1. Introduction
Mesoscale eddies are ubiquitous in global oceans and play a significant role in ocean circulation, heat–salt transport, and biological systems (Chelton et al. 2011; Dong et al. 2014; Z.-G. Zhang et al. 2014; McGillicuddy 2016). Subsurface eddies, usually characterized by a subsurface lens-shaped structure with nearly homogeneous water properties, are a special type of oceanic eddy (Dugan et al. 1982; Gordon et al. 2002; Z.-G. Zhang et al. 2017; Meunier et al. 2018). These eddies are usually termed intrathermocline eddies (ITEs; Dugan et al. 1982) due to the location of the lens core, submesoscale coherent vortices (SCVs; McWilliams 1985) due to their small horizontal scales, or mode-water eddies (McGillicuddy et al. 1999) because of the distinctive water properties trapped in the eddy. The horizontal scales of subsurface eddies range from the submesoscale (<10 km) to the mesoscale (∼100 km) and have been detected in the near-surface, intermediate, and near-bottom layers of the ocean (McWilliams 1985; Takikawa et al. 2005; Nan et al. 2017). Generally, subsurface eddies exhibit anticyclonic lens-shaped structures that depress the main thermocline and raise the seasonal thermocline, while cyclonic cases depress the seasonal thermocline and raise the main thermocline (McGillicuddy 2015; Zhang et al. 2022). The distinct characteristics of subsurface eddies include subsurface velocity maximum, weak (strong) stratification, and low (high) potential vorticity (PV) within the anticyclonic (cyclonic) eddy core (Meunier et al. 2018; Yang et al. 2019; McCoy et al. 2020). In addition, subsurface eddies usually have a long lifetime, which allows them to carry the original water mass properties far away from their formation regions (Armi et al. 1989; Lukas and Santiago-Mandujano 2001; Collins et al. 2013).
Subsurface eddies display weak or even no signals at the sea surface. Thus, it is difficult to identify them only through satellite remote sensing. Most of the subsurface eddies were discovered accidentally through shipboard surveys, Argo floats, gliders, and moorings (e.g., Armi et al. 1989; Takikawa et al. 2005; Baird and Ridgway 2012; Z.-W. Zhang et al. 2015; Nan et al. 2017; Zhang et al. 2022). Despite difficulties in observations, subsurface eddies have been reported in many parts of the World Ocean. Some of them are famous and have specific names, such as the Mediterranean water eddies (Meddies) in the North Atlantic (McDowell and Rossby 1978; Armi et al. 1989), the California undercurrent eddies (Cuddies) in the northeastern Pacific (Lukas and Santiago-Mandujano 2001; Collins et al. 2013), and the Kuroshio Extension intermediate-layer eddies (Kiddies) in the northwestern Pacific (Z.-W. Zhang et al. 2015; Li et al. 2017; Zhu et al. 2021). In addition, subsurface eddies have also been observed in other regions, such as the Sea of Japan (Gordon et al. 2002), the Tasman Sea (Baird and Ridgway 2012), the Bay of Bengal (Gordon et al. 2017), and the eastern equatorial Indian Ocean (Hu et al. 2022). Several formation mechanisms of subsurface eddies have been proposed in previous studies, including baroclinic instability of undercurrents (Jungclaus 1999), subduction of mode water at upper ocean fronts (Spall 1995; Thomas 2008; Li et al. 2017), restratification of preexisting surface anticyclonic eddies due to seasonal surface heating (Hogan and Hurlburt 2006), or local eddy–wind interactions (McGillicuddy 2015).
The South China Sea (SCS) is the largest semienclosed marginal sea in the northwestern Pacific Ocean. Satellite and in situ observations reveal that the SCS has frequent eddy activities (e.g., Wang et al. 2003, 2008; Chen et al. 2011; Nan et al. 2011; Chu et al. 2020), which play important roles in water mass, heat, salt transport, and regional biological systems (Lin et al. 2010; W.-Z. Zhang et al. 2015; He et al. 2018). However, compared to the surface-intensified eddies, only a few subsurface eddies have been reported in the SCS based on sporadic observations. For example, the subsurface eddies were accidentally observed by shipboard surveys, located in the intermediate layer of the Luzon Strait (Xie et al. 2011), below the thermocline of the southern SCS (Z.-X. Zhang et al. 2014), and within the thermocline of the southern SCS (Lin et al. 2017). Recently, Zhang et al. (2022) reported two oppositely rotating SCVs in the northeastern SCS based on high-resolution moored data and proposed a generation mechanism of current–topography interaction in the Luzon Strait. Sun et al. (2022) also observed a train of subsurface mesoscale eddies in the northeastern SCS and pointed out that these eddies may provide a novel route for the intermediate-layer water exchange between the SCS and Pacific. Nevertheless, our knowledge of periodic subsurface eddies in the SCS is still unclear.
This study reported an extraordinary subsurface anticyclonic eddy (SAE) east of the Xisha Islands in May 2021 based on satellite altimetry and hydrographic observations. This SAE exhibited significant surface signals and was different from the subsurface eddies previously reported in the SCS. We analyzed the three-dimensional structure of the SAE, explored its generation mechanism, and discussed the universality of such SAEs in the SCS through a long-term statistical analysis. The rest of this paper is organized as follows: section 2 introduces the data and methodology. Section 3 describes the characteristics of the lens-shaped SAE, investigates its origin and evolutionary process, and estimates its water transport. Section 4 presents the results of multiyear statistical analysis. Finally, discussion and summary are given in sections 5 and 6, respectively.
2. Data and methods
a. Hydrographic observations
A research cruise was conducted by the South China Sea Institute of Oceanology, Chinese Academy of Sciences in the northern SCS from 21 to 28 May 2021. Two sections crossing an anticyclonic eddy were carried out to investigate the eddy’s vertical structure and water properties (Fig. 1). Temperature and salinity profiles in the upper 2000 m with a vertical resolution of 1 m were measured by an SBE 911 conductivity–temperature–depth (CTD) instrument aboard the research vessel Shiyan-1. Vertical distributions of velocities were also obtained from 75-kHz ship-mounted acoustic Doppler current profilers (SADCPs), ranging from 16 to 608 m with a vertical interval of 8 m. The World Ocean Atlas 2018 (WOA18) monthly climatology data were used to calculate the temperature and salinity anomalies along the observation sections in May 2021.
Hydrographic observations in May 2021. Solid green dots represent the CTD stations along the A and B sections. Color shading shows the mean SLA during 21–28 May 2021. Dashed contour lines represent the closed SLA with 5-cm intervals associated with the eddy center. The 200-m isobath is indicated by the black solid line.
Citation: Journal of Physical Oceanography 53, 3; 10.1175/JPO-D-22-0106.1
Two Argo floats (WMO numbers: 2902709 and 2902711) operated west of the Luzon Strait when the eddy was shed from the Kuroshio Loop Current (KLC; Fig. 5c). The Argo profile data were downloaded from the Argo Data Management site. They were used to investigate the water mass properties within and outside of the eddy at the shedding stage during 8–15 March 2021. The profiles passed a series of quality controls, and only those flagged as good (the quality flag was 1) were used for the present study. In addition, the Argo climatology data were used to provide the basic characteristics of the Kuroshio water and the northeastern SCS water, which were downloaded from the Asia–Pacific Data-Research Center (APDRC).
b. Satellite observations
Daily sea level anomaly (SLA) and surface geostrophic velocity data were used to identify and track the mesoscale eddies. The datasets merge all available altimetry satellites and have a global grid with a 1/4° resolution, obtained from the Copernicus Marine Environment Monitoring Service (CMEMS).
The microwave and infrared (MW_IR) optimally interpolated sea surface temperature (OISST) product from the remote sensing systems (RSS), with temporal and spatial resolutions of daily and 9 km, was used to provide the refined structure of SST within and outside the target eddy (Fig. 5). The initial time of this high-resolution dataset was June 2002. To investigate the SST structure of the historical SAEs from 1993, another OISST product from NOAA Physical Sciences Laboratory (PSL) was also used in this paper. The NOAA OISST dataset has been archived daily from September 1981 to the present, with a spatial resolution of 1/4° global grid.
The cross-calibrated multi-platform (CCMP) gridded ocean surface (10 m) vector winds were used to calculate eddy-induced Ekman pumping. The spatial and temporal resolutions of this dataset are 0.25° × 0.25° and 6 h, respectively.
c. Other datasets
The ERA5 surface heat flux data were downloaded from the Climate Data Store (https://cds.climate.copernicus.eu/cdsapp#!/home). The net surface heat flux is calculated as the sum of the net shortwave radiation, net longwave radiation, surface latent heat flux, and surface sensible heat flux. In this study, the ERA5 hourly data on single levels from 1959 to the present were averaged daily with a spatial resolution of 1/4° global grid.
