You are on page 1of 24

Journal of Geophysical Research: Oceans

RESEARCH ARTICLE A New Assessment of Mesoscale Eddies in the South China Sea:
10.1029/2018JC014054
Surface Features, Three-Dimensional Structures,
Key Points:
• Three-dimensional structures of
and Thermohaline Transports
eddies in the South China Sea were Qingyou He1,2,3 , Haigang Zhan1,2 , Shuqun Cai1 , Yinghui He1 , Gaolong Huang1,2,3,
revealed by merging Argo profiles
and daily eddy data and Weikang Zhan1,2,3
• Both the rotation and westward 1
propagation of an eddy contribute to State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences,
12 4
10 W heat transport and 10 kg/s Guangzhou, China, 2Guangdong Key Lab of Ocean Remote Sensing, South China Sea Institute of Oceanology, Chinese
salt transport Academy of Sciences, Guangzhou, China, 3University of Chinese Academy of Sciences, Beijing, China
• The propagations of eddies generate
a westward water transport of 1.4 Sv,
equivalent to about 30% of the
annual-mean Luzon Strait transport
Abstract By analyzing 22 years (1993–2015) of daily eddy data, statistics of surface eddy properties
were refreshed in the South China Sea. More than 7,000 of historical Argo profiles were collocated into
eddy-centered coordinates to reveal the composite mean three-dimensional structure of eddies. The results
indicate that eddies of both polarities have long conical shape, with a maximum (minimum) density anomaly
Correspondence to:
H. Zhan,
of 0.55 kg/m3 (0.51 kg/m3) at 60 m (90 m) in the composite cyclonic (anticyclonic) eddy. Temperature
hgzhan@scsio.ac.cn and salinity anomalies also peak at eddy cores, with values of 1.5 °C and 0.15 psu in the cyclonic eddy and
1.4 °C and 0.16 psu in the anticyclonic eddy. The temperature and density anomalies extend vertically to
Citation:
400–500 m, while the salinity anomalies are apparent only in the upper 150 m. The temperature anomalies
He, Q., Zhan, H., Cai, S., He, Y., Huang, G., contribute about 90% of the density anomalies. Mixed layer depths in cyclonic eddies are on average 15 m
& Zhan, W. (2018). A new assessment of shallower than those in anticyclonic eddies. The rotation of the composite cyclonic (anticyclonic) eddy
mesoscale eddies in the South China
Sea: Surface features, three-dimensional
generates meridional heat transport of 1.4 × 1012 W (3.1 × 1012 W) and salt transport of 4.0 × 104 kg/s
structures, and thermohaline transports. (5.6 × 104 kg/s). More than 90% of the heat and salt transports are concentrated in the upper 300 and 100 m,
Journal of Geophysical Research: respectively. Compared to the meridional transports, the westward propagation of eddies results in zonal
Oceans, 123, 4906–4929. https://doi.org/
10.1029/2018JC014054
heat and salt transports on the same orders of magnitudes. The westward propagation of eddies also
generates a basin-scale westward water transport of 1.4 Sv, equivalent to about 30% of the annual-mean
Received 7 APR 2018 Luzon Strait transport.
Accepted 22 JUN 2018
Accepted article online 2 JUL 2018
Published online 20 JUL 2018
1. Introduction
Oceanic mesoscale eddies, with typical scales of O(100) km and lifespans of several months, occupy 25–30%
of the oceanic surface (Chaigneau et al., 2009) and contain the largest kinetic energy of the oceanic circula-
tion (Grachev et al., 1979). With thousands of eddies spreading across the World Ocean at any time (Chelton
et al., 2011; Faghmous et al., 2015), these features are believed to be important contributors to the global
transport of mass, heat, and salt (Dong et al., 2014; Qiu & Chen, 2005; Zhang et al., 2014). Additionally,
temperature perturbations in eddies can modulate air-sea heat flux (Ma et al., 2015; Villas Bôas et al., 2015),
wind (Byrne et al., 2016), cloud formation, and rainfall (Frenger et al., 2013). Azimuthal and vertical transports
of nutrients and planktons induced by eddies play vital roles in physical-biological-biogeochemical interac-
tions in the ocean (Chelton et al., 2011; Gaube et al., 2014; He et al., 2017; McGillicuddy, 2016). To achieve
accurate estimates of the above mentioned eddy influences, a full understanding of the three-dimensional
structure of eddies are of primary importance.
The accumulation of decades of satellite altimeter data enables long-term global observations of eddy dis-
tributions that are not feasible from classical ship-based observations. More than 10 years of Argo profile
data provide a unique contribution to access temperature and salinity signatures of eddies below the sur-
face. Although altimeter data provide only near-surface information and Argo data are spatially sparse,
the combination of them provides practical means of obtaining the observed subsurface structures of large
numbers of eddies (Chaigneau et al., 2011; Sun et al., 2017; Yang et al., 2015). By collocating six years of Argo
profile data into eddies, Chaigneau et al. (2011) derived the mean vertical structure of eddies in the eastern
South Pacific Ocean and suggested that the cores of cyclonic and anticyclonic eddies differ in depth
©2018. American Geophysical Union. (~150 m versus ~400 m). Using similar method, Yang et al. (2013) and Yang et al. (2015) obtained the mean
All Rights Reserved. vertical structure of eddies in the northwestern subtropical Pacific Ocean and the southeastern tropical

HE ET AL. 4906
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

Indian Ocean, respectively. Sun et al. (2017) investigated the composite three-dimensional structures of
eddies in the Kuroshio Extension and found that the associated anomalies in physical parameters extent
vertically to 800 m.
The South China Sea (SCS), as one of the largest semiclosed marginal sea in the western Pacific, is proved to
have complex current system (Fang et al., 2009; Qu et al., 2009; Su, 2004). Mesoscale eddies are believed to
play vital roles in the transport of heat, salt, and other biogeochemical substances in this system (Chen
et al., 2012; He et al., 2016; Zhang et al., 2014). Numerous efforts have been made to investigate the activity
of eddies using cruise observations in the past decades (Cheng & Qi, 2010; Wang et al., 2005; Zheng et al.,
2017). Recently, the employ of global long-range satellite altimeter observations further confirms the active
activities of eddies. By analyzing eight years (1993–2000) of altimeter data, Wang et al. (2003) gained an over-
view of the distribution of eddies in the SCS. They identified 28 cyclonic and 58 anticyclonic eddies, and most
of them were located in the northeastern and western SCS. Using 17 years (1992–2009) of satellite altimeter
data, Chen et al. (2011) further analyzed the mean properties of eddies in the SCS. They derived a similar spa-
tial pattern of eddy distributions, while the difference between the number of cyclonic and anticyclonic
eddies is much smaller. Using numerical model results and satellite data, Xiu et al. (2010) censused the num-
bers, sizes, lifetimes, and tracks of eddies in the SCS between 1993 and 2007. They identified 32.9 ± 2.4 eddies
per year, and about 52% of them are cyclonic. The different results among these studies are mainly caused by
the differences in eddy identification algorithm and criteria. The limitation in temporal resolution (weekly) of
the altimeter data also leads to uncertainties in eddy identification and tracking.
Efforts had also been made to uncover the subsurface characters of eddies in the SCS. For instance, by com-
bining in situ measurements and satellite observations, Nan et al. (2011) investigated three long-lived antic-
yclonic eddies near the 18°N section in August 2007. These eddies have temperature anomalies of ∼0.65 °C
near eddy cores and extend vertically to ~900 m. Soon afterward, Hu et al. (2011) reported a cyclonic eddy,
with maximum temperature decrease of 8 °C at 50 m and vertical extension of 250 m, in the southwestern
SCS in September 2007. Zhang et al. (2016) deployed a mooring system for eddy observations at the west
of Luzon Strait and found an anticyclonic and cyclonic eddy pair that extended deep to 2,000 m. The vertical
structure of these eddies varies case to case, indicating that case studies that based on rare cruise observa-
tions are not adequate to reveal the general three-dimensional structure of eddies in the whole SCS. By col-
locating three years (2006–2009) of historical Argo profile data into weekly sampled eddies, Chen et al. (2011)
obtained climatological mean temperature anomaly profiles, as functions of depth, in eddies in the SCS. Chen
et al. (2012) further estimated the basin-scale eddy heat and salt transports in the form of eddy diffusion.
However, they did not derive the three-dimensional structure of eddies because of the inadequate in Argo
data (only 63 Argo floats with 375 profiles were associated with eddies in Chen et al., 2012).
By combining more than 7,000 of Argo profiles with daily eddy data, the objective of this work is to reveal the
general three-dimensional structure of eddies in the SCS, and the associated eddy transports. The new daily
data not only provide a refined eddy identification and tracking but also enable an accurate collocating of
transient Argo profiles into eddies. More importantly, the large number of accumulated Argo profiles
enhances the possibility for constructing the mean three-dimensional structure of eddies. The combination
of these two sets of data enables a systematical eddy census including mean surface properties, three-
dimensional structures, and eddy-induced transports. The remainder of the paper is organized as follows.
Section 2 describes the data and computational methods. Since the analysis is based on the new daily eddy
data, we proceed a refreshed statistic of surface eddy properties in section 3. Section 4 focuses on the com-
posite average three-dimensional structure of eddies, including temperature, salinity, and density anomalies,
as well as mixed layer depth (MLD) anomalies. Heat and salt transports that induced by the rotation and
westward propagation of eddies are subsequently estimated based on the derived three-dimensional eddy
structures. In section 5, we compared the results with previous studies and discussed the seasonal and
regional variations of the eddies’ vertical structures, and a simple summary is formulated in section 6.

2. Data and Methods


The eddy data used in this work are the Version 4 eddy data set developed from the early eddy product by
Chelton, Schlax, and Samelson (2011). This series of data has been widely used in the investigation of oceanic
eddies globally (Gaube et al., 2015; Hausmann et al., 2017; He et al., 2016; Pilo et al., 2015; Samelson et al.,

HE ET AL. 4907
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

Figure 1. Geographic distributions of historical Argo profile data in the (a) SCS, (b) composite cyclonic eddy, and
(c) composite anticyclonic eddy. N is the number of profiles. The x and y coordinates in (b) and (c) are normalized by
eddy radius R. Noted that profiles out of 4R are not plotted in (b) and (c).

