Volume 48, Issue 4 e2020GL091277
Research Letter
Open Access

Eddies Connect the Tropical Atlantic Ocean and the Gulf of Mexico

Minghai Huang

Minghai Huang

School of Marine Science and Policy, University of Delaware, Lewes, DE, USA

Search for more papers by this author
Xinfeng Liang

Corresponding Author

Xinfeng Liang

School of Marine Science and Policy, University of Delaware, Lewes, DE, USA

Correspondence to:

X. Liang,

[email protected]

Search for more papers by this author
Yingli Zhu

Yingli Zhu

School of Marine Science and Policy, University of Delaware, Lewes, DE, USA

Search for more papers by this author
Yonggang Liu

Yonggang Liu

College of Marine Science, University of South Florida, St. Petersburg, FL, USA

Search for more papers by this author
Robert H. Weisberg

Robert H. Weisberg

College of Marine Science, University of South Florida, St. Petersburg, FL, USA

Search for more papers by this author
First published: 20 January 2021
Citations: 16

Abstract

Numerical circulation modeling and observational studies have been conducted to understand the Loop Current (LC) system behaviors in the Gulf of Mexico (GoM). One of the factors that may influence the LC are upstream eddies from within the Caribbean Sea. By combining satellite altimetry, sea surface salinity and ocean color data, we demonstrate that mesoscale eddies from the western tropical Atlantic Ocean can eventually make their way to the Gulf of Mexico and likely affect the LC. In addition, our study shows that freshwater of Amazon and Orinoco River origin trapped within mesoscale eddies can also enter the GoM, potentially affecting the GoM stratification. This study provides insights into understanding variations of the LC system and showcases the roles of mesoscale eddies in connecting the open ocean and regional seas.

Key Points

  • Some eddies from the Atlantic Ocean can ultimately reach the Gulf of Mexico and likely affect the Loop Current

  • Freshwater of Amazon and Orinoco River origin and other materials trapped in eddies could reach the Gulf of Mexico

  • Weakening and strengthening of the long-propagating eddies are mostly related to the variation of bathymetry

Plain Language Summary

The Loop Current (LC) is the dominant large-scale oceanic process in the Gulf of Mexico (GoM). However, the mechanism for variations of the LC system is still unsolved. Here, we show that some mesoscale eddies originated in the tropical Atlantic Ocean can pass through the Caribbean Sea and eventually enter the GoM. These remotely generated eddies could be an important upstream factor affecting the behavior of the LC. Also, freshwater and other materials (e.g., chlorophyll) trapped in the eddies could reach the GoM as well. In addition to advancing the understanding of the LC system, this study provides an explicit example showing eddies can serve as a route connecting regional seas and the open ocean.

1 Introduction

The Gulf of Mexico (GoM) is a semi-enclosed sea connecting the Caribbean Sea and the Atlantic Ocean through the Yucatan Channel and the Straits of Florida, respectively. The Loop Current (LC) is the most prominent physical feature in the GoM. It has a significant influence on various processes in the GoM, such as dispersal of spilled oil (e.g., Crone & Tolstoy, 2010; Hazen et al., 2016; Liu et al., 2011; Weisberg et al., 2016, 2017), sediment transport, fish production (e.g., Hazen et al., 2016), and distribution of nutrients (e.g., Hu et al., 2005). In addition, it plays an essential role in the atmosphere-ocean coupling, which influence the prediction of hurricanes and their impacts (e.g., Chen & Curcic, 2016; Curcic et al., 2016; Shay & Jacob, 2006; Sheng et al., 2010).

The past few decades have provided a large number of observational and numerical circulation modeling studies on the LC system (e.g., Candela et al., 2019; Chang & Oey, 2010, 2013; Hirschi et al., 2019; Liu et al., 2016; Oey et al., 2005; Sturges & Leben, 2000; Xu et al., 2013; Weisberg & Liu, 2017). Despite these and others, a fundamental question remains: What controls the trajectory of the LC? Many factors affecting this complex, dynamical system have been proposed, such as the Yucatan Channel transport, bottom topography, and atmospheric forcing. In particular, several modeling studies suggest that upstream factors outside of the GoM could be important (Alvera-Azcárate et al., 2009; Jouanno et al., 2008; Murphy et al., 1999; Oey, 2004).