To investigate the evolution of the lens-shaped structure, two 1/12° ocean reanalysis datasets of the global ocean analysis and forecast system (Lellouche et al. 2018) were acquired from the CMEMS. One of the CMEMS products is GLOBAL_REANALYSIS_PHY_001_030, archived daily from 1 January 1993 to 31 December 2019. The other one is GLOBAL_ANALYSIS_FORECAST_PHY_001_024, archived daily from 1 January 2019 to the present. The CMEMS reanalysis datasets are based on the current real-time global forecasting CMEMS system and have assimilated altimeter data (SLA), satellite SST, sea ice concentration, and vertical in situ temperature and salinity profiles. These datasets are in good agreement with the observations and have been widely used for the study of oceanic mesoscale eddies (e.g., Qian et al. 2018; Potter et al. 2021; Wang et al. 2021). The variables used in this study include daily sea surface height (SSH), three-dimensional velocity, temperature, and salinity.
d. Methods
SLA and geostrophic currents derived from satellite altimetry are widely used to detect mesoscale eddies (Chen et al. 2011; Nan et al. 2017). As the mesoscale eddy can be defined as a vortex with closed SLA (streamline) contours, the innermost closed contour is taken as the center of an eddy. Once the eddy center is determined, the eddy edge is defined as the outermost closed SLA (streamline) contour around its center. The SLA difference between the eddy center and the edge is defined as the eddy amplitude. Taking one day as a step, the center of the same eddy is closest at two adjacent moments. When the amplitude is less than 2 cm, it is considered as the demise time of the eddy. Based on this criterion, the trajectory of the eddy can be tracked. Further details about eddy identification and eddy-tracking procedure can be found in Chen et al. (2011).
The mixed layer is characterized by nearly uniform properties throughout the layer, such as temperature and salinity. The mixed layer depth (MLD) is defined as the depth at which the potential density increase relative to the surface (10 m) equals the increase in surface potential density when the SST decreases by 0.5°C (de Boyer Montégut et al. 2004).
3. Observations of a lens-shaped SAE in the northern SCS
a. Hydrographic observations
Detected from the altimeter data, an anticyclonic eddy was observed east of the Xisha Islands in the northern SCS during 21–28 May 2021 (Fig. 1). The maximum SLA in the eddy center reached 32 cm, and the mean diameter of the eddy exceeded 200 km. Two sections of hydrographic observations across the anticyclonic eddy were carried out to investigate the vertical structure of the anticyclonic eddy. Interestingly, the eddy exhibited a lenticular vertical structure significantly different from that of the general surface-intensified anticyclonic eddy (Fig. 2). A double thermocline structure was observed in the eddy core between 20 and 200 m from section A, 113.4°–114.6°E (A04–A11), and section B, 16.7°–18.0°N (B11–B21). The water mass within the lens core was well mixed and characterized by a temperature of 22°–28°C and a salinity of 34.2–34.7 psu. The lenticular body, composed of nearly homogeneous water, depressed the lower thermocline (halocline/pycnocline) and raised the upper thermocline (halocline/pycnocline). Accompanied by the lens-shaped structure, the eddy’s swirl velocity displayed an anticyclonic subsurface-intensified structure, indicating that this was a typical SAE (Fig. 2c). The maximum velocity was found at ∼80 m and was larger than 0.5 m s−1. It is worth noting that section B does not seem to pass through the entire lenticular structure, as it is expected to extend farther south than B21. It is assumed the position where the velocity contour line equals 0 is the eddy center, and the lens structure is symmetrically distributed on both sides of the center. The complete lenticular structure along section B is obtained, that is, from 16.6° to 18.0°N. Therefore, the horizontal extent of the lens-shaped structure is 139 and 168 km in sections A and B, respectively.
(a) Temperature (°C) and (b) salinity (psu) along section A and section B. (c) Velocity (m s−1) perpendicular to the sections.
Citation: Journal of Physical Oceanography 53, 3; 10.1175/JPO-D-22-0106.1
Considering the distinctive water properties trapped in the SAE, the temperature and salinity anomalies along the isopycnals were calculated referencing the climatological mean values in May derived from WOA18. The lens-shaped structure is clear in Fig. 3. The eddy core was occupied by a thick body of high-temperature and high-salinity water, with its maximum anomalies reaching +1.0°C and +0.4 psu relative to the historical state, respectively. Figure 3c shows the vertical section of Rossby number (ζ/f). The Rossby number is mainly negative in SAE area, which reflects the characteristics of the anticyclonic eddy. In addition, the value of Rossby number in the lens core is basically less than −0.5, indicating that relative vorticity is as important as the planetary vorticity in the eddy region. Therefore, the relative vorticity cannot be ignored when calculating Ertel PV. The vertical distribution of PV is shown in Fig. 3d. Low PV water was trapped inside the lens core, which is a typical feature of the SAE. Large PV values were found near the sea surface and in the vicinity of the main thermocline, consistent with the position of strong stratification. The upper and lower bounds of the lenticular body could be defined by the convex isopycnal σθ = 1022 kg m−3 and the concave isopycnal σθ = 1025 kg m−3, respectively. The largest thickness of the lenticular body was approximately 175 m (from 24 to 199 m) near Station B15. Notably, the horizontal extent of the SAE was not completely consistent with that detected by satellite altimetry data. The size of the lens-shaped structure was smaller than the size of the anticyclonic eddy observed by the satellite, and it was distributed southeast of the anticyclonic eddy (Fig. 5d). This discrepancy between the location of the lens-shaped structure and the anticyclonic eddy was also reflected from the vertical distribution of velocity. The subsurface-intensified velocity only appeared in the southeastern part of the anticyclonic eddy, while its northwestern part still displayed a surface-intensified structure. This phenomenon suggests that, without additional in situ measurements, satellite altimetry cannot reflect the full structure of the subsurface eddies even if they have significant surface signals.
(a),(b) Temperature and salinity anomalies along the isopycnals compared to climatological mean values in May from WOA18. (c) Rossby number (ζ/f). (d) Potential vorticity (PV; 10−9 m−1 s−1). Black lines represent isopycnals.
Citation: Journal of Physical Oceanography 53, 3; 10.1175/JPO-D-22-0106.1
b. The origin and surface features of the SAE
The water properties in the SAE were sharply different from those in the surrounding areas. In particular, a maximum salinity exceeding 34.7 psu was found at ∼175 m at the bottom of the lens-shaped structure (Fig. 2b). This salinity was much saltier than the general maximum salinity (∼34.6 psu) in the SCS, indicating that the SAE may be not generated locally but from the Pacific Ocean with higher salinity. Unlike the ordinary subsurface eddies, this SAE displayed a significant signal at the sea surface. Thus, we can track its origin from satellite observations. The evolution of the SLA maps revealed that the SAE observed east of the Xisha Islands originated from the Luzon Strait and was shed from the KLC last winter (Fig. 4). This result was surprising because the lenticular structure of the Kuroshio shedding eddy has not been reported before. The generation of the SAE can be tracked to late October 2020, when the Kuroshio intruded into the SCS as a loop current, and the anticyclonic eddy was formed in the KLC. As time went by, the anticyclonic eddy gradually developed and reached its strongest extent in early March 2021, when the SLA in the eddy center exceeded 60 cm (Fig. 4e). During this period, the eddy moved slowly southwestward at a speed of 0.02 m s−1 until it completely separated from the KLC on 10 March 2021. Afterward, the eddy gradually decayed and propagated southwest along the continental slope at a speed of 0.08 m s−1 for approximately two months until the Xisha Islands blocked it in May 2021. Although KLC intrusion and its accompanying anticyclonic eddy shedding have been extensively investigated in recent decades (e.g., Z.-W. Zhang et al. 2017; Sun et al. 2020; Sun et al. 2021), the evolution of the Kuroshio shedding eddy to a lens-shaped SAE remains unclear.
(a)–(d) Maps of the SLA (shading; cm) and surface geostrophic current (vectors; m s−1) from December 2020 to May 2021. Black contour lines represent the SLA with 5-cm intervals associated with the eddy center. AE1 and AE2 represent the SAE at the stages of eddy shedding and cruise observations, respectively. (e) Time series of SLA at the eddy center. The dashed lines represent the shedding date of the eddy (10 Mar 2021) and the time of cruise observations (21–28 May 2021).