2014). The new version is identified and tracked based on the DT-2014 daily “two-sat merged” sea level
anomaly fields from the Archiving, Validation, and Interpretation of Satellite Oceanographic data (AVISO;
Pujol et al., 2016). Eddies in the new data are defined on the basis of connected pixels that satisfy
specified criteria (Williams et al., 2011), rather than the closed contours of sea surface height (Chelton,
Schlax, & Samelson, 2011). This version of eddy data is produced by the Collecte Localisation Satellites/
Developing Use of Altimetry for Climate Studies (CLS/DUACS) team and is now available from AVISO. The
data are daily sampled from January 1993 to April 2015. It includes the trajectories of eddies with
additional information such as eddy amplitude, radius, and polarity (cyclonic/anticyclonic). This set of data
is considered to be more accurate than those produced by the Okubo-Weiss method for it reduces the
number of “false” detected eddies in a large extent (Chaigneau et al., 2008, 2011). Moreover, the new daily
product avoids many awful eddy identification and tracking compared to previous eddy data at seven-day
intervals. For more detailed information on the eddy data, please refer to Chelton, Schlax, and Samelson
(2011) and Williams et al. (2011).
Argo profile data that match the eddy signals are used to construct the composite three-dimensional struc-
ture of eddies. This data set is obtained from the China Argo Real-time Data Center, which collects all the Argo
data from the Global Ocean Data Assimilation Experiment and the French Research Institute for Exploitation
of the Sea. The global Argo data are available from July 1997 to present. About 7,597 Argo profiles were col-
lected in the SCS between July 1997 and April 2015 (Figure 1a). The first Argo profile in the SCS was obtained
in April 2006. The number of Argo profiles increased noticeably after 2009, and more than half of the profiles
were obtained after 2011. Each data profile had undergone automatic preprocessing and quality control pro-
cedures by the data collectors. Only profiles with pressure, temperature, and salinity data flagged as good
were used in this analysis. In addition, we proceeded a more rigorous quality control to avoid some suspicious
observations. For example, profiles with first record below 20 m, last record shallower than 100 m, or unique

HE ET AL. 4908
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

records less than 10 were eliminated. Those with records exceeding 3 times of standard deviation were also
removed. Noted that the quality control was set somewhat less rigorous than that in previous works to retain
more profiles in the marginal sea (Chaigneau et al., 2011; Dong et al., 2017). This would not greatly affect the
statistical results.
The retained temperature and salinity profiles were linearly interpolated onto 5 m equally spaced grids from
5 to 1,200 m. Derived quantities, such as density and the MLD, were estimated from these vertically gridded
temperature and salinity profiles. The MLD was defined here as the depth where the density increase com-
pared to density at surface corresponds to a temperature decrease of 0.5 °C (Kara et al., 2000; Monterey &
Levitus, 1997). To better describe perturbations in eddies, anomalies of all these properties were computed
by removing respective climatological mean profiles of all Argo data (Yang et al., 2015). Then, for each profile,
we isolated all synchronous eddies located in the SCS and calculated the distances between the profile and
the centers of the eddies. We associated the profile with the nearest eddy and normalized the distance
between them by the eddy’s radius (R). Then, all the profiles were classified into two categories according
to the polarities (cyclonic/anticyclonic) of the associated eddies. For each category, all associated profiles
were composited onto an eddy-centered coordinate with normalized eddy radius R (Figures 1b and 1c).
The location of each profile in the composite eddy is determined by the relative position between the profile
and the associated eddy. A total number of 970 and 924 Argo profiles are located within 1R of the composite
cyclonic and anticyclonic eddies, respectively. They account for 27% and 23% of the total number of profiles
that are associated with cyclonic and anticyclonic eddies, respectively. These proportions support that eddies
occupy ∼25% of the oceanic surface in the SCS. The profiles within 1R of cyclonic and anticyclonic eddies
have similar uniform spatial distributions in the SCS. Additionally, distributions of the profiles in the compo-
site eddies are almost uniform, especially within twice the eddy radius (Figures 1b and 1c). These results pro-
vide supports to the feasibility of composite analysis. As a consequence, the composite three-dimensional
structures of eddies were derived by interpolating the composited profiles onto normalized high-resolution
grids at each vertical level. Noted that the composite procedure here is a bit different from previous works
(Chaigneau et al., 2011; Sun et al., 2017; Yang et al., 2015), which search profiles in/near each eddy. This is
aimed to avoid the situation that a profile is associated to two or more eddies, that is, located inside one eddy
and at the periphery (i.e., 1-2R) of another.
To achieve an estimation on heat and salt transports that induced by the rotation of eddies, we assumed that
eddies are geostrophic and calculated the three-dimensional geostrophic velocities of the composite eddies.
The velocities are derived from geostrophic potential anomalies, which are a function of temperature, salinity,
and pressure, based on the equation of state for seawater (Millero et al., 1980). The locations of the composite
cyclonic and anticyclonic eddies are assumed to be the mean positions of all cyclonic and anticyclonic eddies,
respectively. The reference depth of no motion is assumed to be 1,200 m (Chen et al., 2012; Yang et al., 2015).
We considered that the chosen of this reference depth should not strongly impact our results since the mean
flow field below 1,000 m depth is negligible small. Additionally, we compared the differences between geos-
trophic velocities that were calculated from different reference depths and found that the velocity fields stay
almost the same when the reference depth is deeper than 1,000 m (figures not shown).

With the composite three-dimensional structures of temperature anomalies (T0 ), salinity anomalies (S0 ), and
swirl velocities (V0 ), heat (Hf) and salt (Sf) fluxes caused by the rotation of eddies can be calculated by

Hf ¼ ρC p V 0 T 0 ; (1)
0 0
Sf ¼ 0:001ρV S ; (2)

where ρ is the density of water and Cp = 4,187 J · kg1 · ° C1 is the specific heat capacity. As a consequence,
meridional heat and salt transports induced by the composite eddies can be estimated by integrated the
meridional component of heat and salt fluxes over depth, respectively.

The above estimation of eddy transports are primarily meridional because large-scale property gradients are
usually meridional. Since eddies often contain cold/warm and salty/fresh water, the propagation of eddies
will also lead to horizontal eddy transports (Early et al., 2011). These transports are primarily zonal since
the propagation of eddies is usually westward. To achieve an estimation on this, defining the depth, to which
an eddy can trap and transport the water within it, has been a challenge (Roemmich & Gilson, 2001; Zhang

HE ET AL. 4909
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

et al., 2017; Zhang, Wang, & Qiu, 2014). Recently, the diagnostic of eddy nonlinearity, as a ratio of swirl over
propagation speed (Chelton, Schlax, & Samelson, 2011), was suggested a reasonable indicator for trapping
efficiency in the upper ocean (Chaigneau et al., 2011; Frenger et al., 2015; Yang et al., 2013). They suggested
that an eddy maintains a coherent structure and traps water as it propagates when its nonlinearity exceeds 1.
Since most of eddies in the SCS are nonlinear (Liu et al., 2013), the volume of water trapped by the composite
eddy (Vol) therefore could be estimated by integrating eddy’s area from the surface to the “trapping depth”
(h0) where maximum swirl speed at eddy’s edge equals to eddy propagation speed (Chaigneau et al., 2011;
Flierl, 1981). As a result, the heat (Qt) and salt (Qs) content anomalies trapped by an eddy can be estimated by

∫ ∫ρC T dA dz;
0
0
Qt ¼ h0 p (3)

∫ ∫0:001ρS dA dz;
0
0
Qs ¼ h0 (4)

where A is the area of the composite eddy and z is depth. The heat and salt transports can be derived by
dividing the corresponding contents by the time (t) the eddy takes to cross a reference section (t = 2R/c,
where c is the westward propagation speed of eddy and R is eddy radius).

3. Eddy Features at the Surface


3.1. Occurrence, Genesis, and Dissipation
Since this study used the new daily eddy data, it is necessary to refresh the statistics of surface eddy proper-
ties at first. A total number of 37,460 cyclonic and 36,812 anticyclonic eddy snapshots, corresponding to 625
and 534 eddies with lifespans longer than 28 days, respectively, were detected in the SCS between January
1993 and April 2015. This means that there are on average 4.6 cyclonic and 4.5 anticyclonic eddies in the SCS
at any given time. The ratio between the number of these two types of eddies is in consistent with the results
of Xiu et al. (2010) but disagrees with those of Wang et al. (2003) and Chen et al. (2011), which suggest that
the number of anticyclonic eddies is more than that of cyclonic eddies. The spatial distribution of the eddies,
at each 0.5° × 0.5° latitude-longitude grid, demonstrates that approximately 40% of eddies are located in the
northern SCS (north of 17°N; Figure 2a). In this region, eddies are most frequently observed to the southwest
of Taiwan, southwest of Luzon Island, and along the northwestern boundary of the SCS. The western SCS
is another region with high eddy occurrence. Approximately 26% of eddies are observed west of 113°E
(7–17°N). With 11% of eddies locate east of 116°E (12–15°N), the eastern SCS is also characterized by active
eddy activities. In contrast, the center and southeastern SCS are of sparse eddy activities. The spatial distribu-
tions of cyclonic and anticyclonic eddies are almost the same (Figures 2b and 2c), differing from the results of
Chen et al. (2011), in which most of the regions tend to be anticyclonic.
The pattern of eddy genesis (locations where eddies are first detected) differs from that of the eddy dis-
tributions (Figure 2d). Approximately 34% of eddies initiate near the eastern boundary (>118°E). In this
region, two peaks are found: northwest of the Luzon Island and west of the Mindoro strait. This is in
agreement with the eddy generation mechanisms that orographic wind jets and instability of nonlinear
Rossby waves radiated from the eastern coast are in favor of the formation of eddies (Cheng et al.,
2016, 2017; He et al., 2016; Wang et al., 2008). The contrasting rare eddy genesis along the southeastern
boundary provides support to this scenario, on the opposite aspect. The western SCS is also characterized
with relatively high eddy genesis, though much lower than the eastern SCS. The distributions of eddy
genesis for cyclonic and anticyclonic eddies are generally similar (Figures 2e and 2f), except in the west
of Luzon strait, where more anticyclonic eddies are observed shedding from the Kuroshio (Nan et al.,
2011; Zhang et al., 2017).
Eddies mainly dissipate in the western SCS, especially along the northwestern and western boundaries
(Figure 2g). Approximately 40% of the eddies dissipate west of 113°E, supporting that the western boundary
acts as a “graveyard” for the westward propagating eddies (Aiki et al., 2016; Zhai et al., 2010). As suggested by
Dewar and Hogg (2010), eddy energy that released at the western boundary is likely scattered into high-
wave-number vertical modes, resulting in locally enhanced diapycnal mixing, which can have important
implications for the abyssal overturning circulation (Aiki et al., 2016; Saenko et al., 2012). The dissipation of

HE ET AL. 4910
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

Figure 2. Geographic distributions of (top) eddy occurrence, (middle) genesis, and (bottom) dissipation at each 0.5° × 0.5° pixel in the SCS between January 1993
and April 2015. The eddy occurrence denotes all the detected eddy snapshots, eddy genesis is the first time an eddy is identified, and eddy dissipation is the
last time an eddy is detected. The left, middle, and right columns are for all eddies, cyclonic eddies, and anticyclonic eddies, respectively. The black contour in each
panel is 200 m isobath.

cyclonic eddies differs somewhat from that of anticyclonic eddies in the northern SCS (Figures 2h and 2i).
Cyclonic eddies dissipate all the way along the northwestern boundary, while few anticyclonic eddies
dissipate until west of 113°E. One important reason might be that the cyclonic eddies are much weaker
than the anticyclonic eddies (He, Zhan, Cai, & Li, 2016). As a result, cyclonic eddies are easier to dissipate
during their interactions with the topography or the northern SCS circulation. This is supported by the
model simulation in Jan et al. (2017), which suggests that weaker eddies are easier to lose energy during
their interactions with the western boundary current. The β-effect provides an alternative explanation for
the difference in eddy dissipation. Previous studies have shown that westward propagating cyclonic

HE ET AL. 4911
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

Figure 3. Histogram distributions of eddy lifespans, amplitudes, radius scales, and propagation speeds (left to right). The top, middle, and bottom rows are
cumulative histograms, histograms, and the ratios of cyclonic to anticyclonic eddies, respectively. The red lines are for cyclonic eddies and blue lines are for
anticyclonic eddies.