Previous studies suggest that mesoscale eddies that may impact the GoM could originate as far upstream as the North Brazil Current (NBC). Rings that are shed from the NBC as it retroflects into the North Equatorial Countercurrent (NECC) can propagate northwestward and interact with the Lesser Antilles. Such NBC rings may be deflected or defracted by the island chain, with the result that some of them may enter the Caribbean Sea (Fratantoni & Richardson, 2006). Numerical modeling studies (e.g., Murphy et al., 1999) show that potential vorticity from NBC rings reforming west of the Lesser Antilles may then grow in the Caribbean Sea gaining energy via mixed barotropic and baroclinic instabilities. Some of those eddies can then enter the GoM through the Yucatan Channel, perhaps impacting the trajectory of the LC and the related eddy-shedding process (e.g., Yang et al., 2020). Besides the modeling studies, drifter track, altimetry observations and Amazon river plume studies also show the advection of these eddies and its filaments cross the Lesser Antilles (e.g., Carton & Chao, 1999; Chérubin & Richardson, 2007; Fratantoni & Richardson, 2006; Goni & Johns, 2001; Richardson, 2005). However, more definitive studies on such long-distance mesoscale eddy connections between the GoM and the tropical Atlantic Ocean remain to be accomplished.

In this study, using various satellite products, including altimetry, sea surface salinity and chlorophyll, we explore roles of mesoscale eddies in connecting the tropical Atlantic and the GoM. The study is organized as follows. A brief description of the data is provided in Section 2. Section 3 presents the propagations of eddies from the tropical Atlantic Ocean to the GoM. At last, the results are summarized and discussed in Section 4.

2 Data and Methods

The Mesoscale Eddy Trajectory Atlas Product (META) distributed by AVISO+ was used to obtain the mean eddy trajectory in the study region as well as a few basic properties of individual eddies. META provides various information along the trajectories of individual eddies over the period January 1993–January 2018, including eddy type, position, amplitude (i.e., magnitude of the difference between the extremum of sea level anomaly (SLA) within the eddy and the SLA at the eddy perimeter) and speed. The “growing method,” which defines each eddy on basis of connected pixels that satisfy specified criteria (Schlax & Chelton, 2016), is used in META to detect and track eddies (Duacs/AVISO+, 2017). Note that the eddy trajectory in META will be stopped if there is land between two consecutive eddies, likely resulting in discontinuities for long-propagating eddies when they encounter topography like island chains.

To avoid discontinuities related to the eddy tracking method mentioned above, we also studied the eddy propagation by examining the absolute dynamic topography (ADT) along the mean eddy trajectory (e.g., Halo et al., 2014; Laxenaire et al., 2018). Note that both ADT and SLA have been used to detect and track eddies, we chose to use ADT in this study because SLA maps are strongly affected by the large sea surface height gradients that are related to strong currents (e.g., Halo et al., 2014). For most part of the study region (from the North Brazil region to Chibcha Channel), the mean eddy trajectory was defined as the mean position of the eddies derived from META. From the Chibcha Channel to Yucatan Channel, where eddy trajectories are complex, we used the 72 cm isoline of sea surface height roughly as the mean eddy trajectory (e.g., Alvera-Azcárate, 2009). In the Atlantic Ocean, a zonal line at 4oN was used. In general, more eddies are detected along the mean eddy trajectory defined above, particularly between the GoM and the NBC retroflection.

The ADT data we used in this study are the daily satellite altimetry data set distributed by the Copernicus Marine Environment Monitoring Service (CMEMS). The data set covers the period from January 1993 to December 2018 and has a spatial resolution of 0.25° × 0.25°. This data set includes a number of variables, including ADT, SLA, and geostrophic currents. More specifically, the altimetry data set was used to derive the Hovmöller plots of ADT and estimate boundaries of sample eddies.

