximum and the flow gets coherently arranged in a southward jet at the rear of the fast-propagating LCEs. As in the slow-propagating case, the two deep southward velocity minima below the eastern rim are also present. Interestingly, the fast-propagating LCEs lead to a weaker upper and deep southward current (|υ| ~ 0.7 m s−1) at the mooring in contrast to the slow-propagating LCEs that lead to stronger southward currents (|υ| ~ 1 m s−1). Notice that the deep southward current has two minima that suggest horizontal vacillations of the deep structures, presumably because of their interaction with the continental slope. Similar results are obtained using ARE3500 mooring data (not shown).
Time series of normalized meridional component of the horizontal velocity |υ/υmax| for the composite of (a) the slower (1,4, 5, 7, and 8 in Fig. 8) and (b) the faster (2, 3, and 6 in Fig. 8) crossing events of Loop Current eddies over the mooring LMP35000. The composites are computed at three different depth levels: at the surface from AVISO (solid gray line) and at 100 m (solid black line) and 2000 m (discontinuous black line) from mooring data. Above each panel a schematic representation of the 2D flow at each layer is shown.
Citation: Journal of Physical Oceanography 48, 1; 10.1175/JPO-D-17-0018.1
Figure 10 upper panel shows the vertical vorticity time series between August and November 2010 at the center of the triangle (LMP3500, EP13500, and EP23650 moorings; see Fig. 2) and four snapshots of AVISO ADT. There was not a typical LCE crossing during this period since the Franklin event never shows a strong eastern rim at the time it crossed the moorings position (see Fig. 3). However, composite images in Fig. 10 show that when the triangle is placed on the east side of a LCE [Ekman (3) in the snapshot for 18 August 2010], the vorticity on the deep layer is positive, and when on the western side of a LCE (Franklin event on snapshots 28 September, 18 October, and 5 November), the vorticity is negative. This result supports that deep vorticity also evolves with the upper-layer conditions, and that is strongly modulated by the presence of LCEs.
Time series of the vertical vorticity component at the center of LMP3500–EP13500–EP23650 mooring triangle after the Ekman (3) crossing and Franklin arrival events period. The hourly velocity data are filtered and smoothed using a Lanczos low-pass filter (48-h day cutoff). In the lower panels four snapshots at selected times (as labeled in figure) of the absolute dynamic topography from AVISO (color) are shown together with the mooring position (black dots). The time labeled in the snapshots is also labeled in the time series (upper panel).
Citation: Journal of Physical Oceanography 48, 1; 10.1175/JPO-D-17-0018.1
b. Deep cyclonic circulation in the western Gulf of Mexico
To get a regional view of the coupling between upper and deep circulation in the WGoM, the data from all the moorings (Fig. 2) are smoothed and interpolated onto a regular grid using optimum interpolation (Bretherton et al. 1976). Scales smaller than the correlation lengths (Lx and Ly) of the covariance function are filtered out in the optimum interpolation process (Le Traon 1990). The data covariance function is assumed to be homogeneous and isotropic so the covariance becomes simply a function of the distance separating the locations of the data and grid points. At 100-m (2000 m) depth, length scales of the data covariance function are determined using 30 (13) moorings and are averaged into 100-km bins and fit to a Gaussian function that yields mean (mesoscale) decorrelation scales of ~148 and ~162 km for the upper and deep zonal velocity and ~171 and ~146 km for the upper and deep meridional velocity, respectively (Table 3).
Decorrelation length scales at 100- and 2000-m depth in the western Gulf of Mexico.
