1. Introduction
Eddies are ubiquitous features of the Caribbean Sea circulation where they play an important role on dynamics and environmental conditions. They contribute to offshore advection and dispersion of larvae (Baums et al. 2006), river plumes (Corredor et al. 2004), or buoyant matter and pollutants (Chérubin and Richardson 2007). They also participate in the dispersion of cool and nutrient-rich upwelled waters from the Guajira upwelling system (Andrade and Barton 2005). They are known to influence the eddy-shedding events of the Loop Current in the Gulf of Mexico (Murphy et al. 1999; Candela et al. 2002; Oey et al. 2003; Athie et al. 2012). The Caribbean Sea is also known to be a major pathway for the return flow of the Atlantic meridional overturning circulation (Rhein et al. 2005; Kirchner et al. 2008) and for the wind-driven flow of the subtropical gyre (Johns et al. 2002). Caribbean eddies are expected to play a key role in the transport and mixing of these waters.
A horizontal map of sea surface height (SSH) in the Caribbean Sea and associated surface currents is shown in Fig. 1. It illustrates that 1) eddies are ubiquitous features of the Caribbean circulation, 2) the more energetic eddies are anticyclones, and 3) the eddies grow in size and strength as they move westward. Several studies suggest or provide evidence that the instability of the mean Caribbean Current is responsible for the production and westward growth of the Caribbean eddies (Carton and Chao 1999; Andrade and Barton 2000; Richardson 2005; Jouanno et al. 2009; Alvera-Azcárate et al. 2009). However, others suggest that wind forcing could contribute to eddy generation and growth in the central Caribbean (Andrade and Barton 2000; Oey et al. 2003). By contrast, other numerical model studies (Murphy et al. 1999; Barnier et al. 2001) propose that there is a strong influence of the rings produced by the North Brazil Current (NBC) retroflection, a hypothesis consistent with process studies on eddy flux through island chains (Simmons and Nof 2002). The role and behavior of the NBC rings near the Lesser Antilles remains unclear, but they are known to influence the circulation east and west of this island chain (Carton and Chao 1999; Richardson 2005). Since their presence near the Lesser Antilles is ubiquitous, they appear to be linked with most of the Caribbean eddies. In Jouanno et al. (2009), a comparison between different regional numerical experiments indicate that mean eddy kinetic energy (EKE) and eddy population in the Caribbean Sea are not substantially different in the absence or presence of NBC rings. The authors conclude that the role of the NBC rings is mostly to act as a finite perturbation for the instability of the mean flow and to influence the frequency of formation of the Caribbean eddies.
Model domain and snapshot of sea surface height (SSH) and surface currents at day 15 May 2007. It illustrates the interaction of a NBC ring with the Barbados (located at 13°N, 58°W) and the occurrence of large anticyclonic eddies in the region from the Lesser Antilles to the Nicaraguan coast. Surface current vectors are only shown for speeds greater than 0.15 m s−1.
Citation: Journal of Physical Oceanography 42, 11; 10.1175/JPO-D-12-048.1
The seasonal and interannual variability of the Caribbean eddies have been poorly documented. Analyzing 15 months of altimetry data, Andrade and Barton (2000) found an enhancement of the eddy activity in the central Caribbean during July–October. They suggest that this maximum could be due to direct action of both wind stress and wind stress curl, together with a strong meridional surface gradient of salinity caused by intense precipitation during summer. In the Eastern Caribbean, the analysis of 10 looping drifters in Richardson (2005) suggests a maximum of anticyclone population in September–November and a minimum in March.
Both the atmospheric conditions over the Caribbean Sea and the large-scale oceanic systems connected with the circulation in the Caribbean basin show significant seasonal variations. The intense and persistent easterly winds of the Caribbean low-level jet (CLLJ) are known to vary semiannually, with two maxima in the summer and winter and two minima in the fall and spring (Wang 2007). Johns et al. (2002) found that the inflow through the Windward Islands passages has a large seasonal cycle, with a maximum in June and a minimum in September–October. Since the growth of the eddies appears to be related to the instability of the mean flow, one could expect that the variability in size and strength of the Caribbean eddies are related to variability of the speed, transport, and shear of the mean flow. In the Atlantic Ocean roughly five to nine rings per year are shed by the NBC retroflection at nearly regular intervals, depending on whether subsurface rings are considered or not (e.g., Barnier et al. 2001; Johns et al. 2003). Johns et al. (2003) do not find a seasonal signal in the surface rings generation rate but they show that rings are deeper and larger in fall and early winter and relate that to the maximum strength and depth of the NBC retroflection in August–September. Although results of Jouanno et al. (2009) suggest that the advection of NBC rings is not a necessary condition to produce energetic Caribbean eddies, we cannot discard the possibility that the seasonal variability in the energy of the perturbations impinging the Lesser Antilles could modulate the EKE in the Caribbean basin.