Citation: Journal of Physical Oceanography 53, 3; 10.1175/JPO-D-22-0106.1
According to the SST maps, the SAE had significant SST anomalies during its lifespan (Fig. 5). In the early stage, the SAE presented a warm-core structure related to the characteristics of high-temperature Kuroshio water. The SST in the eddy center was 0.8°C higher than that in the surroundings in the middle of December 2020. However, with the passage of time, the SAE gradually converted to a distinctive surface cold-core structure, which was different from the traditional warm-core anticyclonic eddies. The cold-core structure appeared in January 2021 and reached its strongest extent in March with a center SST of 23.9°C.
(a)–(d) Maps of SST (shading; °C) and surface geostrophic current (vectors; m s−1) from December 2020 to May 2021. Black contour lines represent the SLA with 5-cm intervals associated with the eddy center. The thick red contour line represents the eddy edge. The color bar of (a)–(c) ranges from 20° to 30°C, while the color bar of (d) ranges from 26° to 32°C. Solid pink and green dots in (c) denote the location of Argo float 2902711 and Argo float 2902709 during 8–15 Mar 2021, respectively. Solid dots in (d) represent the CTD stations, and the solid red dots represent the stations where the lens-shaped structure is located. (e) Time series of SST at the eddy center (solid red line), at the eddy edge (solid green line), and their difference (solid black line). The vertical dashed lines represent the shedding date of the eddy (10 Mar 2021) and the time of cruise observations (21–28 May 2021).
Citation: Journal of Physical Oceanography 53, 3; 10.1175/JPO-D-22-0106.1
To investigate the possible mechanism of forming the cold-core structure, we analyzed the change of surface heat flux during the study period (Fig. 6). The net surface heat flux in the northeastern SCS was negative from November 2020 to February 2021, indicating that the ocean surface was losing heat during this period. The SST in the northeastern SCS decreased continuously in the winter. However, compared to the eddy center, the water at the eddy edge was easily replenished by the low-latitude warm water from the westward jet of the Kuroshio and the Northwest Luzon Coastal Current (Figs. 5b,c). The SST gradient from the eddy edge to the eddy center was established in January 2021 and reached a maximum of 1.8°C in mid-March. Therefore, the formation of the cold-core structure was related to the winter air–sea interaction and the supply of horizontal advection. This cold-core anticyclonic eddy in winter 2020/21 was also reported by Sun et al. (2021). They pointed out that this cold-core event was attributed to the sustained entrainment supported by the warm water from the Kuroshio intrusion and the Northwest Luzon Coastal Current successively.
(a)–(d) Maps of net surface heat flux (SHF; shading; W m−2) from December 2020 to May 2021. Black contour lines represent the SLA with 5-cm intervals associated with the eddy center. The thick red contour line represents the eddy edge. (e) Time series of SHF at the eddy center (solid red line), at the eddy edge (solid green line), and their difference (solid black line). The vertical dashed lines represent the shedding date of the eddy (10 Mar 2021) and the time of cruise observations (21–28 May 2021).
Citation: Journal of Physical Oceanography 53, 3; 10.1175/JPO-D-22-0106.1
As the season changed from winter to spring, the net surface heat flux changed from negative to positive, indicating that the atmospheric forcing reverted to net heating. The SST gradient between the eddy center and the edge began to decrease with the continuous increase of SST. It is worth noting that the net surface heating in the eddy center was a little stronger than in the surroundings during the spring. However, although the surface cold-core structure weakened, it did not disappear. The maintenance of the surface cold-core structure seems to be related to the vertical lens-shaped structure of the SAE. Although the anticyclonic eddy had separated from the KLC and SST had increased with the seasonal transition, the surface cold core could still be seen in the region of the lens-shaped structure observed in May 2021 (Fig. 5d).
c. Evolution of the lens-shaped structure
Satellite observations could only capture the surface signals of the SAE. The evolution of the vertical lens-shaped structure was still unclear. Fortunately, two Argo floats (WMO numbers: 2902709 and 2902711) were deployed in the northeastern SCS and were located just inside and outside the eddy at the shedding stage (Fig. 5c). Thus, the vertical structure of the anticyclonic eddy in the northeastern SCS could be investigated from Argo observations. The SAE at the stages of eddy shedding and cruise observations were referred to as AE1 and AE2, respectively.
Figure 7 shows the temperature, salinity, and potential density profiles within and outside the eddy at the two stages. At the shedding stage, the anticyclonic eddy had not yet formed a lens-shaped structure. Surface cooling in winter caused strong convective mixing in the northeastern SCS. The thermocline and mixed layer within the eddy were much deeper than the surrounding areas. The well-mixed water in the eddy was characterized by a temperature of ∼24.5°C, a salinity of ∼34.5 psu, and a density of ∼1023.1 kg m−3. This was the source of the low PV water observed in May 2021. In particular, the cold-core structure observed by the satellite was proven to be as deep as 70 m at stage AE1. The water within the anticyclonic eddy was also saltier than the surrounding water in the upper 90 m, indicating that the eddy was surrounded by warmer and fresher water in the upper layer. Consequently, the water density in the eddy center was greater above 80 m and lower below 80 m than in the surroundings. According to geostrophic equilibrium [Eqs. (4) and (5)], the horizontal pressure gradient is mainly determined by the density difference of seawater. Therefore, the pressure gradient inside and outside the eddy first increases and then decreases with depth, reaching the maximum at about 80 m. Based on the thermal wind relations, the geostrophic velocity at the eddy edge that perpendicular to the connecting line of two Argo floats was calculated (Fig. 7d). It is clear that the eddy exhibited a subsurface-intensified structure. The maximum velocity appeared at ∼80 m and exceeded 1.0 m s−1 at the shedding stage.
Vertical profiles of (a) temperature (°C), (b) salinity (psu), and (c) potential density (kg m−3) within and outside the SAE at the two stages. The profiles of AE1 and AE2 are obtained from the Argo observations (float 2902709 and float 2902711; Fig. 5c) and the CTD measurements (Station B10 and Station B12; Fig. 1), respectively. (d) The vertical profiles of geostrophic velocity (m s−1) at the edges of AE1 and AE2. The geostrophic velocity is calculated based on the temperature and salinity profiles within and outside of the eddy.
Citation: Journal of Physical Oceanography 53, 3; 10.1175/JPO-D-22-0106.1
After shedding from the KLC, the eddy propagated southwest along the continental slope and decayed rapidly. When observed in May 2021, the SLA at the eddy center decreased from 60 to 32 cm. The depth of the main thermocline (halocline/pycnocline) at the eddy center was uplifted by nearly 50 m (Fig. 7a). The eddy swirl velocity continuously decreased due to the dissipation of the eddy. Nevertheless, the upper cold, salty core of the anticyclonic eddy seems to be well preserved, maintaining the subsurface-intensified structure. The depth of the maximum velocity was still ∼80 m, while the maximum velocity was reduced to 0.44 m s−1 (Fig. 7d). The geostrophic velocity calculated by temperature and salinity was slightly smaller than the absolute velocity measured by SADCP (Fig. 2c). This is because the absolute velocity also contained the Ekman current and other high-frequency processes. As the season changed from winter to spring, the sea surface was heated up, and the stratification of the upper ocean gradually strengthened. The cold, salty core in the upper layer caused a higher density inside the eddy than in the surrounding areas. In other words, it will drive a cyclonic geostrophic flow anomaly in the upper part of the anticyclonic eddy, so the newly formed isopycnal surface was dome-shaped in the eddy center. When the domed seasonal thermocline formed near the surface that capped the original mixed layer water, the Kuroshio shedding eddy was ultimately isolated from the surface and became a lens-shaped SAE.
The CMEMS analysis data were used to explore the complete evolutionary process of the SAE. First, validation of CMEMS analysis is shown in Fig. 8. The SST maps from the CMEMS data were similar to those from remote sensing images (Figs. 5c,d). The development of the anticyclonic eddy and its surface cold-core structure were well reproduced in the CMEMS product. In addition, the lens-shaped and subsurface-intensified structure of the eddy on 25 May was also consistent with that observed by the cruise survey (Fig. 2), which validated the reliability of the CMEMS dataset.
Data validation. (a),(b) Maps of SST (shading; °C) and surface current (vectors; m s−1) on 10 Mar 2020 and 25 May 2021, based on the CMEMS product. Black contour lines represent the SSH with 10-cm intervals associated with the eddy center. The thick pink contour line represents the eddy edge. (c) Temperature (°C) and (d) salinity (psu) along the cross sections on 25 May 2021 (blue lines in Fig. 8b). (e) Velocity (m s−1) perpendicular to the cross sections.