The 5-yr mean circulations at 100- and 1500-m depth in the WGoM obtained from the objectively interpolated maps are shown in Fig. 11. Notice that the deep circulation is shown at 1500 m rather than 2000 m. This level was selected to avoid the effects of the topography on the current at the moorings located along the 2000-m isobath. The large vertical coherence of the flows below 1000-m depth (Hamilton 1990) indicates the results are not too sensitive to the choice of deep level. Figure 11 only shows values where the interpolation error is less than 15%. The mean WGoM circulation at 100 m (Fig. 11a) agrees very well with the mean surface circulation obtained using altimetry (Fig. 3). The mean surface circulation is mainly dominated by the arrival of the westward-propagating LCEs to the central WGoM and the presence of a semipermanent cyclonic circulation in the BoC. Consequently, the mean circulation in the northern (southern) part is anticyclonic (cyclonic) with maximum normalized vertical vorticity ω/f or Rossby number Ro of −0.4 (0.6), with ω being the relative vorticity and f the Coriolis parameter. Both regions of oppositely signed vertical vorticity are separated by a strong westward jet located at 21.5°N, extending zonally from 94°W to the coast. The variance ellipses indicate that the upper-ocean current variability at the shallower moorings (z ≤ 500 m) is anisotropic and follows the bathymetry. In contrast, at the deeper moorings (z ≥ 1000 m), especially in the central WGoM, the current variability is more isotropic (almost circular rather than ellipses). The mean circulation at 1500-m depth is quite different (Fig. 11b). The deep circulation in the WGoM is mostly cyclonic (DeHaan and Sturges 2005; Hamilton et al. 2016). This result is also obtained in the present study, with the exception of the northernmost region (above 24°N; Fig. 11b). Although the mean deep circulation in this particular region is dominated by positive vertical vorticity, the current vectors are divergent: the flow is directed northward (southward) above (below) 24°N. The deep circulation in this region is very complex and not clearly connected with the northern- and southern-central WGoM as observed using RAFOS drifter data (P. Perez-Brunius 2017, personal communication). Another interesting result is that the deep mean cyclonic circulation (ω/f ~ 0.05) is mostly located in the central region of the WGoM and in the most southern region of the BoC (below 20°N), while elsewhere the vertical vorticity is negative or nearly zero. A qualitative comparison between vertical vorticity maps at 100- and 1500-m depth shows a deep region of anticyclonic vorticity beneath the upper ocean where the flow bifurcates around ~(21.5°–22°N, 96°–97°W). The deep mean circulation in this particular region is very complex. For example, the 5-yr average current below 1500 m merges between moorings ARE2000 and LNK2000, with northward (southward) flows at ARE2000 (LNK2000; not shown).
The 5-yr time-averaged (October 2008–October 2013) horizontal distributions of horizontal velocity (gray arrows) at (a) 100- and (b) 1500-m depth obtained from mooring data with superimposed normalized vertical vorticity or Rossby number Ro (color map). The maps are obtained using optimum interpolation (Bretherton et al. 1976). The mean current (black arrows) and variance ellipses at the mooring locations are included as a reference.
Citation: Journal of Physical Oceanography 48, 1; 10.1175/JPO-D-17-0018.1
The horizontal distributions of the Ro in the WGoM at 100- and 1500-m depth, averaged over two different time periods, are shown in Fig. 12. The first period corresponds to the time of maximum upper-ocean northward current associated with the passage of the western edge of the LCEs over the mooring LMP3500. For this particular time composite, the mean Ro map at 100-m depth is negative in the most central WGoM because of the LCEs crossing and is positive in the BoC. At 1500 m, the vorticity map shows cyclonic circulation in the most northern-central WGoM and all the way along the continental slope and anticyclonic circulation in the northeastern part of the BoC. The second time period corresponds to the time of the maximum upper-ocean southward current associated with the passage of the eastern edge of the LCEs over the mooring LMP3500 (i.e., the LCEs are centered in the central WGoM). The vorticity at 100 m (black contours in Fig. 12b) is strongly anticyclonic above 21.0°N and cyclonic below (this vorticity map qualitatively resembles the 5-yr-averaged circulation map; Fig. 11a). At 1500 m, the Ro map (color map in Fig. 12b) is quite different to the previous scenario (color map in Fig. 12a). The anticyclonic vorticity is found below ~21°N in the most eastern part of the BoC and a dipolar structure above 21°N with anticyclonic (cyclonic) vorticity on the upper (lower) continental slope. The change in sign of the deep vorticity is clearly aligned with the LCE’s core at 100-m depth. Notice that the deep dipole axis located at 95.5°–96°W is aligned south–north and just beneath the core of the surface-intensified LCEs.