The aim of the present study is to examine the seasonal and interannual variations of the Caribbean mesoscale eddies using both altimetry data and a high-resolution numerical model. The connection between eddy energy and local or remote winds, NBC rings variability, and the instability of the mean flow is investigated. The paper is organized as follows. A description of model and data is given is section 2. Section 3 describes the seasonal variations of the eddy activity and investigates the mechanisms underlying this variability. Section 4 discusses the role of the incoming NBC rings, and section 5 focuses on the interannual fluctuations of the eddy activity in the Colombia Basin (Fig. 2). Finally, section 6 provides a discussion and summary of the results.
Description of the Caribbean basins and area used to build the longitude–time diagram in Fig. 4. The thin lines represent the 100 and 1000 m isobaths.
Citation: Journal of Physical Oceanography 42, 11; 10.1175/JPO-D-12-048.1
2. Model and data
A high-resolution regional model encompassing the Caribbean Sea and the Gulf of Mexico has been developed for this study. The numerical code is the oceanic component of the Nucleus for European Modeling of the Ocean program (NEMO) (Madec 2008). It solves the three-dimensional primitive equations in spherical coordinates discretized on a C grid and fixed vertical levels (z coordinate). The regional grid extends from 5° to 33°N, 100° to 55°W as illustrated by the snapshot of surface currents in Fig. 1. Horizontal resolution of the model is 
The model is forced at its boundaries using a radiative open boundary condition with outputs from the global interannual experiment ORCA025-MJM95 developed by the DRAKKAR team (Barnier et al. 2006). These open boundary conditions radiate perturbations out of the domain and relax the model variables to 5-day averages of ORCA025-MJM95 outputs. Details of the method are given in Tréguier et al. (2001). At the surface, the atmospheric fluxes of momentum, heat, and freshwater are provided by bulk formulae (Large and Yeager 2004). The model is forced with the European Centre for Medium-Range Weather Forecasts Interim Re-Analysis (ERA-Interim) (3-h fields of wind, atmospheric temperature, and humidity; daily fields of longwave and shortwave radiation and precipitation). The shortwave radiation forcing is modulated by a theoretical diurnal cycle (Bernie et al. 2007). A monthly climatological runoff based on the dataset of Dai and Trenberth (2002) is prescribed near the river mouths as a surface freshwater flux. The regional model was initialized with temperature and salinity fields from the global experiment on 31 December 1990 and was integrated over the period 1991–2009. Five-day averages from 1993 to 2009 are used in the present study. Two-year spinup is consistent with baroclinic Rossby wave transit time to establish some kind of equilibrium at the latitudes of the simulation.
In addition to this reference experiment, an experiment forced with climatological lateral boundary conditions has been carried out. In this experiment, referred to as CLIM in the following sections, the boundary data were built as the climatological seasonal cycle of the boundary data used to force the reference experiment. The use of climatological lateral forcing has two direct consequences on the simulation: 1) the interannual variability at the boundaries is removed and 2) there are no more NBC rings entering the regional model (because the model domain does not encompass the NBC retroflection region).
Numerical results are compared with geostrophic velocities based on Archiving, Validation, and Interpretation of Satellite Oceanographic data (AVISO) altimetry data for the period January 1993–December 2009. Model and altimetry data are decomposed as the sum of a low-passed component and anomaly with respect to this low-passed component as, u = ULP + u′, where u is a model or altimetry variable, ULP the low-passed values of u, and u′ the anomaly. Data were filtered with a 120-day running mean. The characteristic period of the Caribbean eddies is between 50 and 100 days (Jouanno et al. 2008). The energy of these eddies is known to dominate the energy spectra of the surface currents in the region (e.g., Jouanno et al. 2008), so such decomposition allows to properly separate the signal of the mesoscale eddies from that of the seasonal and interannual currents. The EKE is computed from velocity anomalies as (½)(u′2 + υ′2). It is representative of the energy of the mesoscale eddy field. The kinetic energy of the low-frequency currents is based on low-pass velocities, as mean kinetic energy (MKE) = (½)(ULP2 + VLP2).
3. Seasonal cycle of EKE
Surface EKE in the Caribbean Sea, computed from altimetry observations and averaged between 1993 and 2009, is shown in Fig. 3a. The westward increase of EKE from the Lesser Antilles is the signature of the strengthening of the Caribbean eddies when traveling westward. The eddies reach their major strength in the Colombia Basin. Afterward, they lose their energy along the coasts of Nicaragua. The mean energy of the low-frequency currents (MKE) presents a different spatial distribution (Fig. 3c); it is intense at the southern passages of the Lesser Antilles, north of the Guajira Peninsula, and through the Chibcha Channel. Downstream of these regions MKE decreases and EKE increases, in agreement with previous findings that the energy of the caribbean eddies is mostly provided by release of energy of the mean flow through baroclinic and barotropic instability processes (e.g., Jouanno et al. 2009). The strength and spatial distribution of surface EKE and MKE in the model (Figs. 3b,d) are in good agreement with satellite observations (Figs. 3a,c).