Citation: Journal of Physical Oceanography 53, 3; 10.1175/JPO-D-22-0106.1
Figure 9 shows the temporal variation of temperature within and outside of the eddy and their difference. Due to the convergence of the anticyclonic eddy, the isopycnal surfaces in the eddy center were depressed and much deeper than the surrounding areas. The development and intensity of the eddy could also be reflected in the fluctuation of isopycnals. The eddy was generated at the end of October 2020 and developed to its strongest extent in March 2021. As the eddy detached from the KLC, its strength weakened rapidly until May. Afterward, the eddy maintained and existed until November 2021. This anticyclonic eddy survived for more than a year, which is rare in the SCS. Due to its long lifetime, the water trapped in the eddy experienced sea surface cooling in autumn–winter and heating in spring–summer.
Temporal variation of temperature (shading; °C) in the (a) eddy center and (b) eddy edge. Black lines represent isopycnals. (c) Temperature difference between the eddy center and eddy edge. The pink and blue lines denote the MLD in the eddy center and edge, respectively. The vertical dashed lines represent the shedding date of the eddy (10 Mar 2021) and the time of hydrographic observations (21–28 May 2021).
Citation: Journal of Physical Oceanography 53, 3; 10.1175/JPO-D-22-0106.1
Regarding the evolution of the lens-shaped structure, the upper cold-core structure first appeared in January, which was related to winter sea surface cooling and the replenishment of low-latitude warm water at the eddy edge. The cold-core structure generated a geostrophic flow opposite to the clockwise direction of the anticyclone and decelerated it in the upper part of the eddy, thus leading to the subsurface-intensified structure. In winter, the MLD in the eddy center was deeper than the surrounding areas, causing a thicker body of nearly homogeneous water inside the eddy. After shedding from the KLC, the MLD in the eddy center decreased rapidly from 130 to 30 m in only 20 days as the sea surface heated. Since 30 March, the MLD inside and outside the eddy was almost the same, and the MLD inside the eddy was even shallower than that outside the eddy. Rapid shoaling of the MLD in the eddy center caused its original thick and homogeneous water to be well preserved below the mixed layer. Although the temperature difference between the eddy center and the edge decreased above the MLD, it remained well below the MLD. The existence of the upper cold core maintained the subsurface-intensified structure. With the continuous warming of the upper ocean, the isopycnal σθ = 1022 kg m−3 in the eddy center began to appear in early May, indicating the emergence of the seasonal thermocline, that is, the formation of a lens-shaped structure. Note that the water trapped in the Kuroshio shedding eddy was saltier than the surrounding areas, which also contributed to the subsurface-intensified structure by changing the density gradient (Fig. 7b).
d. High-salinity water transport
When shedding from the KLC, the anticyclonic eddy carried a large amount of high-salinity water southwestward into the interior of the SCS, which could influence the regional heat–salt balance and ecological processes (Wang et al. 2021; Z.-W. Zhang et al. 2017). Although the reanalysis data reproduced the evolution of the lens-shaped structure well, the simulated salinity was smaller than the actual observed value (Figs. 2b, 8d). Therefore, the high-salinity water transport caused by this SAE was estimated based on the Argo profiles and hydrological CTD observations. According to Fig. 10a, the high-salinity water trapped in the eddy was a mixture of Kuroshio water and SCS water, which was mainly distributed above the isopycnal of 1025.5 kg m−3. Over time, the proportion of Kuroshio water in the SAE decreased. Nevertheless, the water properties in the lens core of AE2 were very close to those in AE1 and exhibited a change from the core, edge, and outside of the lenticular body. These results indicated that the proportion of Kuroshio water decreased gradually from the eddy center to the outside. If we only consider isopycnal mixing, the percentage of Kuroshio water a trapped within the eddy can be generally estimated using SAE = aSKur + (1 − a)SSCS, where SAE, SKur, and SSCS denote the salinity of the SAE, Kuroshio, and northern SCS water, respectively. The basic characteristics of Kuroshio water and northeastern SCS water are obtained from Argo climatology data in spring. The water properties trapped in AE1 and AE2 are from Argo (VMO: 2902711) and CTD (Stations B13–B20) observations, respectively. Substituting the salinities in Fig. 10a into this formula, the percentages of Kuroshio water trapped in AE1 and the lens core of AE2 are shown in Fig. 10b.
(a) T–S diagrams of water mass within the SAE at the two stages. Black and gray lines show the basic characteristics of Kuroshio water (black box in the inset) and northeastern SCS water (NSCS, gray box in the inset), respectively, based on Argo climatology data in spring. The red line represents the T–S curve within AE1 obtained from Argo 2902711. Blue, green, and yellow dots denote the water properties in the core (Stations B13–B20), at the edge (Stations A4–A10, B11–B12, and B21), and outside (Stations A1–A3, A11–A15, and B1–B10) of the lens-shaped structure in AE2, respectively. The pink line is the mean T–S curve within the lens core of AE2. Thin gray lines represent isolines for potential density. (b) The percentages of Kuroshio water trapped in AE1 and the lens core of AE2 (see the formula in section 3d).
Citation: Journal of Physical Oceanography 53, 3; 10.1175/JPO-D-22-0106.1
Assuming that AE1 was a regular cylinder, its surface area and depth at 1025.5 kg m−3 isopycnal surface within the eddy can be obtained from satellite altimetry data and Argo profiles, respectively. Therefore, the volume of high-salinity water trapped in AE1 was calculated to be 1.70 × 1013 m3. Considering the percentage of Kuroshio water in each layer, the volume of Kuroshio water in AE1 was estimated to be 1.02 × 1013 m3, accounting for 60% of the mixed high-salinity water. According to Qu (2000), the annual mean Luzon Strait transport (LST) in the upper layer is 3.0 Sv (1 Sv = 106 m3 s−1), which is equivalent to an annual transport volume of 9.46 × 1013 m3. Therefore, the Kuroshio water transport caused by this shedding eddy accounted for ∼11% of the annual-mean LST in the upper layer. This result is similar to that estimated by Z.-W. Zhang et al. (2017), who pointed out that the Kuroshio water transport caused by the KLC shedding eddy accounts for 6.8%–10.8% of the upper-layer LST.
When observed in May 2021, the anticyclonic eddy had moved nearly 600 km southwest from its shedding area and had almost halved in strength. The lens-shaped structure was distributed southeast of the anticyclonic eddy, as derived from the satellite observations (Fig. 5d). Based on the existing CTD observations, the horizontal extent of the lens-shaped structure was 139 and 168 km in sections A and B, respectively. It is assumed that this is a lens prism with the base of section A and the side of section B. The base area of the lens prism in section A (1022–1025.5 kg m−3) was calculated to be 2.59 × 107 m2. By multiplying the length of the side edge in section B (168 km), the volume of high-salinity water trapped within the lens prism was obtained to be 4.35 × 1012 m3. Considering the percentage of Kuroshio water in each layer, the volume of Kuroshio water in AE2 was estimated to be 2.21 × 1012 m3, accounting for 51% of the mixed high-salinity water. Compared with the shedding stage, the volume of Kuroshio water trapped in AE2 accounted for only 21.7% of that in AE1. This was due to the decay of the shedding eddy as it moved southwestward, and its intensity and radius were reduced by almost half. However, the proportion of Kuroshio water in the lens core changed little during the two stages. It indicates that the water properties trapped in the core of the SAE were well preserved even after long-distance propagation.
4. Statistical analysis
According to previous studies, KLC intrusion and its accompanying anticyclonic eddy shedding occur nearly annually, especially in boreal fall and winter (e.g., Z.-W. Zhang et al. 2017; Sun et al. 2020; Wang et al. 2021). This study reveals the linkage between the Kuroshio shedding eddy and the lens-shaped SAE, providing a new perspective for the formation of subsurface eddies in the northern SCS. The evolution of the SAE from a preexisting surface eddy due to seasonal change was also reported in the Sea of Japan/East Sea (Hogan and Hurlburt 2006; Lee 2018), indicating that this may be a common phenomenon in marginal seas. Whether the historical KLC shedding eddies eventually transited to lens-shaped structures and if this is a periodic SAE in the northern SCS remains an open question.