(anticyclonic) eddies prefer to deflect poleward (equatorward) under the combined effects of the planetary
vorticity gradient and self-advection (Chelton, Schlax, & Samelson, 2011; Morrow, 2004). As such, cyclonic
eddies are more likely to hit against the northwestern topography, and thus dissipate along the boundary.
It is worthy to note that the central SCS is also a region of high eddy dissipation, and this is true for both
cyclonic and anticyclonic eddies.
3.2. Amplitude, Radius, and Propagation Speed
Cumulative histograms and histograms of eddy lifespans are shown separately for cyclonic and anticyclonic
eddies in Figures 3a and 3e. Cyclonic eddies have an average lifespan of 62 days, and anticyclonic eddies
have an average lifespan of 70 days. Eddies with lifespans shorter than two months account for 64% and
59% for cyclonic and anticyclonic eddies, respectively. And those with lifespans longer than four months
account for 9% and 15%, respectively. A significant characteristic of the histograms is that the lifespans of
anticyclonic eddies are more skewed toward large values than those of cyclonic eddies (Figure 3e). As a con-
sequence, eddies with lifespans longer than 120 days are preferentially anticyclonic and those with lifespans
shorter than 50 days are preferentially cyclonic (Figure 3i). This is consistent with the finding that anticyclonic
eddies tend to propagate farther westward than cyclonic eddies do before dissipation (Figures 2h and 2i).
Histograms and cumulative histograms of eddy amplitudes show that 74% of cyclonic eddies and 68% of
anticyclonic eddies have amplitudes smaller than 10 cm (Figures 3b and 3f). In contrast, eddies with ampli-
tudes larger than 20 cm account for only 3% and 4% of cyclonic and anticyclonic eddies, respectively. The
average amplitudes for these two types of eddies are 7.7 and 8.5 cm, respectively. Similar to the histograms
of eddy lifespans, amplitudes of anticyclonic eddies are more skewed toward large values than those of cyclo-
nic eddies (Figure 3j). The spatial distribution of eddy amplitudes demonstrates that large-amplitude eddies
occur preferentially in the northeastern and western SCS (Figure 4a), where the regions are of energetic

HE ET AL. 4912
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

Figure 4. Geographic distributions of eddy amplitudes, radius scales, and propagation speeds (top to bottom). The left, middle, and right columns are for all eddies,
cyclonic eddies, and anticyclonic eddies, respectively. The black contour in each panel is 200 m isobath.

currents (Qu et al., 2009; Su, 2004). Amplitudes of anticyclonic eddies are larger than those of cyclonic eddies
in most regions of the SCS (Figures 4b and 4c). The difference is most apparent in the northern SCS, where the
Kuroshio often intrudes westward into the SCS in forms of intense anticyclonic eddies (Nan et al., 2015).
The histogram distributions of eddy radii are similar to those of eddy amplitudes (Figures 3c and 3g). There
is a preference for eddies with radii smaller than 90 km to be cyclonic and larger than 100 km to be anticy-
clonic (Figure 3k). The average radii are 104 and 111 km for cyclonic and anticyclonic eddies, respectively
(Figure 3g). The spatial distribution of eddy radii demonstrates that eddies in the southern SCS generally have
larger radii than those in the northern SCS (Figure 4d). This is true for both cyclonic and anticyclonic eddies
(Figures 4e and 4f). The decrease of eddy radius, with the increase of latitude, is consistent in trend with the

HE ET AL. 4913
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

Figure 5. Seasonal variation of eddy propagation speeds. The upper row is for winter (December to the next May), and the lower row is for summer (June
to November). The left, middle, and right columns are for all eddies, cyclonic eddies, and anticyclonic eddies, respectively. The black contour in each panel is
200 m isobath.

decrease of Rossby radius of deformation (Chen et al., 2011). Meanwhile, eddies near the eastern and western
boundaries, locations of eddy genesis and dissipation (Figures 2d and 2g), have radii smaller than those in the
center SCS.
The daily eddy data support a refined analysis of eddy propagation. Propagation speeds of eddies are com-
puted by dividing the distance between the locations of eddy at t + 1 and t  1 by the sampling time. With a
mean value of 8 cm/s, the histograms of propagation speeds of cyclonic and anticyclonic eddies are almost
the same (Figures 3d and 3h). The spatial distribution of eddy propagation speeds demonstrates that eddies
west of Luzon strait propagate westward or northwestward at an average speed of ~5 cm/s (Figure 4g). The
speed increases dramatically to ~10 cm/s when eddies encounter with the topography. Additionally, the pro-
pagate direction deflects to southwestward along the northwestern boundary. Eddies in the eastern SCS pro-
pagate generally westward at an average speed of ~5 cm/s. Whereas, eddies in the western SCS generally
have small propagation speed (less than 3 cm/s) because of the impediment of the western boundary. It is
interesting to note that eddies west of the Luzon Island also have propagation speeds smaller than 3 cm/s.
This is a region where Luzon Cold Eddy often forms (Qu, 2000; Yang & Liu, 2003), and this eddy rarely migrates
in winter (He, Xie, & Cai, 2016). The propagations of cyclonic and anticyclonic eddies are similar in most
regions except in the western SCS, where the westward propagation of cyclonic eddies deflects southward
while that of anticyclonic eddies deflects northward (Figures 4h and 4i).
To gain a better understanding of the accelerated eddy propagation speed along the northwestern bound-
ary, we compared the propagation speeds of eddies between summer (June to November) and winter
(December to the next May). As shown in Figure 5a, eddy propagation speeds are significantly increased
along the northwestern boundary in winter. In contrast, no apparent speed increase is found in this region

HE ET AL. 4914
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

in summer (Figure 5d). This is true for both cyclonic and anticyclonic
eddies (Figures 5b, 5c, 5e, and 5f). One important reason might be that
the SCS is characterized with cyclonic circulation in winter (Quan et al.,
2016). The intense southwestern boundary current may facilitate the
southwest propagation of eddies in a large extend. In contrast, the bound-
ary current weakens in summer, thus has little influence on the propaga-
tion speed of eddies. The interaction between eddy and topography (the
topographic β effect) might also impact the propagation of eddies.
The lifespans of eddies in the SCS range from 30 to 300 days (Figure 3e).
This means that the intensification stage of a long-lived eddy can be the
dissipation stage of another short-lived eddy. The fact makes the statistic
of eddies’ time evolution difficult to achieve. The normalization of eddies’
properties, that is, amplitudes, radii, and propagation speeds, by eddies’
lifespans provides a possible access. As shown in Figure 6a, both cyclonic
and anticyclonic eddies form with small eddy amplitude (~3 cm). The
amplitudes of eddies of both polarities increase and decrease rapidly in
the first and last 30% of eddies’ lifetime, respectively. The trend is similar
to the results in the open oceans (Samelson et al., 2014), except for a much
shorter maintaining stage. This is consistent with the short eddy lifespans
in the SCS compared to those in the open oceans. Considering that the
cyclonic and anticyclonic eddies have average lifespans of 62 and 70 days,
the intensification stage for them are on average 19 and 21 days, respec-
tively. The evolution of eddy radius is similar to that of eddy amplitude
(Figure 6b). A significant characteristic is that the radius of anticyclonic
eddies is 5–10 km lager than that of cyclonic eddies throughout eddies’ life
cycle. The propagation speeds of eddies of both polarities do not show sig-
nificant correlation with eddies’ life cycles (Figure 6c).

4. Three-Dimensional Structures and


Associated Transports
4.1. Three-Dimensional Structures
As shown in Figure 7a, profiles within cyclonic eddies (defined hereafter as
the distance between Argo profile and eddy center r < 1R) generally
recorded lower temperature (on average  1.1 °C at 100 m) than those
Figure 6. Normalized life cycles of (top) eddy amplitude, (middle) radius, and
(bottom) propagation speed for cyclonic (blue) and anticyclonic (red) eddies. outside eddies (defined hereafter as the distance between Argo profile
The eddy lifetime is normalized to 1.pThe
ffiffiffiffiffiffiffiffierror
ffi bars denote the standard and eddy center r > 1R). In contrast, profiles within anticyclonic eddies
error of the mean defined as ±σðtÞ= NðtÞ, where σ(t) is the standard generally recorded higher temperature (on average 1.3 °C at 100 m) than
deviation and N (t) is the number of eddies at dimensionless time t. the ambient (Figure 7b). These results are consistent with the hypothesis
that divergence (convergence) in cyclonic (anticyclonic) eddies generates
upwelling (downwelling) of cold (warm) water in eddy centers (Chaigneau
et al., 2011; Sun et al., 2017; Yang et al., 2015). The upwelling (downwelling) in cyclonic (anticyclonic) eddies
also results in uplift (depress) of salty (fresh) water. As a results, profiles within cyclonic eddies generally
recorded high salinities, and those within anticyclonic eddies generally recorded low salinities (Figures 7c
and 7d).
Though 57% of profiles have negative values, the histogram of temperature anomalies in cyclonic eddies
shows two peaks at 30 m, suggesting that positive and negative temperature anomalies coexist in eddies
(Figure 8, left panel). This is in agreement with the process of eddy stirring, which produces opposite
cross-frontal transport in two opposite sides of an eddy (Chelton, Gaube, et al., 2011; He, Zhan, Cai, & Zha,
2016), resulting in positive temperature anomalies in one side and negative temperature anomalies in the
other side. The positive peak of temperature anomaly decreases with depth and disappears at 75 m. The
number of profiles with negative temperature anomalies accounts for 76% at this depth. The ratio rarely
increases with depth hereafter, and the shape of the histogram adjusts to normal distribution at 150 m