Mean eddy kinetic energy (EKE, m2s−2) and mean kinetic energy (MKE, m2s−2) of the low-passed surface currents, computed from (a),(c) altimetry-derived surface velocities and (b),(d) model surface velocities. Low-pass filtered currents were computed using a 120-day running mean. Data from January 1993 to December 2009 are used for both datasets.
Citation: Journal of Physical Oceanography 42, 11; 10.1175/JPO-D-12-048.1
Composite seasonal cycles of EKE and MKE along the Colombia and Venezuela Basins (Fig. 2) are shown in Fig. 4. The longitude–time diagrams are constructed with EKE and MKE averaged between 13° and 17°N, as indicated in Fig. 2. Seasonal variation of EKE in the Colombia Basin (west of 70°W) depicts one peak in February–March and another in August–October (Fig. 4a). In both model and observations, the fall peak of EKE is more energetic than the spring peak (Figs. 4a and 4c). By contrast, EKE in the Venezuela region (between 70° and 62°W) and east of the Lesser Antilles (east of 62°W) follows an annual cycle. In both Colombia and Venezuela Basins, this seasonality has a clear link with the energy of downstream low-frequency currents one month before. First, the two peaks of EKE in the Colombia Basin follow in space and time the two peaks of MKE occurring north and downstream of the Guajira Peninsula (between 75° and 70°W in Figs. 4b,d) during January–February and June–August. Second, the seasonal peak of EKE in the Venezuela Basin follows the annual peak of MKE occurring through the Lesser Antilles (at 62°W in Figs. 4b,d) in April–July.
Comparison of model and altimetry energy seasonal cycles in the Caribbean Sea: (a),(b) surface EKE (m2 s−2) and (b),(d) surface MKE (m2 s−2). A meridional average of EKE and MKE is performed between 13° and 17°N for both altimetry and model data. Longitude–time diagrams are based on monthly composites computed from 5-day model data and 7-day altimetry data over the period 1993–2009. The area used to build the longitude–time diagrams is shown in Fig. 2.
Citation: Journal of Physical Oceanography 42, 11; 10.1175/JPO-D-12-048.1
Composite seasonal cycles of the three terms mentioned above are shown in Figs. 5a,b. The data are averaged between the ocean surface and 150 m. This range of depths is chosen since most of the barotropic and baroclinic energy conversions occur in this part of the water column, as shown in Jouanno et al. (2009). The Colombia Basin presents the highest values of baroclinic conversion for the whole Caribbean (Fig. 5b). High values occur all over the year, consistent with the continuous westward increase of EKE throughout the year between 75° and 70°W (Figs. 4a,c). In terms of seasonal variability, BC presents a semiannual cycle in the Colombia Basin with a first maximum in January–February and a second in July–September. This suggests that the enhanced release of available potential energy during these periods contributes to enhanced eddy growth. The term of barotropic conversion also peaks twice a year at the entrance of the Colombia Basin (near 72°W in Fig. 5a). It is thus expected to contribute to the semiannual cycle of the eddy energy. The summer maximum of BT is stronger than the winter maximum. This could contribute to the seasonal asymmetry of the observed EKE peaks. The different energy transfer terms averaged over the Colombia Basin (Fig. 6c) confirm these findings and suggest that baroclinic conversion is the dominant source of energy for the eddy field. In addition, it is shown that the term of advection of EKE is negative throughout the year except in July (Fig. 6c). This means that there is more EKE radiated out of the averaging area (12°–17°N, 78°–72°W) than EKE advected into this area. This is in agreement with the idea that the Colombia Basin is a region of production of EKE. In the Venezuela Basin, both energy conversion terms follow an annual cycle, the highest values being reached during April–July at the entrance of the basin (between 62° and 60°W). This is coherent with the summer increase of EKE in this part of the Caribbean Sea.
Seasonal cycles of (a) barotropic conversion of EKE, (b) baroclinic conversion of EKE, (c) advection of EKE by the turbulent and low-frequency currents, (d) horizontal velocity shear, (e) vertical velocity shear, and (f) zonal wind stress . In (a)–(c) positive means a transfer of energy toward the turbulent field. Black contours in (f) indicate values of westward zonal wind stress equal to 0.1 N m−2. As in Fig. 4, longitude–time diagrams are based on monthly composites computed from 5-day model data over the period 1993–2009 averaged between 13° and 17°N. Energy transfer terms and velocity shears are averaged over the upper 150 m of the water column.