The sea surface salinity (SSS) from European Space Agency Earth Explorer mission (SMOS) was used to explore the impacts of mesoscale eddies on the freshwater transport. The SSS data cover a period from January 2010 to December 2017 (Boutin et al., 2018). The SSS data have a temporal interval of 4 days and a spatial resolution of about 25 km. In addition, we examined the chlorophyll data from GlobColour, which merges several products from SeaWiFS, MERIS, MODIS, VIIRS NPP, OLCI-A, VIIRS JPSS-1 and OLCI-B to achieve better spatial and temporal coverage (Maritorena et al., 2010). The data set is from 1997 to 2019 with a spatial resolution of 4 km and a temporal resolution of 8 days. Also, we used ETOPO5 bathymetry data to examine the impacts of bathymetry on the propagating mesoscale eddies.

Lagged correlations of each variable (ADT, SSS, chlorophyll) between its values along the mean eddy trajectory and from a sample point on the trajectory at 55oW were calculated to examine how far the corresponding signals can remain coherent along the mean eddy trajectory. All variables were band-pass filtered (40–200 days). By definition, correlations will approach one from the beginning position in the NBC region (zero lag in space and time).

3 Results

The trajectories of mesoscale eddies that lasted more than 60 days in the study region and occurred between January 1993 and January 2018 were derived from META and shown in Figures 1a and 1b. From the western tropical Atlantic Ocean to the GoM, the eddy trajectories are continuous except for three regions: Lesser Antilles, Chibcha Channel and Yucatan Channel. The main pattern of the anticyclonic and cyclonic eddy track is similar. The numbers of the anticyclonic and cyclonic eddies are on the same order over the examined period (∼200 in the North Brazil, ∼400 in the Caribbean Sea, and ∼250 in the GoM). Note that since in this study we consider the long-distance propagation of mesoscale eddies, only eddies lasting more than 60 days were selected and presented.

Details are in the caption following the image

Eddy tracks of (a) anticyclonic eddies, (b) cyclonic eddies in the study region. Only eddies last more than 60 days are shown. The thick black line stands for the mean eddy trajectory, which is used to further examine the eddy propagation. (c) Hovmöller plot of the band-pass filtered (40–200 days) ADT following the mean eddy trajectory. The black dot line marks one individual eddy case that will be further examined later. The dash vertical lines mark the Lesser Antilles, Chibcha Channel, Yucatan Channel and NBC retroflection (NBCR). (d) Lagged correlations between ADT along the mean eddy trajectory and ADT at one point on the mean eddy trajectory at 55oW. ADT stands for absolute dynamic topography.

Since the eddy detection method in META arbitrarily stops the trajectory if land is found between consecutive eddies, the discontinuity shown in Figures 1a and 1b could be misleading. We then used a second way to track the propagations of mesoscale eddies and examined if the eddies on the two sides of the discontinuous locations were actually connected. We first calculated the mean eddy trajectory (thick black lines in Figures 1a and 1b) as described above. Then, following the mean eddy trajectory, we derived a Hovmöller diagram of the band-pass filtered (40–200 days) ADT (Figure 1c). Here, for better visualization we only present a few years of the data and the other years show similar patterns. It is clear that many ADT signals appear propagating continuously from the western tropical Atlantic Ocean to the GoM even in the regions of discontinuity shown in Figures 1a and 1b (i.e., Lesser Antilles, Chibcha Channel and Yucatan Channel).

ADT signals on the two sides of the Lesser Antilles are closely related, displaying a significant correlation of 0.69 with a time lag of one month between two sample points (62.9oW and 60.4oW) on the mean eddy trajectory. Thus, some eddies inside the Caribbean Sea, particularly those originating near the Lesser Antilles are likely related to eddies in the tropical Atlantic Ocean. We also calculated the lagged correlations of ADT between values along the mean eddy trajectory and a sample point on the trajectory at 55oW (Figure 1d). It further confirms that some of the ADT signals in the tropical Atlantic Ocean can pass the Lesser Antilles, Chibcha Channel and part of them can even reach the GoM. Based on the Hovmöller plot of ADT, we found that between January 1993 and December 2018 there were at least 27 eddies that had an Atlantic Ocean origin and reached the GoM. That is roughly one eddy per year.