Time-averaged horizontal distributions of Rossby number Ro at 100-m depth (solid black contours positive Ro, dashed contours negative Ro, contour interval 0.1; Ro = ω/f = ±0.1) and 1500-m depth (color map) during (a) the maximum upper-ocean northward current (western edge of the Loop Current eddy) and (b) the maximum upper-ocean southward current (eastern edge of the Loop Current eddy) obtained from the objectively interpolated daily maps of circulation.
Citation: Journal of Physical Oceanography 48, 1; 10.1175/JPO-D-17-0018.1
Figure 13 shows the dispersion diagram between the propagation of anticyclonic vorticity signals at 100- and 1500-m depth. The translation speed of the vortical structures at upper and deep layers is of similar order of magnitude, but the deep ones propagate consistently slower. For example, LCEs Cameron and Darwin propagated at about 3–3.5 km day−1, approximately 2 times faster than the corresponding deep vortical structures (~1.5 km day−1). This result suggests that the deep vortical structures are deep eddies that are topographically constrained, decreasing their translation speed and probably dissipating like TRWs. Notice that only seven westward-propagating LCEs are presented in Fig. 13 because of unresolved signals for two particular events: upper (lower) LCE Franklin (Galileo and Hadal).
Dispersion diagram for the 100- and 1500-m depth westward-propagating anticyclonic vorticity signal obtained from Hovmöller diagrams of the optimally interpolated data. The gray dashed line is a least squares fit, and the black dashed–dotted line is DLCE = LCE.
Citation: Journal of Physical Oceanography 48, 1; 10.1175/JPO-D-17-0018.1
4. Discussion and conclusions
The coupling between surface (100 m) and deep (1500–2000 m) circulation in the central WGoM and BoC caused by LCEs crossing is analyzed using mooring- and satellite-derived velocity data. Though we cannot fully resolve the horizontal structure of the deep eddies with the available (coarse) horizontal resolution of the mooring array, the results strongly suggest that broad regions of deep anticyclonic (cyclonic) vorticity lead (follow) the arrival of the LCEs in the central WGoM. A simple explanation using potential vorticity in a two-layer ocean model (Cushman-Roisin et al. 1990) argues that the lower (upper) layer is compressed (stretched) at the front (rear) of the translating upper-ocean LCEs. The deep circulation (deeper than 1000 m) is clearly coupled to the life-span of the LCEs; that is, the intensification and decay of the deep circulation is associated with the intensification and decay of the upper ocean caused by the presence of LCEs. Similar results were found by Hurlburt and Thompson (1982) and Sturges et al. (1993) using two-layer and multilayer ocean models of the Gulf of Mexico, respectively. The presence of a deep anticyclonic eddy leading (10–20 days) the upper-ocean LCEs is a very robust result obtained from the study of nine LCE crossings over two deep moorings (LMP3500 and ARE3500). On one hand, a deep northward current is set on the lower continental slope 10–20 days before the upper northward current reaches its maximum (western edge of an LCE). At this moment, the deep current is nearly zero, which is presumably related to the core of the deep anticyclone. On the other hand, the upper southward current associated with the eastern edge of the LCE is phase locked with a deep southward current, which appears to be the deep axis of the dipoles composed by the deep anticyclone and the cyclone behind. This coherent flow structure throughout the water column occurs at the rear part of the eddy (Kolodziejczyk et al. 2011). The presence of the deep cyclone following the upper-ocean LCE to form a deep dipole is suggested from an Eulerian (Figs. 4, 5, 6, 7, 9a,b) and regional point of view (Fig. 12b) and sketched (Fig. 1 and Fig. 9, upper panel). Deep dipoles beneath the LCEs have been identified numerically in several works but, to our knowledge, this is one of the first observational studies describing the vertical coupling between upper and deeper layers in the WGoM.