Citation: Journal of Physical Oceanography 42, 11; 10.1175/JPO-D-12-048.1
Composite seasonal cycles of energy and energy transfer terms averaged over the Colombia Basin (12°–17°N, 78°–72°W) between the surface and 150 m depth: (a) model (black) and altimetry (gray) EKE (m2 s−2), (b) model (black) and altimetry (gray) MKE (m2 s−2), and (c) energy transfer terms (kg m−1 s−3) including the barotropic conversion of EKE (green), the baroclinic conversion of EKE (black), and the advection of EKE by the turbulent and low-frequency currents (blue).
Citation: Journal of Physical Oceanography 42, 11; 10.1175/JPO-D-12-048.1
The seasonal variations of the energy conversion terms clearly relate to variations of the vertical and horizontal velocity shear of the low-frequency currents shown in Figs. 5d and 5e. The horizontal velocity shear is computed as 
Maps of surface currents in Fig. 7 show that the semiannual variation of MKE at the Guajira Peninsula is mainly due to the semiannual acceleration of the coastal jet, with peaks in January–February and July–August. A meridional section of the zonal currents north of the Guajira Peninsula illustrates the strong vertical and horizontal shears associated with this surface intensified coastal jet (Fig. 8). At the subsurface the Caribbean Countercurrent contributes to enhance the vertical shear near the coast throughout the year. There is no clear evidence that the variability of the subsurface current would modulate the vertical and horizontal shears. Between 65° and 60°W the flow entering the Venezuela Basin by the southern Passages also takes the form of a jet (Fig. 7). This jet has a significant meridional component and its strength is modulated by an annual cycle.
Seasonal low-frequency surface currents (vectors) and anomalies from the annual mean (colors, m s−1). The corresponding fields are shown by periods of two months from January–February to November–December. Only vectors representing speeds above 0.15 m s−1 are shown.
Citation: Journal of Physical Oceanography 42, 11; 10.1175/JPO-D-12-048.1
Seasonal cycle of zonal currents (colors) and density (contours) between the Guajira Peninsula and Hispaniola: positive means eastward flow. Currents and density were low-passed filtered (120-day running mean) and zonally averaged between 72.5° and 70°W. Bold black lines are zero velocity contours.
Citation: Journal of Physical Oceanography 42, 11; 10.1175/JPO-D-12-048.1
The seasonal cycle of the flow entering the Caribbean Sea by the southern passages of the Lesser Antilles is consistent with the description of the seasonal variations of the Caribbean inflow in Johns et al. (2002). These authors found a maximum of Caribbean inflow in spring/summer and a minimum in fall, with most of the seasonal variation,~3 Sv (Sv ≡ 106 m3 s−1), concentrated in the Grenada Passage. They argue that this seasonal cycle is consistent with the meridional migration of the intertropical convergence zone and associated changes of wind stress curl in the region. The annual peak of surface current speed at Grenada Passage in May–June is also in agreement with in situ observations by Rhein et al. (2005).
The semiannual cycle of the jet north of the Guajira Peninsula, already seen in horizontal maps of surface currents (Fig. 7) and vertical section of zonal currents (Fig. 8), is highlighted in a latitude–time diagram of surface currents averaged between 72.5° and 70°W (Fig. 9c). Its semiannual cycle contrasts with the dominant annual cycle in the Venezuela Basin. This discrepancy between seasonal cycles suggests that the variability of the coastal jet is forced locally as part of the Guajira upwelling system. At first order, the seasonal cycle of the coastal jet is explained by the strength of the local wind. First, both wind and current exhibit a semiannual cycle. Second, the maximum of surface currents occurs at the latitudes of the core of the CLLJ, between 12° and 14°N. Third, the winds of the CLLJ are strong and parallel to the coast in this region (Fig. 9a). The accelerated westward winds are expected to force an anomalous offshore Ekman drift. The signature of an anomalous rising of the isopycnals at the coast is not clear in Fig. 8. But the SSH at the Guajira (contours in Fig. 9c) show coastal depressions during January–February and June–August, which indicate that the strengthening of the surface coastal current is in geostrophic balance. The mean wind stress curl is cyclonic in this region (Fig. 9b) and thus also favors the upwelling and associated westward surface jet. Nevertheless, the wind stress curl at the Guajira Peninsula varies with an annual cycle, which cannot explain the semiannual cycle of the coastal jet (Fig. 9e). But, note that the maximum of wind stress curl coincides in time with a maximum of the sCC in July–August (see Fig. 7d). The large-scale broadening and strengthening of the sCC is probably linked to large-scale seasonal variations of the wind stress curl in the Caribbean and western tropical Atlantic. The sCC maximum is expected to contribute to the asymmetry of the semiannual peak of barotropic conversion and EKE in the Colombia Basin.