To investigate the evolution of shedding eddies, we first examined all prominent eddy shedding events based on 29-yr (1993–2021) satellite altimetry data. Anticyclonic eddies generated in winter and shed from the KLC were considered. The eddy’s shedding time is defined as the time when the eddy’s SLA contours are totally disconnected from those of the KLC. The eddy’s lifespan is defined as the duration between the shedding time and demise time when the eddy amplitude is lower than 2 cm. Eddies with a lifespan shorter than 3 weeks were excluded. Based on these criteria, 25 shedding anticyclonic eddies were identified from 1993 to 2021 (Fig. 11). Except for a few years, the eddy shedding event occurred once a year. Similar results were also reported by Z.-W. Zhang et al. (2017), who identified 19 prominent KLC eddy shedding events between October 1992 and October 2014 based on altimeter SSH and SLA maps. The strength of eddies at the shedding time and their lifespans were different from year to year. It seems that the lifespan of the eddy was not directly related to its strength at shedding time, which may have also been affected by the topography and background currents in the SCS. In particular, the eddy in 2021 was the most unique, with the largest center SLA and nearly the longest lifespan. Because the final formation of the lens-shaped structure was related to the capping of the seasonal thermocline, the survival time of shedding eddies in the SCS may be an important condition for the evolution of SAEs. Nearly half of the Kuroshio shedding eddies (12/25) were generated in winter and persisted to the following May. The 28°C isotherm usually appears at the beginning of May (Fig. 13a), which defines the formation of the seasonal thermocline. Using the CMEMS reanalysis data, we found that these eddies in May all exhibited a lens-shaped structure. The earliest time for the formation of SAE was 27 April (case in 2010), when the seasonal thermocline began to appear, separating the shedding eddy from the surface. Notably, the lens-shaped SAEs in the late spring of 2014 and 2018 were also observed by cruise surveys (J. Liu et al. 2020) and Argo 2901480 (Fig. S1 in the online supplemental material), respectively, which further confirmed the reliability of the reanalysis results.
Statistical analysis of Kuroshio eddy shedding events. (a) The SLA at the eddy center when it was shed from the KLC. The dashed line represents the mean SLA of the shedding eddies. Eddies that evolved into SAEs are filled with red. (b) The lifespan of the shedding eddy. The bottom and the top of the bar represent the eddy shedding time and its demise time, respectively. The length of the bar denotes the lifespan of the shedding eddy in the SCS. The two dashed lines correspond to 1 Jan and 1 May.
Citation: Journal of Physical Oceanography 53, 3; 10.1175/JPO-D-22-0106.1
To explore the commonality of these SAEs, a composite analysis of their vertical structures was carried out. Take the vertical structure of each SAE on the day when the lens core was first separated from the surface, normalize it on the x axis using the eddy radius, and then combine all the normalized structures together. The composite results of 12 SAEs are shown in Fig. 12. It exhibits an obvious lens-shaped structure, with a subsurface-intensified velocity at ∼75 m. The water trapped in the lens core is characterized by high salinity and low PV, which is different from the surrounding water. The lens-shaped structure is asymmetric, and its east side is stronger than the west. This is because there is usually an accompanying cyclonic eddy on the east side of the SAE, which strengthens the southward velocity (Fig. 4d). Statistical results indicate that the transition of the Kuroshio shedding eddy to a lens-shaped SAE is a common phenomenon in the northern SCS. The evolution of the SAE is related to the generation and demise time of the shedding eddy. If the eddy is generated in winter and survives until the seasonal thermocline appears in late spring, it will form a lens-shaped structure. In fact, the evolution of the SAE started as early as the appearance of the surface cold core, but it was not until the formation of seasonal thermocline that the eddy was separated entirely from the sea surface. Notably, the occurrence probability of SAE has increased since 2010, and they occur nearly annually. This will be beneficial to the targeted field observations of subsurface eddies in the future.
Vertical sections of the composite SAEs based on CMEMS reanalysis. (a) Temperature (°C). (b) Salinity (psu). (c) Meridional velocity (m s−1). (d) Potential vorticity (10−9 m−1 s−1). Black lines in (d) represent isopycnals. The x axis is normalized by eddy radius R.
Citation: Journal of Physical Oceanography 53, 3; 10.1175/JPO-D-22-0106.1
Although most subsurface eddies display weak or even no signals at the sea surface (e.g., McCoy et al. 2020; Zhu et al. 2021; Sun et al. 2022), the SAE observed in spring 2021 exhibited significant surface signals with positive SLA and negative SST anomaly (Figs. 4, 5). The surface cold-core structure seems to be a typical feature of this kind of SAE (Fig. 13). Based on the strength of the eddies at their shedding time (Fig. 11a), that is, whether the SLA at the eddy center exceeds the average SLA, we divided these 12 SAEs into strong SAEs and weak SAEs. From generation to dissipation, the shedding eddies experienced SST decreases in winter and SST increases in spring. In the early stage, the eddies had a warm-core structure similar to the traditional anticyclonic eddy. However, with the change of seasons, the SST anomaly between the eddy center and edge gradually changed from a positive value to a negative value, especially in the stage of rapid sea surface warming from late February to late May. The appearance of the cold-core structure seems to be related to the different water properties within and outside the eddy. The cold water generated in winter was better preserved in the anticyclonic eddy, contrasting with the surrounding warm waters from low latitudes. More importantly, the cold-core structure was conducive to the transition of eddies from surface-intensified to subsurface-intensified structures. The upward doming of the seasonal thermocline was also beneficial for maintaining the surface cold-core structure. The cold-core structure in strong SAEs appeared earlier and stronger than that in weak SAEs. This is probably because the winter cold water mass trapped in strong eddies is better preserved, which strongly contrasts with the surrounding warm water.
(a) Time series of SST at the center of 12 SAEs. (b),(c) The SST anomaly between the eddy center and eddy edge for weak SAEs and strong SAEs, respectively. The solid black lines in (a)–(c) represent their average values. The black vertical dashed lines indicate 20 Feb and 20 May. The NOAA OISST data were processed by a 7-day moving average.
Citation: Journal of Physical Oceanography 53, 3; 10.1175/JPO-D-22-0106.1
5. Discussion
Statistical analysis confirms that the evolution of Kuroshio shedding eddy to a lens-shaped SAE with seasonal changes is a common phenomenon in the SCS. It is natural to ask whether this seasonal evolution mechanism can be applied to all the anticyclonic eddies generated in winter. To answer this question, we found a similar anticyclonic eddy in the SCS that was generated in winter and survived until spring, and then examined its evolution. The result is shown in Fig. 14. This anticyclonic eddy was generated east of the SCS in January 2010, then propagated westward, and finally dissipated in the central SCS in April 2010. Although this eddy’s generation and demise time was similar to that of the Kuroshio shedding eddy, it had not evolved into a subsurface lens-shaped structure. This result indicates that not all the anticyclonic eddies generated in winter will evolve into subsurface lens-shaped structures with seasonal changes. Compared to the Kuroshio shedding eddy, this eddy was generated in a lower latitude area with higher SST in winter. It propagated westward along the same latitude and exhibited a surface warm-core structure. More importantly, there was no significant difference in the water properties inside and outside the eddy. The temperature inside the eddy was even warmer than the surroundings. Therefore, it cannot evolve to a lens-shaped structure. As for Kuroshio shedding eddies, they are generated in the northeast of the South China Sea, where the SST is lower in winter due to the stronger air–sea interaction. Accordingly, they have a deep mixed layer and contain a thick body of cold and low PV water. In addition, the Kuroshio shedding eddies propagate southwestward along the continental slope, moving from high to low latitude. Therefore, the Kuroshio shedding eddies are more likely to form a cold-core structure in the upper layer. Moreover, the water trapped in the Kuroshio shedding eddies originates from the western Pacific, which is much saltier than the surrounding SCS waters. The cold, salty core will generate a cyclonic geostrophic flow anomaly, which is conducive to the formation of subsurface-intensified velocity and the lens-shaped structure. Similar evolution processes also occurred in the Sea of Japan/East Sea (Hogan and Hurlburt 2006; Lee 2018). This result indicates that this seasonal evolution mechanism is not applicable to all anticyclonic eddies, but is more like a unique mechanism for those eddies with anomalous water properties.
Development of an anticyclonic eddy in the central SCS in 2010 based on the CMEMS reanalysis. (a)–(c) Maps of SST (shading; °C) on 28 Jan 2010, 15 Mar 2010, and 10 Apr 2010. Black contour lines represent the SSH with 5 cm intervals. Blue lines show the cross sections passing the eddy center. (d)–(f) Temperature (°C) of the cross sections passing through the eddy center in three stages. (g)–(i) As in (d)–(f), but for salinity (psu). (j)–(l) As in (d)–(f), but for meridional velocity (m s−1).