We also explored the propagation of individual eddies with the CMEMS ADT data. Three cases of long-distance eddy propagation from the western tropical Atlantic Ocean are presented in Figure 2. Note that for any individual eddy, its boundary was generally defined as the outmost closed ADT contour. However, sometimes, the existence of merging, splitting, and deforming, makes it difficult to use the closed contours. Then, the perimeter was chosen manually to mark the eddy boundary. For case 1 (Figures 2a and 2b), a clearly defined eddy was identified in the middle of the tropical Atlantic Ocean in July 2003 and then propagated westward until encountering the continent of the South America around December 2003. Then, the eddy propagated northwestward following the corridor along the north Brazil coastline and eventually encountered the Lesser Antilles (Figure 2a), where the intensity of the eddy was significantly reduced. However, after part of the eddy was diffracted into the Caribbean Sea in April 2004, the eddy eventually grew into a much stronger eddy as it propagated into the middle Caribbean Sea in December 2004. At last, the eddy entered the GoM and became part of the LC in May 2005 (Figure 2b). The long-distance propagation from the middle of the tropical Atlantic Ocean to the GoM took about 22 months. Similar to case 1, the eddy shown in case 2 (Figure 2c) was first identified in the western tropical Atlantic Ocean in March 2014, and after about 14 months it eventually entered the GoM. It should be noted that eddies originated in the western tropical Atlantic Ocean can also enter the Caribbean Sea but fail to reach the GoM. Case 3 (Figure 2d) is such an example.

Details are in the caption following the image

Cases of long-distance propagation of individual eddy. (a and b) from July 2003 to May 2005, (c) from March 2014 to May 2015, and (d) from January 1995 to December 1995. The dates are shown on top of each snapshot. The absolute dynamic topography (ADT) is superimposed by the geostrophic current. The outmost contour of ADT marks the boundary of each eddy. The 1,000 m isobath and mean eddy trajectory are marked.

We further examined the evolution of the three individual eddies along their propagation trajectories (Figure 3a). When crossing the Lesser Antilles, the eddies were relatively weak with an amplitude of ∼5 cm, which is defined as magnitude of the difference between the extremum of ADT within the eddy and at the eddy perimeter. When the eddies reached the eastern Caribbean Sea, the amplitude sharply increased to 20–25 cm. After that, the eddies intensity decreased again around the Chibcha Channel. To examine if this evolution of eddy intensity is robust or not, we also calculated the statistical results along the mean eddy trajectory with most of the detected cyclonic and anticyclonic eddies in META. The results revealed similar features to the three individual eddies (Figure 3b). In general, the amplitude of the eddies decreased sharply when reaching the Lesser Antilles, the Chibcha Channel and the Yucatan Channel and increased when moving away from these large topographic features. In fact, not only the amplitude, but other variables, like the eddy radius (Figure 3c), also showed similar influence of topography. The evolution of eddy characteristics with topography is likely controlled by the conservation of potential vorticity of the propagating eddies (e.g., Volkov & Fu, 2008). It should be noted that besides topography, local processes (e.g., upwelling, river plumes) could affect the detailed features of the propagating eddies (van der Boog et al., 2019).

Details are in the caption following the image

(a) Evolution of eddy amplitude for the three individual eddies shown in Figure 2. (b) Amplitude evolution for all the anticyclonic eddies (black solid line) and cyclonic eddies (black dashed line) following the mean eddy trajectory. (c) Same as (b), but for eddy radius. The positions of the Lesser Antilles, Chibcha Channel, Yucatan Channel and Beata Ridge are marked as vertical dash lines. The topography along the mean eddy trajectory is marked in orange.

