On the translation of Agulhas rings to the western South Atlantic Ocean

Highlights

• A new Agulhas ring census using 23.6 years of satellite altimetry data is presented.

• A total of 140 rings were shed by the Agulhas Retroflection between 1993 and 2016.

• Seventy-four rings with lifetimes longer than 360 days left the Cape Basin towards the west.

• Only 3 rings reached the western boundary and interacted with the Brazil Current.

• In situ data show one of the rings increasing the local current velocity by three times.

Abstract

The shedding of Agulhas rings is the primary process connecting the Indian and Atlantic oceans. The rings transport warm and salty waters that feed the surface limb of the Atlantic Meridional Overturning Circulation. Early studies suggest that Agulhas rings decay and diffuse their contents within the South Atlantic subtropical gyre. In this paper, we update the ring census using an automated algorithm to detect and track eddies over more than 23 years of satellite altimetry data (1993–2016) and calculate their main characteristics. While 140 rings spawned from the Agulhas Retroflection, their following splitting and merging resulted in 74 long-lived rings that crossed the Walvis Ridge and translated towards the west. Eventually, three rings reached the western boundary. For one of them, we use in situ measurements to document its interaction with the Brazil Current and two cyclonic eddies, which resulted in a current velocity increase by three times. Although already hypothesized, this interaction had not been demonstrated with in situ evidence until now.

Introduction

The heat flux towards the equator makes the South Atlantic unique among the other oceans (Bennett, 1978, Hastenrath, 1982). One of the reasons for such flux is the warm inflow at its southeastern corner provided by the leakage of Indian Ocean water through the Agulhas Retroflection. This warm water primarily feeds the surface route of the Atlantic Meridional Overturning Circulation (AMOC), which eventually returns to the North Atlantic to compensate the production and export of North Atlantic Deep Water (NADW) (Gordon, 1986, Gordon et al., 1992).

The Agulhas Retroflection regulates the heat and salt transfer between the Indian and Atlantic oceans playing a key role in the global climate system (de Ruijter et al., 1999, Matano and Beier, 2003). It is formed when the Agulhas Current, the western boundary current that closes the subtropical gyre in the South Indian Ocean, flowing southwestward along the Southeast African coast until the southernmost tip, detaches and bends upon itself in a tight anticlockwise curve back to the Indian Ocean. Part of Agulhas Current then makes its way into the Atlantic through filaments, direct leakage, and ring shedding (Bang, 1970, Gordon, 1985, Gordon, 1986, Lutjeharms and Van Ballegooyen, 1988a, Lutjeharms and Cooper, 1996). The retroflection may be shifted westward or eastward according to the zero wind stress curl latitude and the magnitude of the Agulhas Current transport. In response, the Agulhas leakage may increase or almost cease with impact on the thermohaline circulation on a global scale and hence on the earth's climate, as revealed by paleo records (de Ruijter and Boudra, 1985, Lutjeharms and Van Ballegooyen, 1988b, Biastoch et al., 2009, van Sebille et al., 2009, Beal et al., 2011, Nof et al., 2011). However, investigations have been contradictory in their conclusion on whether the leakage has shown an upward trend or no trend over the past decades (Biastoch et al., 2009, Beal et al., 2011, Garzoli et al., 2013, Le Bars et al., 2014).

The majority of the transfer of tropical and subtropical water masses from the South Indian Ocean to the South Atlantic occurs through the shedding of Agulhas rings at the retroflection (Gordon, 1986, Lutjeharms, 1996, de Ruijter et al., 1999). Juvenile Agulhas rings are large anticyclonic eddies with a mean diameter of 242 ± 38 km (Duncombe Rae, 1991) and depths that can reach 4.5 km (van Aken et al., 2003). There is no apparent periodicity on shedding events (Goni et al., 1997), and rings are formed on average every two months (van Ballegooyen et al., 1994, Lutjeharms, 1996, Dencausse et al., 2010). The warm and salty waters conveyed from the Indian Ocean contribute to increased evaporation rates in the South Atlantic and affect the AMOC (de Ruijter et al., 1999, Weijer et al., 2002, Beal et al., 2011). According to Casanova-Masjoan et al. (2017), due to a decay time of approximately 1 year, most of the property anomalies contained within the rings will be diffused into the South Atlantic subtropical gyre and hence advected to the north through the AMOC. The ratio between the fluxes of warm and salty water from the leakage and the cooler and low-salinity water from the Drake Passage defines the thermocline characteristics of the South Atlantic (Gordon and Haxby, 1990, Gordon et al., 1992). Considering waters warmer than 10 °C, the flux of Indian Ocean water into the Atlantic by Agulhas rings is estimated at 6.3 Sv (1 Sv = 10⁶ m³/s), and the fluxes of heat and salt at 0.045 PW and 78 Pg/year, respectively (van Ballegooyen et al., 1994). To put these numbers in perspective, the total meridional heat at mid-latitudes in the South Atlantic has been estimated as + 0.43 ± 0.08 PW (McDonagh and King, 2005), and the volume transport of the AMOC as ~11 Sv (Wunsch and Heimbach, 2013).