Citation: Journal of Physical Oceanography 53, 3; 10.1175/JPO-D-22-0106.1
Previous studies have revealed that local eddy–wind interaction can create lens-shaped disturbances in the thermocline (McGillicuddy 2015; Gordon et al. 2017). Eddy-induced Ekman pumping is typically caused by the curl of the surface stress due to the relative motion of the surface wind and eddy current, which generates Ekman upwelling in anticyclonic eddies and downwelling in cyclonic eddies with vertical velocities on the order of 0.1 m day−1 (Gaube et al. 2015). To check the effect of eddy–wind interaction on the formation of SAE, we also calculated the eddy-induced Ekman pumping during the period of SAE evolution in 2021 (Fig. S2). In the winter of 2020, the northeast monsoon produced a strong negative wind stress curl in southwest Taiwan, promoting the development of the anticyclonic eddy. The maximum downward velocity caused by eddy-induced Ekman pumping at the eddy center was nearly 6.0 m day−1. With the strengthening and moving of the eddy, the eddy-induced Ekman pumping velocity at the eddy center changed from negative to positive value since February 2021. The average vertical velocity in March–April 2021 was about 0.2 m day−1. It should be noted that the strength of the anticyclonic eddy weakened rapidly since it separated from the KLC in March. During the process of eddy decay, the sense of the vertical motions was opposite to that during formation/intensification. Relaxation of the density perturbations associated with eddy decay resulted in upper-ocean upwelling within anticyclonic eddies (McGillicuddy 2016). At this stage, the vertical upward velocity caused by the eddy-induced Ekman pumping could be an important factor that contributes to the doming of the near-surface isopycnals and the maintenance of the surface cold-core structure. Therefore, the eddy-induced Ekman pumping may also be a potential mechanism accounting for the formation of SAE.
Most of the subsurface eddies have weak or even no signals at the sea surface because of their small horizontal scales and the deep location of the lens cores. The SAE reported in this study has a diameter of more than 100 km and presents significant sea surface signals, thus it will have an important impact on local air–sea fluxes like surface-intensified mesoscale eddies (Leyba et al. 2017; Liu et al. 2018; Y. Liu et al. 2020). Besides, this kind of SAE has a cold-core structure in the upper layer, so the effect of air–sea interaction may also be different from the traditional warm-core anticyclonic eddies. In addition, subsurface eddies can trap and transport the original water masses to remote regions due to the isolation of their lens cores and long lifespans (Armi et al. 1989; Lukas and Santiago-Mandujano 2001; Sun et al. 2022). As this type of SAE originates from the KLC, a substantial amount of Kuroshio water is trapped inside the eddy and then brought to the interior of SCS, which can significantly influence the local heat–salt balance. Furthermore, this SAE may play an important role in biogeochemical processes. Previous studies have revealed that the upward doming of the seasonal thermocline can deliver nutrients into the euphotic zone and greatly enhance primary production (Ledwell et al. 2008; Eden et al. 2009; McGillicuddy 2016). Moreover, some studies also pointed out that there was a multipolar structure of ageostrophic secondary circulation at the periphery of SAE (Barceló-Llull et al. 2017), and a high concentration chlorophyll-a ring appeared around the SAE during the transformation of anticyclonic eddy to the SAE (Lee 2018). This study reports a periodic SAE in the northern SCS, which will benefit targeted observations in the future.
6. Summary
Subsurface eddies are a special type of oceanic eddy that display a lens core and maximum velocity in the subsurface layer. In May 2021, an extraordinary SAE with a diameter of more than 100 km was captured in the northern SCS by in situ observations. The SAE was located between 20 and 200 m, with a doming of the seasonal thermocline and bowling of the main thermocline. The maximum velocity of the eddy was larger than 0.5 m s−1 and appeared at a depth of ∼80 m. Compared to the climate state, the lens core was occupied by well-mixed water with high temperature, high salinity, and low PV. The core properties of the SAE strongly contrast with surrounding waters, indicating that the SAE was not locally generated but from remote regions.
Satellite images showed that this SAE had significant surface signals with positive SLA and negative SST anomaly. Eddy track revealed that this SAE originated from the Luzon Strait and was generated in the KLC in winter 2020. After several months of development, the eddy was shed from the KLC in March 2021, and then propagated southwest along the continental slope until it was blocked by the Xisha Islands in May 2021. The anticyclonic eddy exhibited a distinct surface cold-core structure since January 2021, which was different from the traditional warm-core anticyclonic eddies. The appearance of the cold-core structure was related to the continuous sea surface cooling in winter and the replenishment of low-latitude warm water at the eddy edge.
Argo profiles and CMEMS analysis data revealed the vertical structure of this eddy. A deep mixed layer in winter caused a thick body of nearly homogeneous water inside the eddy, which was the source of low PV water trapped in the lens-shaped structure. Compared to the surrounding water, the well-mixed water in the eddy center was colder and saltier in the upper layer, thus resulting in a subsurface maximum velocity through geostrophic adjustment. As the season changed from winter to spring, the MLD in the eddy center decreased rapidly. The shoaling of the MLD in the eddy center caused its original thick and homogeneous water to be well preserved below the mixed layer. With the continuous warming of the upper ocean, a domed seasonal thermocline gradually appeared near the surface that capped the original homogeneous water. The anticyclonic eddy was ultimately isolated from the surface, forming the lens-shaped SAE with a subsurface maximum velocity. The evolution of SAE is related to the anomalous water properties inside the eddy and the seasonal change of sea surface heat flux, which is schematically summarized in Fig. 15.
Schematic diagram of SAE evolution. The shading color denotes the seawater temperature. The length of the green arrow indicates the magnitude of the pressure gradient.
Citation: Journal of Physical Oceanography 53, 3; 10.1175/JPO-D-22-0106.1
When shedding from the KLC, the anticyclonic eddy carried a large amount of high-salinity water to the interior of SCS, which is equivalent to 11% of the annual-mean LST in the upper layer. The shedding eddy decayed rapidly as it moved southwestward, and its intensity was almost reduced by half when observed in May 2021. However, compared to the shedding stage, the water properties trapped in the lens core were well preserved even after propagating over a long distance.
Statistical analysis revealed that the transition of KLC shedding eddies to lens-shaped SAEs is a common phenomenon in the northern SCS. Nearly half of the Kuroshio shedding eddies (12/25) were generated in winter and persisted to the following May during 1993–2021. These eddies eventually evolved into SAEs in late spring. The occurrence probability of SAEs has increased since 2010, occurring nearly annually. The surface cold-core structure seems to be a typical feature of these SAEs, which could be utilized as an index to identify SAEs in the SCS. This study reports a periodic SAE and elucidates the underlying processes of its generation and evolution, which provides new knowledge for subsurface eddies in the SCS and will benefit targeted observations in the future.
Acknowledgments.
Data and samples were collected on board R/V Shiyan-1 implementing the open research cruise NORC2021-302 supported by NSFC fund and NSFC Shiptime Sharing Project (42090042, 42049907, and 41976024). This study is supported by the Southern Marine Science and Engineering Guangdong Laboratory (Guangzhou) (2019BT02H594), the Chinese Academy of Sciences (XDB42010305, 133244KYSB20190031, SCSIO202201, and SCSIO202204), and the Open Project Program of State Key Laboratory of Tropical Oceanography (LTOZZ2101, LTOZZ2203). We thank the data archive support from the National Earth System Data Center, National Science and Technology Infrastructure of China (http://www.geodata.cn). The numerical computation is supported by the High Performance Computing Division in the South China Sea Institute of Oceanology.
Data availability statement.
The voyage observation data can be obtained from the South China Sea Ocean Data Center (http://data.scsio.ac.cn/metaData-detail/1496751152051986432). The SLA and surface geostrophic velocity data are available at the CMEMS (https://resources.marine.copernicus.eu/product-detail/SEALEVEL_GLO_PHY_L4_MY_008_047/DATA-ACCESS). The MW_IR OISST data and CCMP vector winds were derived from the RSS (https://data.remss.com/SST/daily/mw_ir/; https://data.remss.com/ccmp/v02.1.NRT/). The NOAA OISST data were provided by the NOAA PSL (https://psl.noaa.gov/thredds/catalog/Datasets/noaa.oisst.v2.highres/catalog.html). The ERA5 surface heat flux data were downloaded from Climate Data Store (https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-single-levels?tab=form). The WOA18 data were downloaded from https://www.nodc.noaa.gov/OC5/woa18/. The Argo profile data were downloaded from Argo Data Management (https://www.ocean-ops.org/board/wa/InspectPtfModule?ref=2902711; https://www.ocean-ops.org/board/wa/InspectPtfModule?ref=2902709). The Argo seasonal climatology data were obtained from APDRC (http://apdrc.soest.hawaii.edu/data/data.php). The global ocean reanalysis and forecast products were acquired from the CMEMS (https://resources.marine.copernicus.eu/product-detail/GLOBAL_ANALYSIS_FORECAST_PHY_001_024; https://resources.marine.copernicus.eu/product-detail/GLOBAL_MULTIYEAR_PHY_001_030/).
