Dietrich (1937) provided perhaps the first comprehensive depiction of the surface circulation in the Gulf of Mexico, which shows the major circulation features now known as the LC in the eastern Gulf and a detached mesoscale anticyclonic eddy in the western Gulf. Austin (1955) provided the first use of the term “loop” to refer to the behavior of the current as it passes from the Yucatan Channel to the Florida Straits. DiMarco et al. (2005) provide a 10-year climatology of the mean surface circulation based on nearly 1,400 near-surface drifter records. The LC arrives in the Gulf through the Yucatan Channel, mostly above 800 meters (Sheinbaum et al., 2002), at a rate of approximately 23–27 Sv (1 Sv = 106 m3/s) and brings warm (25–26°C) and salty (36.7–36.8 ppt) waters into the Gulf (Badan et al., 2005; Hamilton et al., 2005). Once in the Gulf, the LC and associated surface flows remain an upper ocean feature (i.e., approximately 800–1,200 meters and above) in the eastern Gulf (Hamilton, 2009; Oey, 2008). The degree of LC intrusion into the Gulf varies from between about 24°N to 28°N on time scales of 0.5–18.5 months without a clear annual signal (Alvera-Azcárate et al., 2009; Hamilton et al., 2011; Sturges and Lugo-Fernández, 2005; Zeng et al., 2015b). The lowest degree of intrusion (i.e., 24°N) is often called a “port-to-port” position where the LC extends into the Gulf only minimally; the “average” position marks a northward position of approximately 26°N, roughly equivalent to the middle of the Gulf; and the “fully extended” position (approximately 28°N) of the LC abuts the Louisiana-Texas Shelf (e.g., Gopalakrishnan et al., 2013). In each of the LC extension phases, the western side of the LC closely follows the eastern side of the Campeche Bank (the continental shelf extending north of the Yucatan Peninsula).
In general, the greater the LC extension, the greater the likelihood that the LC will form an anticyclonic warm-core eddy that separates from the main LC flow and propagates westward (e.g., Schmitz, 2005). When separation events do occur, they are not quick (i.e., taking several months) and often, an eddy will reattach to the main LC before separating for good (Sturges and Leben, 2000); after a separation, the main LC intrusion retreats significantly. These eddies are very large (approximately 300 kilometers in diameter), spin at about 3–4 knots, and move at speeds of a few centimeters per day (approximately 0.1 knot) (Sturges and Lugo-Fernández, 2005). The interval at which an LCE separates varies considerably (approximately 0.5–19 months), but it averages about every 8 to 9 months (Leben, 2005). There is some evidence of a higher probability of separation during the spring and fall (Hall and Leben, 2016), but there is
clearly much variability in the timing of separations. In terms of mechanisms, Sturges et al. (2010) suggested that these separation events may be influenced by 20- to 30-day signals propagating upstream into the Gulf from the Carribbean Sea. This is based on observations of increased eastward transport through the Florida Straits, as well as increases in sea level on the offshore side of the LC flow in the weeks prior to a separation, and builds on earlier work by Maul (1977). There is also evidence that cyclonic eddies along the LC front influence both LC extension and eddy separation (Gopalakrishnan et al., 2013; Schmitz, 2005). Bottom flows out of the Yucatan are also linked to eddy separation (Oey, 1996). Additionally, bottom topography is likely to have a role in LCS dynamics (e.g., Chérubin et al., 2005).
At any given time, the LC and potentially associated large anticyclonic eddies exist within a larger eddy field containing many smaller eddies (approximately 40–150 kilometers in diameter), including both anticyclones and cyclones (Hamilton et al., 2002). At the northern edge of the LC penetration, pairs of counter-rotating eddies may be a major mechanism for transporting material on and off the Louisiana and Texas continental shelves (Hamilton et al., 2002). In addition, cold-core cyclonic eddies (approximately 80–200 kilometers in diameter) that move downstream along the outside edges of the LC have also been observed (Huang et al., 2013; Le Hénaff et al., 2014; Rudnick et al., 2015). These eddies may be present on the western, northern, or eastern (i.e., West Florida Shelf side) edge of the LC and may play a significant role in the transportation of particles (e.g., Hamilton et al., 2002). The especially intense cyclonic eddies form near the Dry Tortugas and are called simply Tortuga eddies (Huang et al., 2013). Understanding the origin of cyclonic eddies, as well as their physics, interaction with the LC, and effect on Gulf biology, marine mammals (Biggs et al., 2005), and chemistry, require further observations and analyses.