Just after shedding, the Agulhas rings present a distinct warm circular signature at the surface (Lutjeharms and Gordon, 1987) that is erased within a few weeks by intense heat exchange with the cooler atmosphere and mixing due to wind stress (Mey et al., 1990, Olson et al., 1992). Despite losing their surface thermal signature, the rings maintain a positive sea level anomaly (SLA) which can be tracked by satellite altimetry while translating westward across the South Atlantic Ocean (Fu and Zlotnicki, 1989, Gordon and Haxby, 1990, van Ballegooyen et al., 1994, Byrne et al., 1995, Gründlingh, 1995, Witter and Gordon, 1999, Schouten et al., 2000).

In the Cape Basin, the Agulhas rings move northwestward embedded in the Benguela Current creating an eddy corridor to the Walvis Ridge (Garzoli and Gordon, 1996, Goni et al., 1997). This ridge constitutes an obstacle to the translation of deep-reaching eddies into the Atlantic as they usually show a deflection of trajectory and reduction of drift speed while they manage to pass through the deepest valleys (Olson and Evans, 1986, Gordon and Haxby, 1990, Byrne et al., 1995, Arhan et al., 1999). Modeling studies reveal that baroclinic eddies can cross the ridge while the barotropic or near-barotropic cannot (Kamenkovich et al., 1996, Beismann et al., 1999, Matano and Beier, 2003). Early investigations of the period between 1993 and 1996 showed that about two-thirds of the Agulhas rings left the Cape Basin, traveling westward at speeds of 5–8 cm/s between 25°S and 35°S (Gordon and Haxby, 1990, Schouten et al., 2000).

The increase in surface water salinity due to evaporation and the wind forced mixing converts the Indian Ocean water brought to the Cape Basin into South Atlantic typical water (Gordon and Haxby, 1990, Olson et al., 1992). Nevertheless, thermostads and halostads with water characteristics from the Indian Ocean have been identified within rings both in the Cape Basin and west of the Walvis Ridge (McCartney and Woodgate-Jones, 1991, Byrne et al., 1995, Garzoli et al., 1999). This fact indicates that below the surface part of their original thermohaline characteristics may be preserved during translation into the South Atlantic Ocean.

The Agulhas rings are among the largest and most energetic eddies in the world's ocean. The energy contribution of a single ring is equivalent to 7% of the annual wind energy influx over the South Atlantic basin between 10°S and 45°S (Olson and Evans, 1986). Their propagation across the South Atlantic Ocean drives significant interannual variability in the upper layer transport at the northern branch of the subtropical gyre (over 2 Sv RMS for a mean flow on the order of 10 Sv) (Garnier et al., 2003).

In early studies, Geosat SLA maps were used to track Agulhas rings for long distances far from the retroflection region (Gordon and Haxby, 1990, Byrne et al., 1995). However, it is only after the TOPEX/Poseidon mission, that a sufficiently long time-series was produced to permit the tracking of an Agulhas ring from its formation site until 38°W (Schouten et al., 2000). More recently, automatic tracking algorithms have been used over merged multi-satellite altimetric maps to identify and follow ocean eddies (Chelton et al., 2007, Chelton et al., 2011). Long ring trajectories and the analysis of energy decay curves support the hypothesis that Agulhas rings might collide with the Brazil Current and trigger transport anomalies (Gordon and Haxby, 1990, Byrne et al., 1995, Nof, 1999, Azevedo et al., 2012), but no in situ evidence has been put forward so far.

In this paper, we present the results of a new census of Agulhas rings performed using an automated hybrid eddy detection algorithm (Halo et al., 2014) on 23 years of merged multi-satellite altimetry data (1993–2016). We focus on the 74 Agulhas rings with lifetimes longer than 360 days that crossed the Walvis Ridge while moving westward. Of these 74 rings, at least three eventually reached the western boundary. For one of them, first detected at 39.5°S, 14.6°E and hereafter referred to as Ring Lilian, we present direct in situ measurements at the Brazilian continental slope that document its interaction with the Brazil Current, and two cyclonic eddies. While moving across the South Atlantic, this ring captured some surface drifters, and also Argo floats that sampled its vertical temperature and salinity structure. The combination of different techniques and sampling instruments provides a robust and comprehensive view of the lifetime of the Ring Lilian since its origin until the eventual demise into the Brazil Current, after a 4-year journey across the South Atlantic.

The paper is organized as follows. In Section 2, we present the eddy detection and tracking algorithm, the satellite and in situ data sets, as well as the method for estimate volume and energy of the rings. Section 3 summarizes the results of the ring census, and discusses the interaction of Ring Lilian with the Brazil Current. Concluding remarks are then presented at Section 4.