The eddy cases and the ADT propagation indicate that in contrast to the previous studies, mesoscale eddies from the tropical Atlantic Ocean, at least some of them, can cross the discontinuity regions and finally reach the GoM. These eddies serve as vorticity flux passing the Yucatan Channel and could trigger the Loop Current retraction, extension and eddy-shedding (e.g., Yang et al., 2020). Previous studies suggest that the negative (anticyclonic) vorticity flux is related to the Loop Current extension while the positive (cyclonic) vorticity flux causes retraction and sometimes shedding (e.g., Athié et al., 2012; Candela et al., 2002). Our case eddies shown in Figures 2a–2c are examples of the anticyclonic eddies inducing LC extension events. Furthermore, a recent study (Andrade-Canto et al., 2020) clearly shows that some eddies entering the GoM through the Yucatan Channel can eventually result in LCE shedding.

The eddies under consideration carried other material properties along with energy and momentum. Figure 4 shows the Hovmöller diagram of the band-pass filtered (40–200 days) SSS anomalies (SSSA) and the accompanied ADT. Similar to the ADT, the SSSA can propagate from the NBC region, across the Lesser Antilles and into the Caribbean Sea. One can see the propagation across Lesser Antilles even there are some data missing near the Lesser Antilles. In some years, these SSSA features can also propagate cross the Chibcha and Yucatan Channels to enter the GoM. The white dot line marked in Figure 4a shows such a case, which also corresponds to the second eddy case shown in Figure 2c. In addition, lagged correlations between SSSA along the mean eddy trajectory and a point on the trajectory at 55oW (Figure 4c) confirm the long propagation of SSSA signals that originate in the tropical Atlantic Ocean. Therefore, individual eddies can carry freshwater from the western tropical Atlantic Ocean to not only the Caribbean Sea as suggested by previous studies (e.g., Hellweger & Gordon, 2002) but also further into the GoM.

Details are in the caption following the image

Hovmöller plots of (a) sea surface salinity anomalies (SSSA) and (b) ADT. Both were band-pass filtered to keep their variations between 40 and 200 days. The case 2 eddy presented in Figure 2 are marked as dot lines in (a) and (b). (c) Lagged correlations between SSSA along the mean eddy trajectory and the SSSA at one point (55oW) on the trajectory. ADT stands for absolute dynamic topography.

In addition, chlorophyll and colored dissolved organic matter (CDOM) also show the propagation of mesoscale eddies (e.g., Fratantoni & Glickson, 2002; Hu et al., 2004). The contrast between the high nutrient Amazon-influenced water and the surrounding relatively low nutrient mid-ocean water mark the NBC retroflection and the rings. The low-chlorophyll rings core and high-chlorophyll boundaries show the evolution and propagation of the NBC rings. Here, the propagation of chlorophyll is shown in Figure 5. For example, from 2014 to 2015, the chlorophyll anomalies propagated from the NBC retroflection regions, crossed the Lesser Antilles into the Caribbean Sea, and then finally entered the GoM. It should be noted that the mechanisms for chlorophyll anomalies are complex and are at least from two parts: horizontal advection and vertical upwelling of nutrient or chlorophyll itself (e.g., Killworth et al., 2004; O'Brien et al., 2013). Variability in river discharge will also impact the chlorophyll distribution. Therefore, nosier patterns than the ADT signals are expected. Nevertheless, lagged correlations between chlorophyll along the mean eddy trajectory and one point on the trajectory at 55oW clearly show the long-distance propagation of chlorophyll anomalies (Figure 5c). Therefore, our results show that mesoscale eddies can affect the bioproductivity along their long-propagation trajectory that connects the tropical Atlantic Ocean and the GoM.

Details are in the caption following the image

Hovmöller plots of (a) chlorophyll anomalies (CHLA) and (b) ADT. Both were band-pass filtered to keep their variations between 40 and 200 days. (c) Lagged correlations between chlorophyll anomalies along the mean eddy trajectory and values at one point (55oW) on the trajectory. ADT stands for absolute dynamic topography.