REFERENCES
Armi, L., D. Hebert, N. Oakey, J. F. Price, P. L. Richardson, H. T. Rossby, and B. Ruddick, 1989: Two years in the life of a Mediterranean salt lens. J. Phys. Oceanogr., 19, 354–370, https://doi.org/10.1175/1520-0485(1989)019<0354:TYITLO>2.0.CO;2.
Baird, M. E., and K. R. Ridgway, 2012: The southward transport of sub-mesoscale lenses of Bass Strait Water in the centre of anti-cyclonic mesoscale eddies. Geophys. Res. Lett., 39, L02603, https://doi.org/10.1029/2011GL050643.
Barceló-Llull, B., E. Pallàs-Sanz, P. Sangrà, A. Martínez-Marrero, S. N. Estrada-Allis, and J. Arístegui, 2017: Ageostrophic secondary circulation in a subtropical intrathermocline eddy. J. Phys. Oceanogr., 47, 1107–1123, https://doi.org/10.1175/JPO-D-16-0235.1.
Chelton, D. B., M. G. Schlax, and R. M. Samelson, 2011: Global observations of nonlinear mesoscale eddies. Prog. Oceanogr., 91, 167–216, https://doi.org/10.1016/j.pocean.2011.01.002.
Chen, G., Y. Hou, and X. Chu, 2011: Mesoscale eddies in the South China Sea: Mean properties, spatiotemporal variability, and impact on thermohaline structure. J. Geophys. Res., 116, C06018, https://doi.org/10.1029/2010JC006716.
Chu, P. C., C. Fan, C. J. Lozano, and J. L. Kerling, 1998: An airborne expendable bathythermograph survey of the South China Sea, May 1995. J. Geophys. Res., 103, 21 637–21 652, https://doi.org/10.1029/98JC02096.
Chu, X., G. Chen, and Y. Qi, 2020: Periodic mesoscale eddies in the South China Sea. J. Geophys. Res. Oceans, 125, e2019JC015139, https://doi.org/10.1029/2019JC015139.
Collins, C. A., T. Margolina, T. A. Rago, and L. Ivanov, 2013: Looping RAFOS floats in the California Current system. Deep-Sea Res. II, 85, 42–61, https://doi.org/10.1016/j.dsr2.2012.07.027.
de Boyer Montégut, C., G. Madec, A. S. Fischer, A. Lazar, and D. Iudicone, 2004: Mixed layer depth over the global ocean: An examination of profile data and a profile-based climatology. J. Geophys. Res., 109, C12003, https://doi.org/10.1029/2004JC002378.
Dong, C., J. C. McWilliams, Y. Liu, and D. Chen, 2014: Global heat and salt transports by eddy movement. Nat. Commun., 5, 3294, https://doi.org/10.1038/ncomms4294.
Dugan, J. P., R. P. Mied, P. C. Mignerey, and A. F. Schuetz, 1982: Compact, intrathermocline eddies in the Sargasso Sea. J. Geophys. Res., 87, 385–393, https://doi.org/10.1029/JC087iC01p00385.
Eden, B. R., D. K. Steinberg, S. A. Goldthwait, and D. J. McGillicuddy Jr., 2009: Zooplankton community structure in a cyclonic and mode-water eddy in the Sargasso Sea. Deep-Sea Res. I, 56, 1757–1776, https://doi.org/10.1016/j.dsr.2009.05.005.
Gaube, P., D. B. Chelton, P. G. Strutton, and M. J. Behrenfeld, 2013: Satellite observations of chlorophyll, phytoplankton biomass, and Ekman pumping in nonlinear mesoscale eddies. J. Geophys. Res. Oceans, 118, 6349–6370, https://doi.org/10.1002/2013JC009027.
Gaube, P., D. B. Chelton, R. M. Samelson, M. G. Schlax, and L. W. O’Neill, 2015: Satellite observations of mesoscale eddy-induced Ekman pumping. J. Phys. Oceanogr., 45, 104–132, https://doi.org/10.1175/JPO-D-14-0032.1.
Gordon, A. L., C. F. Giulivi, C. M. Lee, H. H. Furey, A. Bower, and L. Talley, 2002: Japan/East Sea intrathermocline eddies. J. Phys. Oceanogr., 32, 1960–1974, https://doi.org/10.1175/1520-0485(2002)032<1960:JESIE>2.0.CO;2.
Gordon, A. L., E. Shroyer, and V. S. N. Murty, 2017: An intrathermocline eddy and a tropical cyclone in the Bay of Bengal. Sci. Rep., 7, 46218, https://doi.org/10.1038/srep46218.
He, Q., H. Zhan, S. Cai, Y. He, G. Huang, and W. Zhan, 2018: A new assessment of mesoscale eddies in the South China Sea: Surface features, three-dimensional structures, and thermohaline transports. J. Geophys. Res. Oceans, 123, 4906–4929, https://doi.org/10.1029/2018JC014054.
Hogan, P. J., and H. E. Hurlburt, 2006: Why do intrathermocline eddies form in the Japan/East Sea? A modeling perspective. Oceanography, 19, 134–143, https://doi.org/10.5670/oceanog.2006.50.
Hu, Z., and Coauthors, 2022: A large subsurface anticyclonic eddy in the eastern equatorial Indian Ocean. J. Geophys. Res. Oceans, 127, e2021JC018130, https://doi.org/10.1029/2021JC018130.
Jungclaus, J. H., 1999: A three-dimensional simulation of the formation of anticyclonic lenses (meddies) by the instability of an intermediate depth boundary current. J. Phys. Oceanogr., 29, 1579–1598, https://doi.org/10.1175/1520-0485(1999)029<1579:ATDSOT>2.0.CO;2.
Ledwell, J. R., D. J. McGillicuddy Jr., and L. A. Anderson, 2008: Nutrient flux into an intense deep chlorophyll layer in a mode-water eddy. Deep-Sea Res. II, 55, 1139–1160, https://doi.org/10.1016/j.dsr2.2008.02.005.
Lee, D., 2018: High concentration chlorophyll a rings associated with the formation of intrathermocline eddies. Limnol. Oceanogr., 63, 2806–2814, https://doi.org/10.1002/lno.11010.
Lellouche, J.-M., and Coauthors, 2018: Recent updates to the Copernicus marine service global ocean monitoring and forecasting real-time 1/12° high-resolution system. Ocean Sci., 14, 1093–1126, https://doi.org/10.5194/os-14-1093-2018.
Leyba, I. M., M. Saraceno, and S. A. Solman, 2017: Air-sea heat fluxes associated to mesoscale eddies in the southwestern Atlantic Ocean and their dependence on different regional conditions. Climate Dyn., 49, 2491–2501, https://doi.org/10.1007/s00382-016-3460-5.
Li, C., Z. Zhang, W. Zhao, and J. Tian, 2017: A statistical study on the subthermocline submesoscale eddies in the northwestern Pacific Ocean based on Argo data. J. Geophys. Res. Oceans, 122, 3586–3598, https://doi.org/10.1002/2016JC012561.
Lin, H., J. Hu, Z. Liu, I. M. Belkin, Z. Sun, and J. Zhu, 2017: A peculiar lens-shaped structure observed in the South China Sea. Sci. Rep., 7, 478, https://doi.org/10.1038/s41598-017-00593-y.
Liu, H., W. Li, S. Chen, R. Fang, and Z. Li, 2018: Atmospheric response to mesoscale Ocean eddies over the South China Sea. Adv. Atmos. Sci., 35, 1189–1204, https://doi.org/10.1007/s00376-018-7175-x.
Lin, I.-I., C.-C. Lien, C.-R. Wu, G. T. F. Wong, C.-W. Huang, and T.-L. Chiang, 2010: Enhanced primary production in the oligotrophic South China Sea by eddy injection in spring. Geophys. Res. Lett., 37, L16602, https://doi.org/10.1029/2010GL043872.
Liu, J., and Coauthors, 2020: Effect of mesoscale eddies on diazotroph community structure and nitrogen fixation rates in the South China Sea. Reg. Stud. Mar. Sci., 35, 101106, https://doi.org/10.1016/j.rsma.2020.101106.
Liu, Y., L. Yu, and G. Chen, 2020: Characterization of sea surface temperature and air‐sea heat flux anomalies associated with mesoscale eddies in the South China Sea. J. Geophys. Res. Oceans, 125, e2019JC015470, https://doi.org/10.1029/2019JC015470.