Mean (a) wind stress and (b) wind stress curl in the Caribbean Sea and latitude–time composite seasonal cycles of (c) surface currents, (d) wind stress, and (e) wind stress curl at the Guajira Peninsula. Isocontours of SSH (cm) ranging from −18 to 18 cm are also shown in (c). Black lines in (e) are isocontours of zero wind stress curl. Latitude–time diagrams in (c)–(e) are built with data averaged over the box drawn in (b) (12°–18°N, 72.5°–70°W).
Citation: Journal of Physical Oceanography 42, 11; 10.1175/JPO-D-12-048.1
4. Role of the NBC rings
Longitude–time diagrams of surface EKE suggest a propagation of EKE in summer and fall from the east of the basin to the west (Figs. 4a,c). This suggests that during this period anomalously energetic eddies could be advected from the Lesser Antilles/Venezuela Basin and would contribute to the summer peak of EKE in the Colombia Basin. In this section we further examine this issue.
On one hand, there are some elements that suggest the NBC rings could play a significant role on the Caribbean EKE variability. Horizontal maps of EKE anomalies are shown in Fig. 10. Each panel represents two-month average of anomalies taken from the annual mean. Variability of North Equatorial Counter Current [this pattern is referenced as (A) in Fig. 10] and variability of the NBC retroflection (B) are more intense during fall, consistent with Barnier et al. (2001) or Johns et al. (2003). A band of maximum variability along the Brazilian coast starts to extend northwestward during winter and shows a peak in March–April (C). This band is related to the presence of more energetic rings in the region from February to June. In May–June, the high EKE band has been split into two parts. A first part engulfs the Barbados Island (D) and will remain there until July–August (see also Fratantoni and Richardson 2006). The second part is seen as the high EKE band in Venezuela Basin (E), signature of energetic Eastern Caribbean eddies formed near the southernmost Lesser Antilles passages. In September–October, the strength and extent of the EKE tongue in the Colombia Basin is maximum (G). The westward advection of more energetic perturbations, which occur in the Venezuela Basin some months earlier (F during July–August), could contribute to this peak of energy. Maps of EKE anomalies from model data are not shown but present similar patterns at the same periods suggesting this seasonality is a robust feature of the regional variability. The westward movement of EKE maxima from the retroflection region to the Colombia Basin (roughly 3500 km) in about eight months corresponds to a propagation speed of 15 cm s−1, consistent with estimations of translation speeds for NBC rings (~14.5 cm s−1, Johns et al. 2003) and Caribbean anticyclones (13.5 cm s−1, Richardson 2005).
Seasonal cycle of anomalies of EKE from the annual mean (colors, m2 s−2). Low-passed surface currents are shown with vectors. EKE and currents are based on 1993–2009 geostrophic velocity derived from altimetry. Current vectors are shown when current speeds are greater than 0.1 m s−1. Bold vectors mean that the current speed is above the local annual mean speed. See text for explanation of letter labels.
Citation: Journal of Physical Oceanography 42, 11; 10.1175/JPO-D-12-048.1
On the other hand, our numerical results suggest that the link between the level of EKE in the Caribbean Sea and the advection of more energetic NBC rings is not straightforward. First, the term ADV (Figs. 5c and 6c) shows that the contribution of the advection of EKE to the peak of eddy activity in the Colombia Basin is much lower than the contributions of barotropic and baroclinic conversion. Second, the seasonal EKE in CLIM experiment resembles that of the reference experiment (cf. Fig. 4c and Fig. 11a). In the CLIM experiment, there are no intraseasonal perturbations entering the model grid at 55°W, so this eddy energy cannot have been advected from the Atlantic.
Seasonal cycle of EKE and MKE as in Figs. 4c,d but for experiment CLIM. In this experiment the model is forced at the lateral boundaries with climatological fields.
Citation: Journal of Physical Oceanography 42, 11; 10.1175/JPO-D-12-048.1
The largest differences between seasonal EKE in the control and CLIM experiments are seen in the Venezuela Basin and east of the Lesser Antilles. In CLIM high values of EKE during summer are reached from 65°W, whereas high EKE is seen from 60°W in the reference experiment. Moreover the peak of EKE occurring in May at 65°W (Fig. 4a) is absent in CLIM (Fig. 11a). This suggests that the NBC rings have an impact, although limited, on the eddy production in the far Eastern Caribbean. But one must be aware that the processes whereby energetic rings induce eddies on the western side of the Lesser Antilles are still not well understood and are certainly more complex than a simple crossing of the eddies through the Antilles Passages. Note also that the occurrence in CLIM of significant values of EKE in the Venezuela Basin suggests that NBC rings are not necessary for the formation of the eddies in this region. Although the arrival of energetic rings during summer could control part of the EKE annual cycle in the Venezuela Basin, a significant part of this annual cycle is linked with the annual variability of the Caribbean inflow through Grenada Passage. Two facts support this statement. First, there is a marked seasonal cycle in the Venezuela Basin in the CLIM experiment. Second, there is a maximum of inflow through the Grenada Passage in May–June (Figs. 7 and 10). Note that this inflow maximum coincides in time with the annual maxima of EKE on both sides of the Lesser Antilles. As mentioned in the previous section, the increased velocity shear and barotropic/baroclinic energy conversion favor the formation of energetic eddies in the Venezuela Basin during summer.