HE ET AL. 4915
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

Figure 7. Scatters of (top) temperatures and (bottom) salinities in composite (left) cyclonic and (right) anticyclonic eddies
at 100 m. The x and y coordinates are normalized by eddy radius as in Figure 1.

and deeper. These results suggest that the influence of eddy stirring weakens with depth. The primary reason
may be that the horizontal gradient of large-scale temperature decreases with depth and becomes relatively
small beneath the mixed layer. Additionally, the rotational speed of eddies also decreases with depth (Chen
et al., 2012; Zhang et al., 2016). As a result, temperature anomalies caused by the rotation of eddies are
secondary to those caused by eddy upwelling.
For anticyclonic eddies, the number of profiles with positive temperature anomalies accounts for 61% at 30 m
(Figure 8, second column). Whereas, the negative peak of the histogram is markedly smaller than the positive
peak, suggesting that the influence of eddy stirring is not evident. The negative peak weakens with
depth and disappears at 75 m. As a result, 77% of the profiles have positive temperature anomalies at this
depth. Then, the ratio remains stable and the shape of the histogram adjusts to normal distribution
when depth increases. The trend is similar to that of cyclonic eddies.
The situation for salinity anomalies is similar to that of temperature anomalies (Figure 8, right panel). More
than half of the profiles (61% and 60%) have positive and negative salinity anomalies in cyclonic and antic-
yclonic eddies at 30 m, respectively. Two peaks lie in the histograms of salinity in eddies of both polarities.
Additionally, the negative peak in cyclonic eddies and the positive peak in anticyclonic eddies diminish with
depth and disappear at 75 m. Consequently, the ratio of profiles that have positive and negative salinity
anomalies in cyclonic and anticyclonic eddies increase to 73% and 72%, respectively. These results support
the hypothesis of eddy stirring and provide additional evidence that the influence of eddy stirring weakens
with depth.
The near normal distributions of the histograms of temperature and salinity anomalies provide support to the
feasibility of composite analysis. By interpolating the eddy-centered temperature and salinity anomalies at
each depth (i.e., Figure 7) onto standard high-resolution grids, we constructed the composite mean three-
dimensional structures of temperature and salinity anomalies in eddies (Figure 9). As shown in Figure 9a,
the composite cyclonic eddy has a long conical shape, with negative temperature anomalies in eddy’s inter-
ior. The peak of the temperature anomalies (1.5 °C) locates at 70 m at eddy’s center. The temperature

HE ET AL. 4916
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

Figure 8. Histograms of temperature anomalies (left patterns) and salinity anomalies (right patterns) of Argo profiles in eddies (1R) at different depths (30, 75, 150,
and 300 m from top to bottom). Note that the y labels differ from each other.

anomalies increase upward to 0.4 °C at the surface and downward to 0.2 °C at about 500 m (Figure 9g).
Large negative and positive temperature anomalies lie in the southwestern and northeastern parts of the
eddy near the surface, respectively (Figure 9a, sea also Figure 10a). This dipole-like feature provides
evidence to the process of eddy stirring (Chelton, Gaube, et al., 2011), which transports cold water from
north to south and warm water from south to north in the west and east sides of eddies, respectively. The
dipole weakens with depth and disappears at about 50 m, which is generally the depth of mixed layer (He,
Zhan, Cai, & Li, 2016), indicating that the effect of eddy stirring is only apparent in the upper layer.
Additionally, the dipole is asymmetric, with lager anomalies in the western side (Figure 10a), implying net
meridional heat transport by the rotation of eddies.
The shape of temperature anomalies is similar for the composite anticyclonic eddy, except for somewhat flat
(Figure 9d). The composite eddy shows a maximum temperature anomaly of 1.4 °C centered at 90 m. The
temperature anomaly decreases downward to about 0.1 °C at 500 m (Figure 9g). However, the composite fea-
ture does not show significant dipole-like temperature anomalies near the surface. This suggests that eddy
stirring may not have apparent influence on meridional temperature flux in anticyclonic eddies. Instead, a
layer of relatively cold water overlays the core of the composite eddy. Previous studies in oceans at similar
latitudes (the eastern South Pacific Ocean) derived similar composite temperature structures in anticyclonic
eddies (Chaigneau et al., 2011; Colas et al., 2011). They suggest that this is probably due to the doming of iso-
therms above anticyclonic eddy cores.
The structures of salinity anomalies in the composite eddies are less regular (Figures 9b and 9e). The
composite cyclonic (anticyclonic) eddy shows large salinity anomalies of about 0.15 psu (0.16 psu) at the
subsurface (Figure 9h). The positive and negative salinity anomalies are consistent with the upwelling and

HE ET AL. 4917
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

Figure 9. Average three-dimensional (left) temperature, (middle) salinity, and (right) density anomalies of the composite (top) cyclonic and (middle) anticyclonic
3
eddies. The black contours are ±0.4°C, ±0.05 psu, and ±0.1 kg/m for temperature, salinity, and density anomalies, respectively. The bottom rows are the mean
vertical profiles of temperature (left), salinity (middle), and density (right) anomalies inside the composite cyclonic (blue) and anticyclonic (red) eddies (<1R). The
shadows are standard errors as defined in Figure 6. The x and y coordinates are normalized by eddy radius as in Figure 1.

downwelling of salty and fresh water in cyclonic and anticyclonic eddies, respectively. The centers of the
salinity anomalies are similar in depths to those of the temperature anomalies. Whereas, the vertical
extends of the salinity anomalies are apparent only in the upper 150 m, which is much shallower than the
vertical extends of the temperature anomalies (Figure 10). This is primarily attributed to the presence of a

HE ET AL. 4918
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

Figure 10. Vertical sections of (left) temperature, (middle) salinity, and (right) density anomalies of the composite (top) cyclonic and (bottom) anticyclonic eddies at
y = 0 in Figure 9.

salinity barrier layer in the SCS (Nan et al., 2016; Zeng & Wang, 2016). As stated by Nan et al. (2016), large-scale
salinity increases apparently from the surface to ~150 m in the SCS. Then, the salinity decreases gradually
with depth before it increases with depth again at 500 m. The vertical variation of the salinity below 150 m
is much weaker than that above 150 m. As a result, salinity anomalies in eddies are apparent only in the
upper 150 m. In the mixed layer (<50 m), apparent positive (negative) salinity anomalies are observed in
the west (east) side of the composite cyclonic eddy (Figure 10b), demonstrating the importance of eddy
stirring in meridional salt flux.
To obtain a direct insight into the physical structure of eddies, we computed three-dimensional density
anomalies within the composite eddies via temperature and salinity profiles. The results show that eddies
of both polarities have long conical shape (Figures 9c and 9f). The structure is similar to that of the tempera-
ture anomalies rather than the salinity anomalies (Figure 9), suggesting that the vertical structures of eddies
are more dominated by temperature anomalies. Based on the linearized state equation of seawater, it is esti-
mated that the temperature anomalies contribute about 90% of the density anomalies in eddies of both pola-
rities. The composite cyclonic eddy has elevated density in eddy’s interior. The eddy’s core locates at 60 m
with maximum density anomaly of 0.55 kg/m3. The density anomaly decreases downward to 0.03 kg/m3 at
about 500 m (Figure 9i). For the composite anticyclonic eddy, depressed density is found in eddy’s interior.
The eddy’s core locates at 90 m with minimum density anomaly of 0.51 kg/m3. Although this core locates
deeper than that of the cyclonic eddy, the vertical extend of the density anomalies is ~100 m shallower than
that of the later (Figures 10c and 10f).

HE ET AL. 4919
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

Figure 11. (a) Cumulative histogram and (b) histogram distributions of MLDs in cyclonic eddies (blue), anticyclonic eddies
(red), and outside eddies (black). The corresponding dashed blue, red, and black lines in (b) are for the mean values. (c and
d) The composite average MLDs in cyclonic and anticyclonic eddies, respectively.

The temperature and salinity anomalies in eddies may also result in thermocline movements. As presented in
Figure 11b, MLDs in cyclonic eddies (<1R) are more skewed toward small values while those in anticyclonic
eddies are more skewed toward large values. Approximately 63% of the cyclonic eddies are characterized
with MLD less than 40 m (Figure 11a). In contrast, the percentage is nearly halved (38%) for anticyclonic
eddies. Meanwhile, eddies with MLD larger than 80 m account for 13% of anticyclonic eddies and only 2%
of cyclonic eddies. These results support that upwelling in cyclonic eddies uplifts the MLD and downwelling
in anticyclonic eddies depresses the MLD. Specifically, the average MLD in cyclonic eddies is ~14% shallower
than that of the backgrounds (37 m versus 43 m). And that in anticyclonic eddies is 21% deeper than that of
the backgrounds (52 m versus 43 m). Since the seasonal variation of the MLD (ranges from ~30 m in summer
to ~65 m in winter) is generally larger than the regional variation in the SCS, we removed the climatological
monthly mean MLD from the MLD observations to construct the spatial patterns of composite MLD anoma-
lies in eddies (Figures 11a and 11d). The patterns present negative and positive MLD anomalies in cyclonic
and anticyclonic eddies, respectively, and the magnitude of the prior is smaller than that of the later. One
important reason might be that the amplitudes of anticyclonic eddies are generally larger than those of
cyclonic eddies (Figures 4b and 4c). As stated by Hausmann et al. (2017), eddies with larger amplitudes are
more likely to have deeper MLDs.