Much of the work performed over the last several decades has focused on surface flows in the Gulf largely because surface measurements are more accessible. There is a long and rich history of energy industry-sponsored studies of the LC and LCE system dating back to the 1980s (e.g., Lewis et al., 1989). Many of these studies focused on limited surface drifter, shipboard, and aircraft observations of the horizontal and vertical structure of the currents in the upper 1,000 meters of the water column associated with long extensions of the LC into the Gulf, and with the LCEs that detach and propagate westward, some of which propagate along the central Gulf of Mexico escarpment where oil and gas exploration and production occur. The general deepwater LCE current structure increases steadily from the center out to a radius of maximum velocity, then decays rapidly toward the outer edge. Vertically, the maximum currents often occur in the upper 100 meters, and often decay steadily down to a depth ap-
proaching 1,000 meters, below which the slower and relatively low sheared deep flow is observed. A kinematic feature model for the most critical design features of an LCE, specifically the horizontal structure and the current shear in the upper 1,000 meters of the inner core, was developed (e.g., Forristall et al., 1992; Glenn et al., 1990) and has been used as a design tool for establishing LCE eddy climatologies based on satellite and drifter observations of the eddy track and average swirl velocity. But kinematic LCE models, while useful for historical studies along past LCE trajectories, are not forecast models. They do not forecast LC growth or LCE separation and subsequent propagation. They assume the eddies are isolated in deepwater and do not include LCE interactions with the LC or topography, and they use a best-fit symmetric elliptical shape to represent the actual complex LCE horizontal structure. Even individual LCEs have been observed to exhibit a range of different asymmetries over their lifetime, including asymmetries consistent with LC interactions, TRW dispersion, and planetary Rossby wave dispersion (Glenn and Ebbesmeyer, 1993).
Weatherly et al. (2005) utilized data from 17 Argo (PALACE variety) floats set in the Gulf that sampled the intermediate-depth (approximately 900-meter) flow from April 1998 to February 2002. Their analysis revealed a mean cyclonic circulation along the northern, western, and southwestern edges of the Gulf. This flow intensified into an approximately 0.10 m/s current in the western and southern Bay of Campeche and was deflected around a bathymetric feature, called here the “Campeche Bay Bump,” in the southern Bay of Campeche. Associated with this intensified flow was a small cyclonic gyre in the southwestern Bay of Campeche. Floats launched in the eastern Gulf of Mexico tended to stay there and those launched in the western Gulf tended to stay in the western Gulf, suggesting a restricted connection at depths between the eastern and western Gulf. While this conclusion derived from a field study is not the final answer in unraveling the behavior of deep currents, it does demonstrate the complexity of deep Gulf circulation, and how it is heretofore not well observed or understood.
Our knowledge of the deep (greater than 1,000-meter) Gulf flows is driven primarily by data from Lagrangian floats and current meters/moorings, as well as numerical models (DeHaan and Sturges, 2005; Schmitz et al., 2005). The mean deep flow is thought to be opposite in direction from the main surface LC flows (i.e., cyclonic) in both the eastern and western Gulf and to move more slowly (e.g., approximately 1–2 cm/s at 2,000 meters) (DeHaan and Sturges, 2005). While slower than near-surface LCS currents, deep eddy currents are much stronger than the mean, with speeds that frequently grow to 20–30 cm/s (approximately 1/2 knot). They extend nearly independent of depth through the full water column (Hamilton, 2009). These have been observed on numerous moorings in water depths from 2,000 to 3,500 meters south of Alabama to Texas. These deep currents are thought to be associated, at
least in part, with TRWs that have periods of 10 to 100 days and wavelengths of 50 to 200 kilometers (Hamilton, 2009). The TRWs appear to be generated by stretching and squashing of the lower water column (potential vorticity adjustments) when the sloped thermocline of the LC shifts across the sloped seafloor. Deep cyclonic circulation is in part driven in the mean when TRWs approach the more steeply sloped upper continental slope and reflect (Hamilton, 2009). Some modeling work suggests that the deep cyclonic circulation observed may also be related to the formation of a deep cyclone-anticyclone eddy pair under the LC, of which the anticyclone dissipates much more quickly (Welsh and Inoue, 2000). There is also evidence for the role of the Campeche Bank coupled with the baroclinic instability of the LC in the formation of these eddies (Oey, 2008; Weatherly et al., 2005). The deep cyclones in a numerical model by Oey (2008) have swirl speeds of approximately 0.3 cm/s and are approximately 100–200 kilometers across with vertical scales of 1,000–2,000 meters.