4 Conclusions and Discussion

By combining the eddy track, ADT, SSS and chlorophyll data, we show that some of the mesoscale eddies originating in the tropical Atlantic Ocean can eventually enter the GoM. In other words, mesoscale eddies can connect the tropical Atlantic Ocean and the GoM. Although such a long-distance connection by way of mesoscale eddies has been suggested in previous numerical studies (Murphy et al., 1999; van Westen et al., 2018), here we provide direct observational evidence. In addition, earlier studies have suggested that the potential vorticity flux through the Yucatan Channel may influence the LC trajectory and the eddy shedding process in the GoM (Andrade-Canto et al., 2020; Athié et al., 2012; Candela et al., 2002; Murphy et al., 1999; Oey, 2004). So, the long-distance propagating eddies from the tropical Atlantic Ocean may play a role in LC evolution and LCE shedding. Besides the eddies shown along the mean eddy trajectory, eddy propagation along another zonal line in the Atlantic Ocean was also examined. Coherent features from the Atlantic Ocean to the Caribbean Sea were also shown, suggesting that such eddy connections between the Atlantic Ocean and the GoM are common.

A more specific impact of those long-propagating eddies is related to the freshwater transport. As the largest oceanic rings, the NBC rings transport freshwater and other materials, particularly considering that the NBC flows past the Amazon River (Chérubin & Richardson, 2007; Fournier et al., 2017; Fratantoni & Glickson, 2002; Grodsky et al., 2015; Hellweger & Gordon, 2002). Early studies show that the Amazon river plume can influence the Caribbean Sea salinity variation through salt advection (Hellweger & Gordon, 2002; Muller-Karger et al., 1988). In this study, we show that Amazon and Orinoco River freshwater trapped in mesoscale eddies not only can get into the Caribbean Sea but can finally enter the GoM in many cases (Figure 4). The freshwater carried by those eddies could generate barrier layers and raise sea surface temperature in the GoM. As a consequence, those long-distance propagating eddies can potentially contribute to hurricane intensification in the Caribbean Sea and the GoM. Preliminary results based on two Argo floats moving from the Caribbean Sea and finally entering the GoM indeed show that the propagating eddies can affect salinity profiles and hence stratification along their propagating trajectories.

There is similarity between the evolution and propagation of individual ocean eddy and of the atmosphere hurricanes and storms, which also show weakening and strengthening along their trajectories. But the number of studies on the ocean counterpart of hurricanes and storms are much less. In this study, we show a close relationship between topography and variations of eddy intensity between the Lesser Antilles and the Yucatan Channel. However, inside the GoM and in the Tropical Atlantic Ocean, no such clear relations appear. These observations suggest that various factors and mechanisms are involved in any successful eddy connection events between the tropical Atlantic and the GoM, and more carefully designed studies are needed in the future to further explore this complex dynamical process.

Acknowledgments

We thank two anonymous reviewers whose comments and suggestions helped improve this paper. The work was supported by the Gulf of Mexico Research Initiative through Grant G-231804 and the Alfred P. Sloan Foundation through Grant FG-2019-12,536. Partial support was also provided by the Gulf Research Program of the National Academies of Sciences, Engineering, and Medicine (NASEM) UGOS-1 Grant 2000009918 and the NOAA IOOS through SECOORA Program Grant NA16NOS0120028.

    Data Availability Statement

    All the data used in this study are publicly available. The mesoscale eddy trajectory product is from the “Mesoscale Eddy Trajectory Atlas Product” distributed by AVISO+ (https://www.aviso.altimetry.fr/). The altimetry datasets can be obtained from the Copernicus Marine Environment Monitoring Service (CMEMS, https://marine.copernicus.eu/). The sea surface salinity data set is from the European Space Agency Earth Explorer mission (SMOS, https://earth.esa.int/web/guest/missions/esa-operational-eo-missions/smos). And the chlorophyll data set is available at GlobColour (https://hermes.acri.fr/). All the data are also publicly available through the Gulf of Mexico Research Initiative Information and Data Cooperative (GRIIDC; https://data.gulfresearchinitiative.org/pelagos-symfony/data/R6.x820.000:0009).