4.2. Eddy-Induced Transports


It has been mentioned above that the asymmetric of temperature and salinity anomalies in eddies might
result in net meridional heat and salt transports. To achieve an estimation on this, we calculated the three-
dimensional geostrophic velocities of the composite eddies (section 2). As shown in Figures 12a and 12b,
the local minimum and maximum geostrophic potential anomalies coincide with the centers of the compo-
site cyclonic and anticyclonic eddies, respectively. Correspondingly, swirl velocities in these eddy centers are
close to zero. The maximum swirl velocity is of ∼10 cm/s near the edges of the composite eddies of both pola-
rities (at 10 m depth). The spatial distributions of the velocities are consistent with the anticlockwise and
clockwise rotation of cyclonic and anticyclonic eddies, respectively. The zonal sections of meridional veloci-
ties crossing the eddies’ centers demonstrate that both cyclonic and anticyclonic eddies have the largest
swirl velocity near the surface (Figures 12c and 12d). The velocities in eddies of both properties decrease with
depth and become less than 0.01 m/s below 500 m. This provides support to that the flow in deep water is
rather weak and the definition of reference depth as 1,200 m is reasonable. It is worthy to note that the

HE ET AL. 4920
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

Figure 12. (a and b) Geopotential anomalies (colors) and geostrophic velocities (vectors) of the composite eddies near the
surface (10 m). The reference depth is 1,200 m. (c and d) Vertical sections of meridional geostrophic speeds of the com-
posite eddies at y = 0 in (a) and (b). The left and right panels are for cyclonic and anticyclonic eddies, respectively.

sections of the meridional velocities are asymmetric for eddies of both polarities. This asymmetric in
velocities might also contribute to net meridional heat and salt transports (Chen et al., 2012; Yang et al., 2015).
As shown in Figures 13a and 13b, both cyclonic and anticyclonic eddies generate positive (northward) and
negative (southward) heat transports in the west and east parts, respectively (Figures 13a and 13b). The heat
transports are mainly concentrated in the upper 300 m, with peaks of ±2 × 105 W/m2 at the depth (of 70–
90 m) where temperature anomalies are the most apparent (Figures 10a and 10d). Because of the asymmetry
in temperature anomalies and velocities (Figures 10 and 12), the heat transports also show asymmetric pat-
terns. As a result, the integrated transports across the zonal section is positive (northward) for cyclonic eddies
and negative (southward) for anticyclonic eddies (Figure 13e). The integrated heat transports are on the order
of 1010 W/m and are strongly depth dependent. For both cyclonic and anticyclonic eddies, more than 90% of
the heat transports are concentrated in the upper 300 m. The depth-integrated heat transport is 1.4×1012 W
and 3.1 × 1012 W for the composite cyclonic and anticyclonic eddies, respectively. The magnitude is about
an order smaller than that reported by Chen et al. (2012). The reason may be that the anticyclonic eddies
reported by them are exceptional eddy cases that have much larger temperature anomalies and swirl
velocities than the composite eddies in this study. Actually, the heat transport is similar in magnitude to
the composite results at similar latitudes of the open ocean (Yang et al., 2015).
Meridional salt fluxes of the composite eddies are opposite in signs to the heat fluxes. Eddies of both pola-
rities produce negative (southward) salt transports in the west parts and positive (northward) salt transports
in the east parts (Figures 13c and 13d). The salt transports are mainly concentrated in the upper 150 m, with
peaks of ±1 × 102 kg · m2 · s1 at 50–60 m. The integrated profiles of salt transports show generally nega-
tive (southward) and positive (northward) transports by cyclonic and anticyclonic eddies, respectively
(Figure 13f). Similar to the heat transports, the salt transports are also strongly depth dependent. For both
cyclonic and anticyclonic eddies, approximately 95% of the salt transports are concentrated in the upper
300 m and 90% of them are concentrated in the upper 100 m. The depth-integrated total meridional salt

HE ET AL. 4921
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

Figure 13. Vertical sections of meridional (top) heat and (bottom) salt transports of the composite (left) cyclonic and (middle) anticyclonic eddies at y = 0 as in
Figures 12c and 12d, respectively. The right plots are depth-integrated heat and salt transports of the left and middle patterns. The blue lines are for cyclonic
eddies and red lines for anticyclonic eddies.

transport is 4.0 × 104 kg/s and 5.6 × 104 kg/s for the composite cyclonic and anticyclonic eddies,
respectively. The results suggest a southward salt transport by the rotation of cyclonic eddies and a
northward salt transport by the rotation of anticyclonic eddies.
Aside from the transports that are caused by the rotation of eddies, the propagation of eddies might also lead
to horizontal eddy transports (Early et al., 2011). These transports are mainly zonal because the propagations
of eddies are generally westward in the SCS. Based on formulas (3) and (4), the westward propagation of the
composite cyclonic eddy is estimated to generate 3.0 × 1012 W (eastward) heat transport and 7.4 × 104 kg/s
(westward) salt transport. In contrast, the composite anticyclonic eddy is estimated to result in 1.5 × 1012 W
(westward) heat transport and 6.0 × 104 kg/s (eastward) salt transport. The opposite directions of the heat
and salt transport induced by eddies of different polarities might play important roles in regional heat and
salt flux. Considering that there is on average 4.6 cyclonic and 4.5 anticyclonic eddies in the SCS at any given
time (section 3.1), the westward propagation of all the eddies may result in a net eastward heat transport of
7.1 × 1012 W and a net westward salt transport of 7.0 × 104 kg/s. Both of the heat and salt transports are the
same in orders of magnitudes to the corresponding transports caused by eddy rotation, suggesting the same
importance of eddy propagation and rotation in horizontal eddy transports.
Although thousands of eddy trajectories (625 cyclonic versus 534 anticyclonic) were tracked in the last
22 years, many of them dissipate locally and influence only near local regions. To evaluate the contribution
of eddies to basin scale circulation in the SCS, we defined 115°E as a reference section to estimate the zonal
mass transport induced by the propagation of eddies. Among these 1,159 eddies, 94 cyclonic and 114

HE ET AL. 4922
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

anticyclonic eddies were detected to propagate westward from the eastern SCS to the western SCS. Similar to
the estimation of heat transport by eddy propagation, the volume of fluid that is trapped by an eddy is
estimated by integrating eddy’s area from the surface to the “trapping depth.” As a result, the volume of
water trapped by the composite cyclonic and anticyclonic eddies are estimated to be 3.7 × 1012 m3 and
3.9 × 1012 m3, respectively. The values are 2 times of the estimation (of 2 × 1012 m3) based on relative vorticity
(Jia & Chassignet, 2011). Multiplying the numbers of eddies and dividing the study time period, the westward
volume transports are estimated to be 0.67 Sv (1Sv = 106 m3/s) by cyclonic eddies and 0.78 Sv by anticyclonic
eddies. This suggests a total westward water transport of 1.4 Sv by the westward propagation of eddies. The
value is approximately 30% of the annual-mean Luzon Strait transport (4.4 Sv), which is the mean value of
previous observations and model results (Fang et al., 2009; Nan et al., 2015).

5. Discussion
As presented in section 3 that the new daily eddy data provide a refined statistical analysis of eddies in the
SCS. Many of the obtained eddy properties show differences from previous ones ranging from slight to
substantial (Chen et al., 2011; Wang et al., 2003; Xiu et al., 2010). Multifactors are possibly responsible for
the differences. The temporal resolution of altimeter data may be an important one. Eddy data obtained from
daily altimeter data have about 7 times as many eddy snapshots as those obtained from previous weekly alti-
meter data. On one hand, more eddy samples may generally contribute to more accurate statistical results.
On the other hand, higher temporal resolution of altimeter data supports more accurate eddy detection,
especially for eddies of short lifespans. For example, an eddy that exists for 30 days would be detected by
daily altimeter data for 30 times, whereas it could only be detected by weekly altimeter observations for at
most 4 times. If the structure of the eddy is not regular at the formation or dissipation stage, it will possibly
be undetected by the weekly data. This helps explain the underestimation of cyclonic eddies, which generally
have shorter lifespans compared to anticyclonic eddies (Figure 3a), in previous studies (Chen et al., 2011;
Wang et al., 2003). The difference in eddy detection, that is, algorithm and criterion, is another factor that
may lead to differences in eddy statistics (Chen et al., 2011). As discussed in Chelton, Schlax, and Samelson
(2011), eddy detection in terms of sea surface height is superior to that in terms of the Okubo-Weiss para-
meter because of the obviation of noise by the double differentiation of sea surface height when compute
the Okubo-Weiss parameter. The former method yields significantly more eddies than the later (Chelton
et al., 2007; Chelton, Schlax, & Samelson, 2011). Moreover, Wang et al. (2003) filtered out eddies with lifespans
less than 60 days and Xiu et al. (2010) discarded eddies found over water depths shallower than 1,000 m.
These procedures excluded lots of eddies of short lifespans or near the coasts. The difference in data period
might also be a factor for that eddies have large interannual variability in the SCS. To address this, we repro-
duced Figures 2–5 using the same data period (1993–2009) to Chen et al. (2011). The results show that the
differences still hold true, suggesting that the data period does not apparently influence the eddy statistics
(figures not shown).
Previous studies have demonstrated that the features of eddies at the surface present apparent seasonal
variability (Chen et al., 2011; He, Zhan, Cai, & Li, 2016; Xiu et al., 2010). To reveal the seasonality of the eddies
beneath the surface, we compared the mean profiles of temperature and salinity anomalies in eddies
between summer (June to November) and winter (December to the next May). The results show that the pro-
files of temperature and salinity anomalies have qualitatively similar shape in different seasons (Figure 14).
When looking more specifically at the profiles, eddies of both polarities generally have larger temperature
anomalies in winter than in summer, especially bellow the cores of eddies. For cyclonic eddies, the peak of
temperature anomalies in summer is 10 m shallower and 1 °C larger than that in winter. The primary reason
may be that intense stratification in summer shallows the thermocline and increases the temperature gradi-
ent in the thermocline (Zeng & Wang, 2016). A shoaled peak of temperature anomalies is also observed for
anticyclonic eddies in summer, but with smaller temperature anomalies than in winter. Weak seasonality is
observed for salinity anomalies, except for a summer shoaling of the peaks of salinity anomalies in both cyclo-
nic and anticyclonic eddies.
Although Zhang et al. (2013) proposed a universal structure for eddies in the world ocean, previous studies
have derived various composite three-dimensional structures of eddies in different regions of the world
ocean, that is, the Kuroshio Extension (Sun et al., 2017), the southeastern tropical Indian Ocean (Yang et al.,

HE ET AL. 4923
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

Figure 14. Seasonality of mean vertical profiles of (left) temperature and (right) salinity anomalies inside eddies (<1R). The
blue lines are for cyclonic eddies and red lines for anticyclonic eddies. The solid lines denote summer (June to November),
and the dashed lines denote winter (December to the next May). The shadows are standard errors as defined in Figure 6.