Using objectively interpolated regional maps of Ro, we found the following main results:
Deep cyclonic circulation along the lower continental slope of the central WGoM (actually the deep GoM in general) has been related to the deep overflow at Yucatan Channel and TRWs flow rectification (e.g., DeHaan and Sturges 2005) based on the results of Mizuta and Hogg (2004), who showed that an incident TRW over a slope induces a mean along-slope bottom flow of ~0.05 to 0.1 m s−1 with shallower water on the right, that is, cyclonic, as a result of the balance between divergence of vertically integrated Reynolds stresses and bottom friction. Possible sources of TRWs generation are various [see Hamilton (2009) for a review], but the ubiquity and intensification of the upper circulation caused by the presence of LCEs and their associated deep circulation features in this region appear to be a very important source too. Sutyrin et al. (2003) and Frolov et al. (2004) model results indicate that the interaction of an LCE with the slope and shelf depends on several factors including orientation, size, and steepness of the slope and shelf. Those studies as well as Welsh and Inoue (2000) and Romanou et al. (2004) suggest deep cyclones associated with LCEs tend to last longer and intensify more than anticyclones in the Modon-like structures that form below the LCEs. Both (TRWs and intensified abyssal plane cyclones) may contribute to the strengthening of the deep mean cyclonic circulation. The persistent deep cyclonic circulation has also been observed in deep RAFOS float trajectories as well (Pérez-Brunius et al. 2018).
A differential westward propagation speed between the upper and deep vortical structures, the latter being sometimes 2 times smaller. This result indicates that the deep eddies are topographically constrained and suggests that the deep eddies interacting with the topography could be an important path for excitation of TRWs and possibly mean flow dissipation.
Although more studies are needed to prove the validity of these results, the upper and deep circulation in the central WGoM is strongly linked, which increases our predictability of the deep circulation by knowing the life-span of the LCEs in the central WGoM.
Acknowledgments
This work was supported by the Petroleos Mexicanos Grant PEP-CICESE 428229851 as a component of the “Medición y Análisis Metoceánico del Golfo de México, Etapa 2009-2014.” The altimeter products were produced by SSALTO/DUACS and distributed by AVISO, with support from CNES (http://www.aviso.altimetry.fr/duacs/). The authors thank Paula Pérez-Brunius for very helpful discussions regarding these observations and two anonymous reviewers for their insightful comments, which substantially improved the manuscript. We also acknowledge the crew of UNAM’s Research Vessel Justo Sierra and its Captain Leobardo Rios for their professionalism and continuous support over several years. The authors gratefully acknowledge financial support from the Consejo Nacional de Ciencia y Tecnología (Cátedra CONACYT-CICESE).
REFERENCES
Bretherton, F. P., D. E. Russ, and C. B. Fandry, 1976: A technique for objective analysis and design of oceanographic experiments applied to MODE-73. Deep-Sea Res. Oceanogr. Abstr., 23, 559–582, https://doi.org/10.1016/0011-7471(76)90001-2.
Chaigneau, A., A. Gizolme, and C. Grados, 2008: Mesoscale eddies off Peru in altimeter records: Identification algorithms and eddy spatio-temporal patterns. Prog. Oceanogr., 79, 106–119, https://doi.org/10.1016/j.pocean.2008.10.013.
Cushman-Roisin, B., E. P. Chassignet, and B. Tang, 1990: Westward motion of mesoscale eddies. J. Phys. Oceanogr., 20, 758–768, https://doi.org/10.1175/1520-0485(1990)020<0758:WMOME>2.0.CO;2.