Lukas, R., and F. Santiago-Mandujano, 2001: Extreme water mass anomaly observed in the Hawaii Ocean time-series. Geophys. Res. Lett., 28, 2931–2934, https://doi.org/10.1029/2001GL013099.
McCoy, D., D. Bianchi, and A. L. Stewart, 2020: Global observations of submesoscale coherent vortices in the ocean. Prog. Oceanogr., 189, 102452, https://doi.org/10.1016/j.pocean.2020.102452.
McDowell, S. E., and H. T. Rossby, 1978: Mediterranean water: An intense mesoscale eddy off the Bahamas. Science, 202, 1085–1087, https://doi.org/10.1126/science.202.4372.1085.
McGillicuddy, D. J., Jr., 2015: Formation of intrathermocline lenses by eddy–wind interaction. J. Phys. Oceanogr., 45, 606–612, https://doi.org/10.1175/JPO-D-14-0221.1.
McGillicuddy, D. J., Jr., 2016: Mechanisms of physical-biological-biogeochemical interaction at the oceanic mesoscale. Annu. Rev. Mar. Sci., 8, 125–159, https://doi.org/10.1146/annurev-marine-010814-015606.
McGillicuddy, D. J., Jr., R. Johnson, D. A. Siegel, A. F. Michaels, N. R. Bates, and A. H. Knap, 1999: Mesoscale variations of biogeochemical properties in the Sargasso Sea. J. Geophys. Res., 104, 13 381–13 394, https://doi.org/10.1029/1999JC900021.
McWilliams, J. C., 1985: Submesoscale, coherent vortices in the ocean. Rev. Geophys., 23, 165–182, https://doi.org/10.1029/RG023i002p00165.
Meunier, T., and Coauthors, 2018: Intrathermocline eddies embedded within an anticyclonic vortex ring. Geophys. Res. Lett., 45, 7624–7633, https://doi.org/10.1029/2018GL077527.
Nan, F., H. Xue, P. Xiu, F. Chai, M. Shi, and P. Guo, 2011: Oceanic eddy formation and propagation southwest of Taiwan. J. Geophys. Res., 116, C12045, https://doi.org/10.1029/2011JC007386.
Nan, F., F. Yu, C. Wei, Q. Ren, and C. Fan, 2017: Observations of an extra-large subsurface anticyclonic eddy in the northwestern Pacific subtropical gyre. J. Mar. Sci. Res. Dev., 7, 234, https://doi.org/10.4172/2155-9910.1000234.
Potter, H., C.-Y. Hsu, and S. F. DiMarco, 2021: Rapid dissipation of a loop current eddy due to interaction with a severe Gulf of Mexico hurricane. Ocean Dyn., 71, 911–922, https://doi.org/10.1007/s10236-021-01471-y.
Qian, S., H. Wei, J.-G. Xiao, and H. Nie, 2018: Impacts of the Kuroshio intrusion on the two eddies in the northern South China Sea in late spring 2016. Ocean Dyn., 68, 1695–1709, https://doi.org/10.1007/s10236-018-1224-y.
Qu, T., 2000: Upper-layer circulation in the South China Sea. J. Phys. Oceanogr., 30, 1450–1460, https://doi.org/10.1175/1520-0485(2000)030<1450:ULCITS>2.0.CO;2.
Spall, M. A., 1995: Frontogenesis, subduction, and cross-front exchange at upper ocean fronts. J. Geophys. Res., 100, 2543–2557, https://doi.org/10.1029/94JC02860.
Sun, Z., J. Hu, Z. Chen, J. Zhu, L. Yang, X. Chen, and X. Wu, 2021: A strong Kuroshio intrusion into the South China Sea and its accompanying cold-core anticyclonic eddy in winter 2020–2021. Remote Sens., 13, 2645, https://doi.org/10.3390/rs13142645.
Sun, Z.-B., Z. Zhang, B. Qiu, X. Zhang, C. Zhou, X. Huang, W. Zhao, and J. Tian, 2020: Three-dimensional structure and interannual variability of the Kuroshio loop current in the northeastern South China Sea. J. Phys. Oceanogr., 50, 2437–2455, https://doi.org/10.1175/JPO-D-20-0058.1.
Sun, Z.-B., Z. Zhang, B. Qiu, C. Zhou, W. Zhao, and J. Tian, 2022: Subsurface mesoscale eddies observed in the northeastern South China Sea: Dynamic features and water mass transport. J. Phys. Oceanogr., 52, 841–855, https://doi.org/10.1175/JPO-D-21-0177.1.
Takikawa, T., H. Ichikawa, K. Ichikawa, and S. Kawae, 2005: Extraordinary subsurface mesoscale eddy detected in the southeast of Okinawa in February 2002. Geophys. Res. Lett., 32, L17602, https://doi.org/10.1029/2005GL023842.
Thomas, L. N., 2008: Formation of intrathermocline eddies at ocean fronts by wind-driven destruction of potential vorticity. Dyn. Atmos. Oceans, 45, 252–273, https://doi.org/10.1016/j.dynatmoce.2008.02.002.
Wang, G., J. Su, and P. C. Chu, 2003: Mesoscale eddies in the South China Sea observed with altimeter data. Geophys. Res. Lett., 30, 2121, https://doi.org/10.1029/2003GL018532.
Wang, G., D. Chen, and J. Su, 2008: Winter eddy genesis in the eastern South China Sea due to orographic wind jets. J. Phys. Oceanogr., 38, 726–732, https://doi.org/10.1175/2007JPO3868.1.
Wang, X., Y. Du, Y. Zhang, A. Wang, and T. Wang, 2021: Influence of two eddy pairs on high-salinity water intrusion in the northern South China Sea during fall-winter 2015/2016. J. Geophys. Res. Oceans, 126, e2020JC016733, https://doi.org/10.1029/2020JC016733.
Xie, L., J. Tian, S. Zhang, Y. Zhang, and Q. Yang, 2011: An anticyclonic eddy in the intermediate layer of the Luzon Strait in autumn 2005. J. Oceanogr., 67, 37–46, https://doi.org/10.1007/s10872-011-0004-9.
Yang, G.-B., and Coauthors, 2019: Subsurface cyclonic eddies observed in the southeastern tropical Indian Ocean. J. Geophys. Res. Oceans, 124, 7247–7260, https://doi.org/10.1029/2019JC015381.
Zhang, W.-Z., H. Xue, F. Chai, and Q. Ni, 2015: Dynamical processes within an anticyclonic eddy revealed from Argo floats. Geophys. Res. Lett., 42, 2342–2350, https://doi.org/10.1002/2015GL063120.
Zhang, X., Z. Zhang, J. C. McWilliams, Z. Sun, W. Zhao, and J. Tian, 2022: Submesoscale coherent vortices observed in the northeastern South China Sea. J. Geophys. Res. Oceans, 127, e2021JC018117, https://doi.org/10.1029/2021JC018117.
Zhang, Z.-G., W. Wang, and B. Qiu, 2014: Oceanic mass transport by mesoscale eddies. Science, 345, 322–324, https://doi.org/10.1126/science.1252418.
Zhang, Z.-G., Y. Zhang, and W. Wang, 2017: Three-compartment structure of subsurface-intensified mesoscale eddies in the ocean. J. Geophys. Res. Oceans, 122, 1653–1664, https://doi.org/10.1002/2016JC012376.
Zhang, Z.-W., P. Li, L. Xu, C. Li, W. Zhao, J. Tian, and T. Qu, 2015: Subthermocline eddies observed by rapid-sampling Argo floats in the subtropical northwestern Pacific Ocean in Spring 2014. Geophys. Res. Lett., 42, 6438–6445, https://doi.org/10.1002/2015GL064601.
Zhang, Z.-W., W. Zhao, B. Qiu, and J. Tian, 2017: Anticyclonic eddy sheddings from Kuroshio loop and the accompanying cyclonic eddy in the northeastern South China Sea. J. Phys. Oceanogr., 47, 1243–1259, https://doi.org/10.1175/JPO-D-16-0185.1.
Zhang, Z.-X., F. Qiao, and J. Guo, 2014: Subsurface eddies in the southern South China Sea detected from in-situ observation in October 2011. Deep-Sea Res. I, 87, 30–34, https://doi.org/10.1016/j.dsr.2014.02.004.
Zhu, R., Z. Chen, Z. Zhang, H. Yang, and L. Wu, 2021: Subthermocline eddies in the Kuroshio extension region observed by mooring arrays. J. Phys. Oceanogr., 51, 439–455, https://doi.org/10.1175/JPO-D-20-0047.1.