2015), and the eastern South Pacific Ocean (Chaigneau et al., 2011). Moreover, significant differences can
even exists in the subregions of one region (Chaigneau et al., 2011; Mason et al., 2017). We repeated the
composite analysis in the northern (17°N–22°N and 109°E–120°E), western (6°N–17°N and 109°E–114°E),
and eastern (11°N–16°N and 114°E–120°E) SCS and derived the corresponding average profiles of
temperature and salinity anomalies in eddies (Figure 15). The results show that temperature anomalies in
the eastern SCS decrease much faster with depth than those in the other two regions, suggesting that
eddies have the shallowest vertical extends in the eastern SCS. Another important feature is that cyclonic
eddies have much larger temperature and salinity anomalies in the western SCS than those in the other
two regions. This result suggests that the intensities of cyclonic eddies in the western SCS are much larger

Figure 15. Regional variability of mean vertical profiles of (left) temperature and (right) salinity anomalies in eddies (<1R).
The blue lines are for cyclonic eddies and red lines for anticyclonic eddies. The solid lines denote the northern SCS
(17°N–22°N and 109°E–120°E), the dashed lines the western SCS (6°N–17°N and 109°E–114°E), and the dotted lines the
eastern SCS (11°N–16°N and 114°E–120°E). The shadows are standard errors as defined in Figure 6.

HE ET AL. 4924
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

Figure 16. Mean vertical profiles of temperature anomalies inside (a) cyclonic and (b) anticyclonic eddies as a function of
their surface signature in eddy amplitude. The right plots (c) are maximum temperature anomalies in the cores of
cyclonic (blue) and anticyclonic (red) eddies as a function of eddy amplitude. The asterisks are mean values, the bars are
standard errors, and the solid lines are linear regressions.

than those in the northern and eastern SCS. This scenario is supported by the surface signal that the
amplitudes of cyclonic eddies in the western SCS are much larger than those in the northern and eastern
SCS (Figure 4b). Previous case study reported a cyclonic eddy with maximum subsurface temperature
decrease of 8 °C (Hu et al., 2011), which is larger than the observations in other regions of the SCS (Chu &
Fan, 2001; Nan, He, et al., 2011; Shu et al., 2016). We noted that the number of available profiles located in
cyclonic eddies in the western SCS is only 167, which is much less than that in the northern SCS (414)
(Figure 1). Insufficient observational profiles might enlarge the uncertainty of the composite result. This is
also a reason why we constructed the composite three-dimensional structures of eddies in the whole SCS
rather than in the subregions currently. Further studies on the regional variability of eddies in the SCS are
expected when adequate data is accumulated.
The structure of eddies varies regionally and seasonally, while this does not mean the composite analysis is
meaningless. In fact, the composite analysis reveals that the structure of eddies in the SCS presents specific
characters compared with those in the open oceans. The most apparent one is that these eddies have
large density anomalies but small vertical extents. The maximum magnitudes of density anomalies exceeds
0.4 kg/m3 in the cores of eddies of both polarities. In contrast, the value is only ~0.1 and ~0.27 kg/m3 for
eddies in the eastern South Pacific (Chaigneau et al., 2011) and the Kuroshio Extension region (Sun et al.,
2017), respectively. The vertical extents of eddies are 400–500 m in the SCS, while 800–1,000 m in the eastern
South Pacific and the Kuroshio Extension region (Chaigneau et al., 2011; Sun et al., 2017). This is opposite to
our general understanding that eddies with large density anomalies should have deep vertical extents. One
important reason is that eddies in the SCS produce positive salinity anomalies in cyclonic eddies and negative
salinity anomalies in anticyclonic eddies (Figure 10). The salinity anomalies intensify the positive and negative
density anomalies in cyclonic and anticyclonic eddies, respectively. This is opposite to the situations in the
eastern South Pacific and the Kuroshio Extension region, where negative (positive) salinity anomalies in
cyclonic (anticyclonic) eddies weaken the density anomalies (Chaigneau et al., 2011; Sun et al., 2017).
An alternative explanation for the shallow vertical extents of the eddies is that the intensities of eddies in the
SCS are smaller than those in the open oceans, that is, the Kuroshio Extension region (Chelton, Schlax, &
Samelson, 2011). Eddies with small amplitudes are suggested to produce weak temperature and salinity
anomalies, as well as shallow vertical extents (Chaigneau et al., 2011). To confirm this scenario, we plotted
the profiles of mean temperature anomalies in eddies as functions of eddy amplitude (Figure 16). The results
provide evidences that both the magnitude and vertical extent of temperature anomalies increase dramati-
cally with eddy amplitude. The magnitudes of temperature anomalies increase approximately 0.9 °C for

HE ET AL. 4925
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

cyclonic eddies and 1.1 °C for anticyclonic eddies when the amplitudes of eddies increase by 10 cm
(Figure 16c). Additionally, the average vertical extent of eddies exceeds 800 m when eddy amplitude is larger
than 20 cm. This is proved by in situ observations that intense eddies are able to extend vertically to more
than 600 m or even 1,000 m in the SCS (Chu et al., 2014; Nan, He, et al., 2011; Zhang et al., 2016). Noted that
most of the previous eddy case studies focused on prominent eddy phenomena with apparent perturbations
that are easy to be captured by ship-based surveys. These eddies often have amplitudes of more than 20 cm
and last for more than 2 months (Chu et al., 2014; Shu et al., 2016; Zhang et al., 2016). However, the number of
eddies with amplitude larger than 20 cm accounts for only 4% in the SCS (Figure 3). These results suggest that
compared with case studies, the composite results are more representable for the common structure of
eddies in the SCS.
It should be noted that there are some biases in the composite analysis. First, the Argo profiles are spatially
and temporally sparse compared to the satellite data, and the eddies are not entirely circular as well. The
compositing of thousands of eddies and Argo profiles onto a normalized eddy-centered coordinate may miti-
gate the bias but not eliminate it. Second, some intense eddies can cause relatively high velocities below
1,000 m (Zhang et al., 2016), the definition of 1,200 m as the reference depth of no motion is not suitable
for these cases. In addition, mesoscale eddies are usually not exactly geostrophic, which means the derived
geostrophic velocities may underestimate the velocity fields of eddies. Third, the estimation of heat and salt
transports induced by the propagation of eddies did not consider diapycnal mixing between eddies’ interiors
and the ambient. As stated by Lehahn et al. (2011), although a strong nonlinear eddy acts as a quasi-isolated
system, particles inside it may decrease at an rate of ~1‰ per day because of diapycnal mixing. The eddy
transports derived in this study therefore should be an upper estimate.

6. Summary
Based on the new version of daily eddy data, a refreshed eddy assessment in the SCS is proposed, in which
many new and/or refined eddy characteristics that are not resolved by previous weekly eddy data are
revealed. For instance, the number of cyclonic eddies is close to that of anticyclonic eddies; eddy lifespan,
amplitude, and radius for anticyclonic eddies are more skewed toward large values than those for cyclonic
eddies; the intensification stage accounts for about 30% of both cyclonic and anticyclonic eddies’ entire life;
the propagation speed of eddies along the northwestern boundary is significantly accelerated in winter
because of the background cyclonic circulation; the westward propagation of cyclonic eddies deflects south-
ward while that of anticyclonic eddies deflects northward in the western SCS; much more cyclonic eddies
dissipate along the northwestern boundary than anticyclonic eddies do; and the central SCS is also a region
of high eddy dissipation. These results contribute to a refined overview of mesoscale eddies in the SCS.
A full understanding of the three-dimensional structure of eddies are of primary importance for the investi-
gation of eddy dynamics and the associated heat and salt transports. The compositing of thousands of Argo
profiles onto normalized eddy-centered coordinates enables the construction of eddies’ mean three-
dimensional structure in the SCS. The composite results reveal that the core of cyclonic eddies is on average
30 m shallower than that of anticyclonic eddies (60 m versus 90 m). The composite cyclonic (anticyclonic)
eddy has a temperature anomaly of 1.5 °C (1.4 °C), salinity anomaly of 0.15 psu (0.16 psu), and density
anomaly of 0.55 kg/m3 (0.51 kg m3) near the eddy’s core. The temperature anomalies contribute about
90% of the density anomalies. The temperature and density anomalies extend vertically to 400–500 m, while
the salinity anomalies are apparent only in the upper 150 m. Compared to eddy peripheries, the average MLD
in cyclonic eddies is shoaled for 6 m and that in anticyclonic eddies is deepened for 9 m. Apparent dipole-like
temperature and salinity anomalies in the mixed layer demonstrate the importance of eddy stirring in mer-
idional heat and salt fluxes, especially for cyclonic eddies. The meridional heat flux induced by the rotation of
eddies is mostly concentrate in the upper 300 m, and the salt flux is concentrate in the upper 150 m. Depth-
integrated heat transport is of 1.4 × 1012 W for the composite cyclonic eddy and 3.1 × 1012 W for the com-
posite anticyclonic eddy. The values for salt transport are 4.0 × 104 kg/s and 5.6 × 104 kg/s, respectively.
Simultaneously, the westward propagation of the cyclonic (anticyclonic) eddy results in a zonal heat trans-
port of 3.0 × 1012 W (1.5 × 1012 W) and salt transport of 7.4 × 104 kg/s (6.0 × 104 kg/s). The propagations
of cyclonic and anticyclonic eddies contribute to a total westward water transport of 1.4 Sv, which is equiva-
lent to about 30% of the annual-mean Luzon Strait transport. These results emphasize the important roles of

HE ET AL. 4926
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

eddies in regional oceanic circulation in the SCS. Recent studies have indicated that eddy-induced horizontal
and vertical transports of biogeochemical properties are of great importance for the spatial redistribution of
surface chlorophyll. The uncover of eddies’ three-dimensional structure may provide references for the
future estimation, as well as model simulation, of eddies’ impacts on biogeochemical environments beneath
the surface.