Finally, since the EKE in the Colombia Basin is similar in both experiments, we conclude that the levels of EKE in the Eastern Caribbean have low impact on the EKE in the Colombia Basin. This highlights a dynamical separation between the two regions. This is in agreement with findings of the previous section: most of the eddy energy in the Colombia Basin is produced locally. In conclusion, the advection of NBC rings only has a limited impact on the overall Caribbean eddy activity.
5. Interannual variability
Interannual variability of the EKE is now investigated in the Colombia Basin. We focus on this region since it presents the largest values of EKE and its semiannual cycle is the dominant feature in the whole Caribbean EKE cycle. The time series of EKE and MKE are shown in Figs. 12a,b for both model and altimetry data. Interannual fluctuations in the model are in good agreement with observations. We note that the amplitude of the interannual variability of EKE and MKE is of the same order as their seasonal fluctuations. The interannual variability of the EKE east of the Lesser Antilles in model and altimetry (Fig. 12c) are also in good agreement. The EKE in this region is dominated by the incoming NBC rings. This suggests that the model adequately resolves their energy and interannual variability. There is no obvious relationship between this variability and that of the EKE in the Colombia Basin (Fig. 12a). A 365-day running mean filter is applied to the model time series shown in Figs. 12a,b. At interannual scale, anomalies of EKE (Fig. 13a) are related to anomalies of MKE (Fig. 13b). We also evidence a close relationship between the interannual variability of the low-frequency currents and interannual variations of the zonal wind stress at the Guajira Peninsula (average between 12° and 15°N, 72.5° and 70°W). This means that most of the interannual variability of the eddies in the Colombia Basin is controlled by regional forcing. This hypothesis is supported by the comparison between the reference experiment and CLIM experiments: at first order, the interannual fluctuations of the EKE in the Colombia Basin in CLIM (Fig. 14) are similar to those simulated in the reference experiment (Fig. 13a). The main difference of EKE between both simulations is seen during year 2007 (Figs. 13a and 14). During this year, the westward winds were not anomalously strong at the Guajira Peninsula (Fig. 13c), but the EKE in the control simulation was high. We note that the energy east of the Lesser Antilles was particularly high in the model during 2006–07 (Fig. 12c). So, it is possible that during this year, energetic NBC rings had a direct or indirect influence on the EKE in the Colombia Basin.
Interannual variability of (a) EKE and (b) MKE (m2 s−2) averaged over the Colombia Basin (12°–17°N, 78°–72°W) from model (black) and altimetry derived (gray) surface currents. (c) The interannual variability of EKE averaged east of the Lesser Antilles (7°–13°N, 60°–57°W).
Citation: Journal of Physical Oceanography 42, 11; 10.1175/JPO-D-12-048.1
Interannual anomalies of model (a) EKE and (b) MKE (10−2 m2 s−2) averaged over the Colombia Basin (12°–17°N, 78°–72°W) as in Fig. 12. (c) Anomalies of zonal wind stress (10−2 N m−2) averaged at the Guajira Peninsula (between 12° and 15°N, 72.5° and 70°W). All datasets have been demeaned and smoothed with a 365-day running mean.
Citation: Journal of Physical Oceanography 42, 11; 10.1175/JPO-D-12-048.1
Interannual anomalies of model EKE (m2 s−2) as in Fig. 13 but for experiment CLIM.
Citation: Journal of Physical Oceanography 42, 11; 10.1175/JPO-D-12-048.1
In the previous section, we have found a close relationship between the seasonal variability of the alongshore winds and the seasonal variability of the EKE. It is remarkable that both interannual variability and seasonal variability of the EKE appear to be controlled by similar processes.
6. Discussion and summary
This work describes the seasonal and interannual variations of the eddy energy in the Caribbean Sea and investigates the mechanisms whereby this energy is modulated. A regional model of the Caribbean Sea and Gulf of Mexico, forced with ERA-Interim reanalysis, is analyzed for the period 1993–2009. Comparisons with satellite altimeter observations show that the model adequately represents the variability of the eddy energy in the Caribbean Sea.