DeHaan, C. J., and W. Sturges, 2005: Deep cyclonic circulation in the Gulf of Mexico. J. Phys. Oceanogr., 35, 1801–1812, https://doi.org/10.1175/JPO2790.1.
Donohue, K. A., D. R. Watts, P. Hamilton, R. Leben, and M. Kennelly, 2016: Loop Current eddy formation and baroclinic instability. Dyn. Atmos. Oceans, 969, 195–216, https://doi.org/10.1016/j.dynatmoce.2016.01.004.
Elliott, B. A., 1982: Anticyclonic rings in the Gulf of Mexico. J. Phys. Oceanogr., 12, 1292–1309, https://doi.org/10.1175/1520-0485(1982)012<1292:ARITGO>2.0.CO;2.
Frolov, S. A., G. G. Sutyrin, G. D. Rowe, and L. M. Rothstein, 2004: Loop Current eddy interaction with the western boundary in the Gulf of Mexico. J. Phys. Oceanogr., 34, 2223–2237, https://doi.org/10.1175/1520-0485(2004)034<2223:LCEIWT>2.0.CO;2.
Hamilton, P., 1990: Deep currents in the Gulf of Mexico. J. Phys. Oceanogr., 20, 1087–1104, https://doi.org/10.1175/1520-0485(1990)020<1087:DCITGO>2.0.CO;2.
Hamilton, P., 2009: Topographic Rossby waves in the Gulf of Mexico. Prog. Oceanogr., 82, 1–31, https://doi.org/10.1016/j.pocean.2009.04.019.
Hamilton, P., and A. Badan, 2009: Subsurface jets in the northwestern Gulf of Mexico. J. Phys. Oceanogr., 39, 2875–2891, https://doi.org/10.1175/2009JPO4158.1.
Hamilton, P., G. S. Fargion, and D. C. Biggs, 1999: Loop Current eddy paths in the western Gulf of Mexico. J. Phys. Oceanogr., 29, 1180–1207, https://doi.org/10.1175/1520-0485(1999)029<1180:LCEPIT>2.0.CO;2.
Hamilton, P., A. Bower, H. Furey, R. Leben, and P. Pérez-Brunius, 2016: Deep circulation in the Gulf of Mexico: A Lagrangian study. Bureau of Ocean Energy Management Outer Continental Shelf Study Rep. BOEM 2016-081, 289 pp.
Hurlburt, H. E., and J. D. Thompson, 1982: The dynamics of the Loop Current and shed eddies in a numerical model of the Gulf of Mexico. Hydrodynamics of Semi-Enclosed Seas: Proceedings of the 13th International Liege Colloquium on Ocean Hydrodynamics, J. C. J. Nihoul, Ed., Elsevier Oceanography Series, Vol. 34, Elsevier, 243–297.
Kolodziejczyk, N., J. Ochoa, J. Candela, and J. Sheinbaum, 2011: Deep currents in the Bay of Campeche. J. Phys. Oceanogr., 41, 1902–1920, https://doi.org/10.1175/2011JPO4526.1.
Kolodziejczyk, N., J. Ochoa, J. Candela, and J. Sheinbaum, 2012: Observations of intermittent deep currents and eddies in the Gulf of Mexico. J. Geophys. Res., 117, C09014, doi:10.1029/2012JC007890.
Le Traon, P.-Y., 1990: A method for optimal analysis of fields with spatially variable mean. J. Geophys. Res., 95, 13 543–13 547, https://doi.org/10.1029/JC095iC08p13543.
Mizuta, G., and N. G. Hogg, 2004: Structure of the circulation induced by a shoaling topographic wave. J. Phys. Oceanogr., 34, 1793–1810, https://doi.org/10.1175/1520-0485(2004)034<1793:SOTCIB>2.0.CO;2.