Acknowledgments References
The authors acknowledge the AVISO
database for the eddy data (http://www. Aiki, H., Zhai, X., & Greatbatch, R. J. (2016). Energetics of the global ocean: The role of mesoscale eddies. In S. K. Behera & T. Yamagata (Eds.),
aviso.altimetry.fr) and the China Argo Indo-Pacific climate variability and predictability (pp. 109–134). Tokyo: World Scientific.
Real-time Data Center for the Argo data Byrne, D., Munnich, M., Frenger, I., & Gruber, N. (2016). Mesoscale atmosphere ocean coupling enhances the transfer of wind energy into the
(http://www.argo.org.cn). These Argo ocean. Nature Communications, 7, ncomms11867. https://doi.org/10.1038/ncomms11867
data were collected and made freely Chaigneau, A., Eldin, G., & Dewitte, B. (2009). Eddy activity in the four major upwelling systems from satellite altimetry (1992–2007).
available by the International Argo Progress in Oceanography, 83(1–4), 117–123. https://doi.org/10.1016/j.pocean.2009.07.012
Program and the national programs Chaigneau, A., Gizolme, A., & Grados, C. (2008). Mesoscale eddies off Peru in altimeter records: Identification algorithms and eddy
that contribute to it (http://www.argo. spatio-temporal patterns. Progress in Oceanography, 79(2–4), 106–119. https://doi.org/10.1016/j.pocean.2008.10.013
ucsd.edu, http://argo.jcommops.org). Chaigneau, A., Le Texier, M., Eldin, G., Grados, C., & Pizarro, O. (2011). Vertical structure of mesoscale eddies in the eastern South Pacific
This work was supported by the Key Ocean: A composite analysis from altimetry and Argo profiling floats. Journal of Geophysical Research, 116, C11025. https://doi.org/
Research Program of Frontier Sciences, 10.1029/2011JC007134
CAS, grant QYZDJ-SSW-DQC034; the Chelton, D. B., Gaube, P., Schlax, M. G., Early, J. J., & Samelson, R. M. (2011). The influence of nonlinear mesoscale eddies on near-surface
National Natural Science Foundation of oceanic chlorophyll. Science, 334(6054), 328–332. https://doi.org/10.1126/science.1208897
China (NSFC), grants 41430964 and Chelton, D. B., Schlax, M. G., & Samelson, R. M. (2011). Global observations of nonlinear mesoscale eddies. Progress in Oceanography, 91(2),
41776014; the Strategic Priority 167–216. https://doi.org/10.1016/j.pocean.2011.01.002
Research Program of the Chinese Chelton, D. B., Schlax, M. G., Samelson, R. M., & de Szoeke, R. A. (2007). Global observations of large oceanic eddies. Geophysical Research
Academy of Sciences, grant Letters, 34, L15606. https://doi.org/10.1029/2007GL030812
XDA11020201; the Project of State Key Chen, G., Gan, J., Xie, Q., Chu, X., Wang, D., & Hou, Y. (2012). Eddy heat and salt transports in the South China Sea and their seasonal mod-
Laboratory of Tropical Oceanography, ulations. Journal of Geophysical Research, 117, C05021. https://doi.org/10.1029/2011JC007724
grants LTOZZ1705 and LTOZZ1801; the Chen, G., Hou, Y., & Chu, X. (2011). Mesoscale eddies in the South China Sea: Mean properties, spatiotemporal variability, and impact on
National Postdoctoral Program for thermohaline structure. Journal of Geophysical Research, 116, C06018. https://doi.org/10.1029/2010JC006716
Innovative Talents, grant BX201700303; Cheng, X., McCreary, J. P., Qiu, B., Qi, Y., & Du, Y. (2017). Intraseasonal-to-semiannual variability of sea-surface height in the astern, equatorial
and the China Postdoctoral Science Indian Ocean and southern Bay of Bengal. Journal of Geophysical Research: Oceans, 122, 4051–4067. https://doi.org/10.1002/2016JC012662
Foundation, grant 2018 M631000. Cheng, X., & Qi, Y. (2010). Variations of eddy kinetic energy in the South China Sea. Journal of Oceanography, 66(1), 85–94. https://doi.org/
10.1007/s10872-010-0007-y
Cheng, X., Xie, S.-P., Du, Y., Wang, J., Chen, X., & Wang, J. (2016). Interannual-to-decadal variability and trends of sea level in the South China
Sea. Climate Dynamics, 46(9–10), 3113–3126. https://doi.org/10.1007/s00382-015-2756-1
Chu, P. C., & Fan, C. (2001). Low salinity, cool-core cyclonic eddy detected northwest of Luzon during the South China Sea Monsoon
Experiment (SCSMEX) in July 1998. Journal of Oceanography, 57(5), 549–563. https://doi.org/10.1023/A:1021251519067
Chu, X., Xue, H., Qi, Y., Chen, G., Mao, Q., Wang, D., & Chai, F. (2014). An exceptional anticyclonic eddy in the South China Sea in 2010.
Journal of Geophysical Research: Oceans, 119, 881–897. https://doi.org/10.1002/2013JC009314
Colas, F., McWilliams, J. C., Capet, X., & Kurian, J. (2011). Heat balance and eddies in the Peru-Chile current system. Climate Dynamics, 39(1–2),
509–529. https://doi.org/10.1007/s00382-011-1170-6
Dewar, W. K., & Hogg, A. M. (2010). Topographic inviscid dissipation of balanced flow. Ocean Modelling, 32(1–2), 1–13. https://doi.org/
10.1016/j.ocemod.2009.03.007
Dong, C., McWilliams, J. C., Liu, Y., & Chen, D. (2014). Global heat and salt transports by eddy movement. Nature Communications, 5(1), 3294.
https://doi.org/10.1038/ncomms4294
Dong, D., Brandt, P., Chang, P., Schütte, F., Yang, X., Yan, J., & Zeng, J. (2017). Mesoscale eddies in the northwestern Pacific Ocean:
Three-dimensional eddy structures and heat/salt transports. Journal of Geophysical Research: Oceans, 122, 9795–9813. https://doi.org/
10.1002/2017JC013303
Early, J. J., Samelson, R., & Chelton, D. B. (2011). The evolution and propagation of quasigeostrophic ocean eddies. Journal of Physical
Oceanography, 41(8), 1535–1555. https://doi.org/10.1175/2011JPO4601.1
Faghmous, J. H., Frenger, I., Yao, Y., Warmka, R., Lindell, A., & Kumar, V. (2015). A daily global mesoscale ocean eddy dataset from satellite
altimetry. Scientific Data, 2, 150028. https://doi.org/10.1038/sdata.2015.28
Fang, G., Wang, Y., Wei, Z., Fang, Y., Qiao, F., & Hu, X. (2009). Interocean circulation and heat and freshwater budgets of the South China Sea
based on a numerical model. Dynamics of Atmospheres and Oceans, 47(1–3), 55–72. https://doi.org/10.1016/j.dynatmoce.2008.09.003
Flierl, G. R. (1981). Particle motions in large-amplitude wave fields. Geophysical and Astrophysical Fluid Dynamics, 18(1–2), 39–74. https://doi.
org/10.1080/03091928108208773
Frenger, I., Gruber, N., Knutti, R., & Münnich, M. (2013). Imprint of Southern Ocean eddies on winds, clouds and rainfall. Nature Geoscience,
6(8), 608–612. https://doi.org/10.1038/ngeo1863
Frenger, I., Münnich, M., Gruber, N., & Knutti, R. (2015). Southern Ocean eddy phenomenology. Journal of Geophysical Research: Oceans, 120,
7413–7449. https://doi.org/10.1002/2015JC011047
Gaube, P., Chelton, D. B., Samelson, R. M., Schlax, M. G., & O’Neill, L. W. (2015). Satellite observations of mesoscale eddy-induced Ekman
pumping. Journal of Physical Oceanography, 45(1), 104–132. https://doi.org/10.1175/jpo-d-14-0032.1
Gaube, P., McGillicuddy, D. J., Chelton, D. B., Behrenfeld, M. J., & Strutton, P. G. (2014). Regional variations in the influence of mesoscale eddies
on near-surface chlorophyll. Journal of Geophysical Research: Oceans, 119, 8195–8220. https://doi.org/10.1002/2014JC010111
Grachev, Y., Koshlyakov M., Tikhomirova T., & Yenikeyev Y. (1979). Synoptic eddy field in the POLYMODE area, September 1977 to May 1978,
Polymode News, 69, 1–6.
Hausmann, U., McGillicuddy, D. J., & Marshall, J. (2017). Observed mesoscale eddy signatures in Southern Ocean surface mixed-layer depth.
Journal of Geophysical Research: Oceans, 122, 617–635. https://doi.org/10.1002/2016JC012225
He, Q., Zhan, H., Cai, S., & Li, Z. (2016). Eddy effects on surface chlorophyll in the northern South China Sea: Mechanism investigation and
temporal variability analysis. Deep Sea Research Part I: Oceanographic Research Papers, 112, 25–36. https://doi.org/10.1016/j.dsr.2016.03.004