Our results indicate that the variations of the eddy activity in the Caribbean Sea lies in the mean flow conditions through the Lesser Antilles and along the South American coast. In the Colombia Basin, the two maxima of EKE during March–April and September–October are explained by local enhancement of baroclinic and barotropic energy conversion during these periods. The seasonal variability of the EKE in the Colombia Basin is not clearly connected to the variability further east, which is dominated by an annual cycle. This highlights the importance of local processes.
The EKE in the Venezuela Basin follows an annual cycle. NBC rings or processes related to them act as perturbations that trigger instability and contribute to a maximum of EKE in this region during summer. However, the way whereby NBC rings increase the variability in the Venezuela Basin when they reach the Antilles is a complex process. Agreement between model, altimetry, and drifters concerning the splitting of the EKE maximum near the Lesser Antilles (one around the Barbados and other in the Venezuela Basin) supports the idea of Fratantoni and Richardson (2006) that only filaments of rings enter the Caribbean Sea. Once inside, these perturbations can generate larger eddies that grow in size and strength as they move westward to the Colombia Basin. The seasonal arrival of more energetic rings coincides with a peak inflow through Grenada Passage. The jet resulting from this inflow has strong horizontal and vertical velocity shears. Thus baroclinic and barotropic energy conversion are also increased at this time and contribute to the annual peak of EKE in the Venezuela Basin.
In agreement with previous studies, our results suggest that Caribbean eddies are not generated directly by the local wind. Instead, the importance of the wind field for the Caribbean mesoscale variability is indirect. The semiannual acceleration of the CLLJ, which passes over the Caribbean basin, controls variations in strength of the surface southern Caribbean Current and, in consequence, modifies its stability properties and the amount of energy available for the eddies.
The interannual variability of the eddy energy in the Colombia Basin is linked to the interannual variability of the low-frequency flow. Anomalies of the energy of the low-frequency currents tend to lead by about one or two months the anomalies of EKE (Figs. 6a,b and Fig. 13). We have also shown a close relationship between interannual anomalies of the zonal wind stress along the South American coast and the strength of the coastal jet and Caribbean eddies. So, the mechanisms at play at seasonal scale are also relevant at interannual scale. It is worth mentioning that an attempt to explain part of the seasonal and interannual variability of the EKE and MKE as the result of westward propagation of baroclinic Rossby waves has been unsuccessful. The intensity of the winds of the CLLJ, the constraint imposed by the coast, and the strong nonlinearity of the dynamics appear to favor the development of a regional/local variability at the expense of remote forcing through the propagation of long (and short) Atlantic Rossby waves.
Acknowledgments
We acknowledge the provision of supercomputing facilities by the CICESE (Conacyt Project SEP-2003-C02-44534). The regional configuration was set up in cooperation with the DRAKKAR project (http://www.drakkar-ocean.eu/). Altimetry data were produced by Salto/Duacs and distributed by Aviso, with support from CNES. We are grateful to two anonymous reviewers for their comments and suggestions.
REFERENCES
Alvera-Azcárate, A., A. Barth, and R. H. Weisberg, 2009: The surface circulation of the Caribbean Sea and the Gulf of Mexico as inferred from satellite altimetry. J. Phys. Oceanogr., 39, 640–657.
Andrade, C. A., and E. D. Barton, 2000: Eddy development and motion in the Caribbean Sea. J. Geophys. Res., 105 (C11), 26 191–26 201.
Andrade, C. A., and E. D. Barton, 2005: The Guajira upwelling system. Cont. Shelf Res., 25, 1003–1022.
Athie, G., J. Candela, J. L. Ochoa, and J. Sheinbaum, 2012: Influence of Caribbean cyclonic eddies on the Loop Current eddy shedding. J. Geophys. Res., 117, C03018, doi:10.1029/2011JC007090.
Barnier, B., and Coauthors, 2006: Impact of partial steps and momentum advection schemes in a global ocean circulation model at eddy permitting resolution. Ocean Dyn., 56 (5–6), 543–567.
Barnier, B., T. Reynaud, A. Beckman, C. W. Böning, J. M. Molines, S. Bernard, and Y. Jia, 2001: On the seasonal variability and eddies in the North Brazil Current: Insights from model inter-comparison experiments. Prog. Oceanogr., 48, 195–230.
Baums, I. B., M. W. Miller, and M. E. Hellberg, 2006: Geographic variation in clonal structure in a reef building Caribbean coral, Acropora palmata. Ecol. Monogr., 76, 503–519.
Bernie, D. J., E. Guilyardi, G. Madec, J. M. Slingo, and S. J. Woolnough, 2007: Impact of resolving the diurnal cycle in an ocean-atmosphere GCM. Part 1: A diurnally forced OGCM. Climate Dyn., 29, 575–590.
Blanke, B., and P. Delecluse, 1993: Variabiliy of the tropical Atlantic Ocean simulated by a general circulation model with two different mixed-layer physics. J. Phys. Oceanogr., 23, 1363–1388.