Nof, D., 1981: On the β-induced movement of isolated baroclinic eddies. J. Phys. Oceanogr., 11, 1662–1672, https://doi.org/10.1175/1520-0485(1981)011<1662:OTIMOI>2.0.CO;2.
Oey, L.-Y., 2008: Loop Current and deep eddies. J. Phys. Oceanogr., 38, 1426–1449, https://doi.org/10.1175/2007JPO3818.1.
Oey, L.-Y., T. Ezer, and H.-C. Lee, 2005: Loop Current, rings and related circulation in the Gulf of Mexico: A review of numerical models and future challenges. Circulation in the Gulf of Mexico: Observations and Models, Geophys. Monogr., Vol. 161, Amer. Geophys. Union, 31–56, https://doi.org/10.1029/161GM04.
Pérez-Brunius, P., P. García-Carrillo, J. Dubranna, J. Sheinbaum, and J. Candela, 2013: Direct observations of the upper layer circulation in the southern Gulf of Mexico. Deep-Sea Res. II, 85, 182–194, https://doi.org/10.1016/j.dsr2.2012.07.020.
Pérez-Brunius, P., H. Furey, A. Bower, P. Hamilton, J. Candela, P. García-Carrillo, and R. Leben, 2018: Dominant circulation patterns of the deep Gulf of Mexico. J. Phys. Oceanogr., https://doi.org/10.1175/JPO-D-17-0140.1, in press.
Romanou, A., E. P. Chassignet, and W. Sturges, 2004: Gulf of Mexico circulation within a high-resolution numerical simulation of the North Atlantic Ocean. J. Geophys. Res., 109, C01003, https://doi.org/10.1029/2003JC001770.
Sturges, W., 2005: Deep-water exchange between the Atlantic, Caribbean, and Gulf of Mexico. Circulation in the Gulf of Mexico: Observations and Models, Geophys. Monogr., Vol. 161, Amer. Geophys. Union, 263–278, https://doi.org/10.1029/161GM019.
Sturges, W., J. Evans, W. Holland, and S. Welsh, 1993: The separation of warm-core rings in the Gulf of Mexico. J. Phys. Oceanogr., 23, 250–268, https://doi.org/10.1175/1520-0485(1993)023<0250:SOWCRI>2.0.CO;2.
Sutyrin, G. G., G. D. Rowe, L. M. Rothstein, and I. Ginis, 2003: Baroclinic eddy interactions with continental slopes and shelves. J. Phys. Oceanogr., 33, 283–291, https://doi.org/10.1175/1520-0485(2003)033<0283:BEIWCS>2.0.CO;2.
Vázquez De La Cerda, A. M., R. O. Reid, S. F. DiMarco, and A. E. Jochens, 2005: Bay of Campeche circulation: An update. Circulation in the Gulf of Mexico: Observations and Models, Geophys. Monogr., Vol. 161, Amer. Geophys. Union, 279–294, https://doi.org/10.1029/161GM20.
Vidal, V. M. V., F. V. Vidal, and J. M. Pérez-Molero, 1992: Collision of a Loop Current anticyclonic ring against the continental shelf slope of the western Gulf of Mexico. J. Geophys. Res., 97, 2155–2172, https://doi.org/10.1029/91JC00486.
Vidal, V. M. V., F. V. Vidal, E. Meza, J. Portilia, L. Zambrano, and B. Jaimes, 1999: Ring-slope interactions and the formation of the western boundary current in the Gulf of Mexico. J. Geophys. Res., 104, 20 523–20 550, https://doi.org/10.1029/1999JC900118.
Vukovich, F. M., 2007: Climatology of ocean features in the Gulf of Mexico using satellite remote sensing data. J. Phys. Oceanogr., 37, 689–707, https://doi.org/10.1175/JPO2989.1.
Welsh, S. E., and M. Inoue, 2000: Loop Current rings and the deep circulation in the Gulf of Mexico. J. Geophys. Res., 105, 16 951–16 959, https://doi.org/10.1029/2000JC900054.