HE ET AL. 4927
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

He, Q., Zhan, H., Cai, S., & Zha, G. (2016). On the asymmetry of eddy-induced surface chlorophyll anomalies in the southeastern Pacific: The
role of eddy-Ekman pumping. Progress in Oceanography, 141, 202–211. https://doi.org/10.1016/j.pocean.2015.12.012
He, Q., Zhan, H., Shuai, Y., Cai, S., Li, Q. P., Huang, G., & Li, J. (2017). Phytoplankton bloom triggered by an anticyclonic eddy: The combined
effect of eddy-Ekman pumping and winter mixing. Journal of Geophysical Research: Oceans, 122, 4886–4901. https://doi.org/10.1002/
2017JC012763
He, Y., Xie, J., & Cai, S. (2016). Interannual variability of winter eddy patterns in the eastern South China Sea. Geophysical Research Letters, 43,
5185–5193. https://doi.org/10.1002/2016GL068842
Hu, J., Gan, J., Sun, Z., Zhu, J., & Dai, M. (2011). Observed three-dimensional structure of a cold eddy in the southwestern South China Sea.
Journal of Geophysical Research, 116, C05016. https://doi.org/10.1029/2010JC006810
Jan, S., Mensah, V., Andres, M., Chang, M.-H., & Yang, Y. J. (2017). Eddy-Kuroshio interactions: Local and remote effects. Journal of Geophysical
Research: Oceans, 122, 9744–9764. https://doi.org/10.1002/2017JC013476
Jia, Y., & Chassignet, E. P. (2011). Seasonal variation of eddy shedding from the Kuroshio intrusion in the Luzon Strait. Journal of
Oceanography, 67(5), 601–611. https://doi.org/10.1007/s10872-011-0060-1
Kara, A. B., Rochford, P. A., & Hurlburt, H. E. (2000). An optimal definition for ocean mixed layer depth. Journal of Geophysical Research, 105(C7),
16,803–16,821. https://doi.org/10.1029/2000JC900072
Lehahn, Y., d’Ovidio, F., Lévy, M., Amitai, Y., & Heifetz, E. (2011). Long range transport of a quasi isolated chlorophyll patch by an Agulhas ring.
Geophysical Research Letters, 38, L16610. https://doi.org/10.1029/2011GL048588
Liu, F. F., Tang, S. L., & Chen, C. Q. (2013). Impact of nonlinear mesoscale eddy on phytoplankton distribution in the northern South China Sea.
Journal of Marine Systems, 123, 33–40. https://doi.org/10.1016/j.jmarsys.2013.04.005
Ma, J., Xu, H., Dong, C., Lin, P., & Liu, Y. (2015). Atmospheric responses to oceanic eddies in the Kuroshio Extension region. Journal of
Geophysical Research: Atmospheres, 120, 6313–6330. https://doi.org/10.1002/2014JD022930
Mason, E., Pascual, A., Gaube, P., Ruiz, S., Pelegrí, J. L., & Delepoulle, A. (2017). Subregional characterization of mesoscale eddies
across the Brazil-Malvinas Confluence. Journal of Geophysical Research: Oceans, 122, 3329–3357. https://doi.org/10.1002/
2016JC012611
McGillicuddy, D. J. Jr. (2016). Mechanisms of physical-biological-biogeochemical interaction at the oceanic mesoscale. Annual Review of
Marine Science, 8(1), 125–159. https://doi.org/10.1146/annurev-marine-010814-015606
Millero, F. J., Chen, C.-T., Bradshaw, A., & Schleicher, K. (1980). A new high pressure equation of state for seawater. Deep Sea Research Part A.
Oceanographic Research Papers, 27(3–4), 255–264. https://doi.org/10.1016/0198-0149(80)90016-3
Monterey, G., & Levitus, S. (1997). Seasonal variability of mixed layer depth for the world ocean, NOAA Atlas NESDIS 14 (p. 100). Silver Spring, Md:
Natl. Oceanic and Atmos. Admin.
Morrow, R. (2004). Divergent pathways of cyclonic and anti-cyclonic ocean eddies. Geophysical Research Letters, 31, L24311. https://doi.org/
10.1029/2004GL020974
Nan, F., He, Z., Zhou, H., & Wang, D. (2011). Three long-lived anticyclonic eddies in the northern South China Sea. Journal of Geophysical
Research, 116, C05002. https://doi.org/10.1029/2010JC006790
Nan, F., Xue, H., Xiu, P., Chai, F., Shi, M., & Guo, P. (2011). Oceanic eddy formation and propagation southwest of Taiwan. Journal of Geophysical
Research, 116, C12045. https://doi.org/10.1029/2011JC007386
Nan, F., Xue, H., & Yu, F. (2015). Kuroshio intrusion into the South China Sea: A review. Progress in Oceanography, 137, 314–333. https://doi.
org/10.1016/j.pocean.2014.05.012
Nan, F., Yu, F., Xue, H., Zeng, L., Wang, D., Yang, S., & Nguyen, K.-C. (2016). Freshening of the upper ocean in the South China Sea since the
early 1990s. Deep Sea Research Part I: Oceanographic Research Papers, 118, 20–29. https://doi.org/10.1016/j.dsr.2016.10.010
Pilo, G. S., Mata, M. M., & Azevedo, J. L. L. (2015). Eddy surface properties and propagation at Southern Hemisphere western boundary current
systems. Ocean Science, 11(4), 629–641. https://doi.org/10.5194/os-11-629-2015
Pujol, M.-I., Faugère, Y., Taburet, G., Dupuy, S., Pelloquin, C., Ablain, M., & Picot, N. (2016). DUACS DT2014: The new multi-mission altimeter
data set reprocessed over 20 years. Ocean Science, 12(5), 1067–1090. https://doi.org/10.5194/os-12-1067-2016
Qiu, B., & Chen, S. (2005). Eddy-induced heat transport in the subtropical North Pacific from Argo, TMI, and altimetry measurements. Journal
of Physical Oceanography, 35(4), 458–473. https://doi.org/10.1175/JPO2696.1
Qu, T. (2000). Upper-layer circulation in the South China Sea. Journal of Physical Oceanography, 30(6), 1450–1460. https://doi.org/10.1175/
1520-0485(2000)030%3C1450:ULCITS%3E2.0.CO;2
Qu, T., Song, Y. T., & Yamagata, T. (2009). An introduction to the South China Sea throughflow: Its dynamics, variability, and application for
climate. Dynamics of Atmospheres and Oceans, 47(1–3), 3–14. https://doi.org/10.1016/j.dynatmoce.2008.05.001
Quan, Q., Xue, H., Qin, H., Zeng, X., & Peng, S. (2016). Features and variability of the South China Sea western boundary current from 1992 to
2011. Ocean Dynamics, 66(6–7), 795–810. https://doi.org/10.1007/s10236-016-0951-1
Roemmich, D., & Gilson, J. (2001). Eddy transport of heat and thermocline waters in the North Pacific: A key to interannual/decadal climate
variability? Journal of Physical Oceanography, 31(3), 675–687. https://doi.org/10.1175/1520-0485(2001)031%3C0675:ETOHAT%3E2.0.CO;2
Saenko, O. A., Zhai, X., Merryfield, W. J., & Lee, W. G. (2012). The combined effect of tidally and eddy-driven diapycnal mixing on the large-
scale ocean circulation. Journal of Physical Oceanography, 42(4), 526–538. https://doi.org/10.1175/JPO-D-11-0122.1
Samelson, R. M., Schlax, M. G., & Chelton, D. B. (2014). Randomness, symmetry, and scaling of mesoscale eddy life cycles. Journal of Physical
Oceanography, 44(3), 1012–1029. https://doi.org/10.1175/jpo-d-13-0161.1
Shu, Y., Xiu, P., Xue, H., Yao, J., & Yu, J. (2016). Glider-observed anticyclonic eddy in northern South China Sea. Aquatic Ecosystem Health &
Management, 19(3), 233–241. https://doi.org/10.1080/14634988.2016.1208028
Su, J. (2004). Overview of the South China Sea circulation and its influence on the coastal physical oceanography outside the Pearl River
Estuary. Continental Shelf Research, 24(16), 1745–1760. https://doi.org/10.1016/j.csr.2004.06.005
Sun, W., Dong, C., Wang, R., Liu, Y., & Yu, K. (2017). Vertical structure anomalies of oceanic eddies in the Kuroshio Extension region. Journal of
Geophysical Research: Oceans, 122, 1476–1496. https://doi.org/10.1002/2016JC012226
Villas Bôas, A. B., Sato, O. T., Chaigneau, A., & Castelão, G. P. (2015). The signature of mesoscale eddies on the air-sea turbulent heat fluxes in
the South Atlantic Ocean. Geophysical Research Letters, 42, 1856–1862. https://doi.org/10.1002/2015GL063105
Wang, G., Chen, D., & Su, J. (2008). Winter eddy genesis in the eastern South China Sea due to orographic wind jets. Journal of Physical
Oceanography, 38(3), 726–732. https://doi.org/10.1175/2007jpo3868.1
Wang, G., Su, J., & Chu, P. C. (2003). Mesoscale eddies in the South China Sea observed with altimeter data. Geophysical Research Letters, 30,
2121. https://doi.org/10.1029/2003GL018532
Wang, G., Su, J., & Qi, Y. (2005). Advances in studying mesoscale eddies in South China Sea. Advances in Earth Science, 20(8), 882–886. https://
doi.org/10.11867/j.issn.1001-8166.2005.08.0882

HE ET AL. 4928
21699291, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JC014054 by Sri Lanka National Access, Wiley Online Library on [19/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Oceans 10.1029/2018JC014054

Williams, S., Petersen, M., Bremer, P.-T., Hecht, M., Pascucci, V., Ahrens, J., et al. (2011). Adaptive extraction and quantification of geophysical
vortices. IEEE Transactions on Visualization and Computer Graphics, 17(12), 2088–2095. https://doi.org/10.1109/TVCG.2011.162
Xiu, P., Chai, F., Shi, L., Xue, H., & Chao, Y. (2010). A census of eddy activities in the South China Sea during 1993–2007. Journal of Geophysical
Research, 115, C03012. https://doi.org/10.1029/2009JC005657
Yang, G., Wang, F., Li, Y., & Lin, P. (2013). Mesoscale eddies in the northwestern subtropical Pacific Ocean: Statistical characteristics and
three-dimensional structures. Journal of Geophysical Research: Oceans, 118, 1906–1925. https://doi.org/10.1002/jgrc.20164
Yang, G., Yu, W., Yuan, Y., Zhao, X., Wang, F., Chen, G., et al. (2015). Characteristics, vertical structures, and heat/salt transports of mesoscale
eddies in the southeastern tropical Indian Ocean. Journal of Geophysical Research: Oceans, 120, 6733–6750. https://doi.org/10.1002/
2015JC011130
Yang, H., & Liu, Q. (2003). Forced Rossby wave in the northern South China Sea. Deep Sea Research Part I: Oceanographic Research Papers,
50(7), 917–926. https://doi.org/10.1016/s0967-0637(03)00074-8
Zeng, L., & Wang, D. (2016). Seasonal variations in the barrier layer in the South China Sea: Characteristics, mechanisms and impact of
warming. Climate Dynamics, 48(5–6), 1911–1930. https://doi.org/10.1007/s00382-016-3182-8
Zhai, X., Johnson, H. L., & Marshall, D. P. (2010). Significant sink of ocean-eddy energy near western boundaries. Nature Geoscience, 3(9),
608–612. https://doi.org/10.1038/ngeo943
Zhang, Y., Liu, Z., Zhao, Y., Wang, W., Li, J., & Xu, J. (2014). Mesoscale eddies transport deep-sea sediments. Scientific Reports, 4(1), 5937.
https://doi.org/10.1038/srep05937
Zhang, Z., Tian, J., Qiu, B., Zhao, W., Chang, P., Wu, D., & Wan, X. (2016). Observed 3D structure, generation, and dissipation of oceanic
mesoscale eddies in the South China Sea. Scientific Reports, 6(1), 24349. https://doi.org/10.1038/srep24349
Zhang, Z., Wang, W., & Qiu, B. (2014). Oceanic mass transport by mesoscale eddies. Science, 345(6194), 322–324. https://doi.org/10.1126/
science.1252418
Zhang, Z., Zhang, Y., Wang, W., & Huang, R. X. (2013). Universal structure of mesoscale eddies in the ocean. Geophysical Research Letters, 40,
3677–3681. https://doi.org/10.1002/grl.50736
Zhang, Z., Zhao, W., Qiu, B., & Tian, J. (2017). Anticyclonic eddy sheddings from Kuroshio loop and the accompanying cyclonic eddy in the
northeastern South China Sea. Journal of Physical Oceanography, 47(6), 1243–1259. https://doi.org/10.1175/JPO-D-16-0185.1
Zheng, Q., Xie, L., Zheng, Z., & Hu, J. (2017). Progress in research of mesoscale eddies in the South China Sea. Advance in Marine Science, 35(2),
131–158. https://doi.org/10.3969/j.issn.1671-6647.2017.02.001

HE ET AL. 4929

You might also like