Candela, J., J. Sheinbaum, J. Ochoa, A. Badan, and R. Leben, 2002: The potential vorticity flux through the Yucatan Channel and the Loop Current in the Gulf of Mexico. Geophys. Res. Lett., 29, 2059, doi:10.1029/2002GL015587.
Carton, J. A., and Y. Chao, 1999: Caribbean Sea eddies inferred from TOPEX/Poseidon altimetry and a 1/6° Atlantic Ocean model circulation. J. Geophys. Res., 104 (C4), 7743–7752.
Chérubin, L. M., and P. L. Richardson, 2007: Caribbean current variability and the influence of the Amazon and Orinoco freshwater plumes. Deep-Sea Res. I, 54, 1451–1473.
Corredor, J. E., J. M. Morell, J. M. Lopez, J. E. Capella, and R. A. Armstrong, 2004: Cyclonic eddy entrains Orinoco river plume in Eastern Caribbean. Eos, Trans. Amer. Geophys. Union, 85, 20, doi:10.1029/2004EO200001.
Dai, A., and K. Trenberth, 2002: Estimates of freshwater discharge from continents: Latitudinal and seasonal variations. J. Hydrometeor., 3, 660–687.
Fratantoni, D. M., and P. L. Richardson, 2006: The Evolution and demise of North Brazil Current rings. J. Phys. Oceanogr., 36, 1241–1264.
Johns, W. E., T. L. Townsend, D. M. Fratantoni, and W. D. Wilson, 2002: On the Atlantic inflow to the Caribbean Sea. Deep-Sea Res. I, 49, 211–243.
Johns, W. E., R. J. Zantopp, and G. Goni, 2003: Cross-gyre transport by North Brazil Current rings. Interhemispheric Water Exchange in the Atlantic Ocean, G. J. Goni and P. Malanotte-Rizzoli, Eds., Elsevier Oceanographic Series, Vol. 68, Elsevier, 411–441.
Jouanno, J., J. Sheinbaum, B. Barnier, J.-M. Molines, L. Debreu, and F. Lemarié, 2008: The mesoscale variability in the Caribbean Sea. Part I: Simulations and characteristics with an embedded model. Ocean Modell., 23, 82–101, doi:10.1016/j.ocemod.2008.04.002.
Jouanno, J., J. Sheinbaum, B. Barnier, and J.-M. Molines, 2009: The mesoscale variability in the Caribbean Sea. Part II: Energy sources. Ocean Modell., 26, 226–239, doi:10.1016/j.ocemod.2008.10.006.
Kirchner, K., M. Rhein, C. Mertens, C. W. Böning, and S. Hüttl, 2008: Observed and modeled MOC related flow into the Caribbean. J. Geophys. Res., 113, C03028, doi:10.1029/2007JC004320.
Large, W. L., and S. G. Yeager, 2004: Diurnal to decadal global forcing for ocean and sea-ice models: The data sets and flux climatologies. NCAR Tech. Rep. TN-460+STR, 105 pp.
Madec, G., 2008: NEMO ocean engine. Institut Pierre-Simon Laplace Nôte du Pole de modélisation 27, 367 pp.
Masina, S., G. Philander, and A. Bush, 1999: An analysis of tropical instability waves in a numerical model of the Pacific Ocean. Part II: Generation and energetics. J. Geophys. Res., 104, 637–662.
Murphy, S. J., H. E. Helburt, and J. J. O’Brien, 1999: The connectivity of eddy variability in the Caribbean Sea the Gulf of Mexico and the Atlantic Ocean. J. Geophys. Res., 94 (C1), 1431–1453.
Oey, L.-Y., H.-C. Lee, and W. J. Jr Shmitz, 2003: Effect of winds and Caribbean eddies on the frequency of Loop Current eddy shedding: a numerical model study. J. Geophys. Res., 108, 3324, doi:10.1029/2002JC001698.
Rhein, M., K. Kirchner, C. Mertens, R. Steinfeldt, M. Walter, and U. Fleischmann-Wischnath, 2005: Transport of South Atlantic Water through the passages south of Guadeloupe and across 16°N, 2000-2004. Deep-Sea Res. I, 52, 2234–2249.
Richardson, P., 2005: Caribbean Current and eddies as observed by surface drifters. Deep-Sea Res. II, 52, 429–463.
Simmons, H. F., and D. Nof, 2002: The squeezing of eddies through gaps. J. Phys. Oceanogr., 32, 314–335.
Tréguier, A., and Coauthors, 2001: An eddy permitting model of the Atlantic circulation: Evaluating open boundary conditions. J. Geophys. Res., 106, 22 115–22 130.
Wang, C., 2007: Variability of the Caribbean low-level jet and its relations to climate. Climate Dyn., 29, 411–422.
