A cyclonic eddy detected in the northwestern South China Sea during the summer of 2011 is investigated using both satellite data and several sets of in situ observations including cruise measurements and a mooring line. The eddy was generated to the west of the Zhongsha Islands in early June and then propagated northwestward. In July and August, when the cyclonic gyre originating from the offshore jet recirculation east of Vietnam matured, the eddy was absorbed into the gyre and strengthened by it. The eddy regulated the thermohaline characteristics and the currents of the entire water column. Along-slope boundary currents (toward the northeast in summer) were completely reversed from the surface to about 1300 m below. The thermohaline in deep water (660–950 m) showed a significant response to the presence of the eddy, with the isotherms and isohalines rising up to 100 m. The recurrence of the phenomenon is further explored by a statistical analysis of satellite observations. Results suggest the presence of a cyclonic eddy and abnormal offshore jets are closely associated with the El Nino Southern Oscillation events.
The South China Sea (SCS) is a large, semi-enclosed tropical marginal sea located in the northwest Pacific Ocean (Figure 1a). Driven by the seasonally reversing monsoon, the general circulation pattern in the SCS is cyclonic in winter and anticyclonic in summer (Wyrtki, 1961; Hu et al., 2000; Qu, 2000; Su, 2004). The boundary current exhibits a strong seasonal variation in its direction and pattern (Liu et al., 2002; Fang et al., 2012; Wang et al., 2013). During winter, the strong current flows southwestward along the shelf and slope of the northern SCS, and then turns southward along the central and southern Vietnamese coasts. In summer, separated from the boundary current along the Vietnamese coast (near 11°N), an offshore jet moving east induces an eddy pair to the east of Vietnam, which is composed of an anticyclonic eddy to the south of the eastward jet and a cyclonic eddy anchored to the north (Wu et al., 1998, 1999; Chen et al., 2010). The boundary current in the northwestern SCS is highly complex. It is influenced by various factors, such as prevailing monsoons, irregular topography, the eddy pair, and the Kuroshio intrusion.
Along the shelf edges, many eddies are associated with the boundary current and its variability in the SCS (Wang et al., 2000; Liu et al., 2001; He et al., 2002; Wang et al., 2003, 2008). Also, the baroclinic instability was found to be responsible for the energy transfer between the boundary current and the eddies (Chen et al., 2012). As an important area for the extinction of eddies, the Xisha region has a relatively low mean annual eddy kinetic energy (EKE) (Zhuang et al., 2010; Chen et al., 2011). However, field observations and altimeter measurements have indicated that the summer boundary currents moving northeast are closely related to the anticyclonic eddies or meanders in the northwestern SCS (Hwang and Chen, 2000; Nan et al., 2011; Fang et al., 2012; Chu et al., 2013), where strong interactions between the eddy and current due to the presence of complex topography exist (Wang et al., 2013). Although more attention has been paid to the boundary current and its associated eddies because of altimeter data and ocean models (Hwang and Chen, 2000; Morimoto et al., 2000; Chern and Wang, 2003; Yu et al., 2007), the structures and variations of these eddies and their interactions with boundary currents are still not well understood, especially with respect to their in situ observations.
In this study, a prolonged cyclonic eddy in the northwestern SCS are investigated based on the satellite measurements, cruise observations, and an array of moored temperature, salinity, and current meter records, which are anchored to the northwest of the Xisha Islands and located in the path of the SCS boundary current. The eddy, coupled with the cyclonic gyre originating from the offshore jet recirculation east of Vietnam, lasted for the whole summer of 2011 and significantly influenced the thermohaline structure of the entire water column in the western Xisha Islands.
The sea surface signals of the eddy can be found using the satellite's data. The sea level anomaly (SLA) data used in this study were obtained from the merged satellite Archiving, Validation, and Interpolation of Satellite Oceanographic (AVISO) data project. The merged data from the combination of Jason, TOPEX/Poseidon, Envisat, GFO, ERS, and Geosat altimeters were interpolated onto a global grid of 1/4° resolution and archived in weekly averaged frames. The set includes all data from October 1992 to the present. Sea surface temperature (SST) data from June to September 2011 were collected from the Group for High-Resolution Sea Surface Temperature (GHRSST). The data is available with a 1/4° daily resolution.
The evolution of the eddy features over time is explored using the SLA-based eddy detection algorithm (Chaigneau et al., 2009; Chaigneau et al., 2011). The center of the eddy is determined by a local SLA minimum. Closed SLA contours are searched for with an increase of 0.1 cm. The eddy boundary is defined as the outermost closed SLA contour, embedding only the considered center. Given the accuracy of satellite measurements and AVISO product, we only considered eddies having SLA amplitudes higher than 2 cm, a minimum diameter greater than 35 km, and a lifetime not shorter than 4 weeks. Once the boundary and the center of the eddy were established, its main features could be derived by the equivalent radius, ΔSLA = SLAedge - SLAcenter, and EKE = (u'2 + v'2)/2, where u' and v' are the zonal and meridional components of the geostrophic velocity anomaly, respectively.
In situ data
The vertical structure of the eddy was determined using in situ observations, including the temperature and salinity data from 11 July to 23 August 2011 and the data recorded from June 2007 to August 2012 by a mooring line. The mooring line was deployed at approximately 17.1°N and 110.3°E in the northwestern SCS (Figure 1), which is in the path of seasonally reversing boundary current. Mooring maintenance was conducted once per year in order to recover the data stored in the instruments and change the batteries. An up-looking RDI Workhorse Long Ranger 75 kHz acoustic Doppler current profiler (ADCP) was deployed at a depth of 450–650 m during the operational period. Its depth was ∼650 m from August 2010 to August 2011, and ∼550 m since August 2011. Another down-looking ADCP was deployed at ∼1150 m depth and has remained since May 2006. During September 2010 and August 2011, the line consisted of 20 conductivity–temperature (CT) profilers 665–950 m deep. Observations of thermohaline properties and currents were collected at sampling rates of 25 min for the CT and 30 min or 1 h for the ADCP.
Moreover, the conductivity–temperature–depth (CTD) profiler data and current vectors measured by the shipboard ADCP along the 18°N section are used in this study. This data was derived from the cruise observation conducted in the northern SCS from 27 August to 2 September 2011. There are 11 observational stations 18°N between 110°E and 115°E. Furthermore, an Argo buoy (No. 2901143) was found near the eddy from 26 June to 24 July 2011. Its trajectory is shown in Figure 1c. The thermohaline profiles were obtained from the Global Argo Profiling Observations Dataset provided by the China Argo Real-time Data Center (ftp://ftp.argo.org.cn/pub/ARGO/global/).
Results and discussion
Evolution of the eddy associated with boundary current
The eddy was detected in the northwestern SCS between early June and late September 2011 using the satellite measurements (Figure 1). Figures 1b–d illustrate the main features of the eddy, how they evolved with the topography and the change of boundary current, and the associated SLA image and corresponding surface geostrophic velocity on 31 August. The eddy was generated southwest of the Zhongsha Islands, propagated northwestward slowly, and became stationary around the Xisha Islands in July. During its propagation, the eddy weakened, with the ΔSLA sharply decreasing, probably due to the negative impact of the complex topography (Sutyrin and Grimshaw, 2010) around the Xisha Islands and the continental shelf and slope (Figures 1a–c). The cyclonic gyre east of the Vietnamese coast formed in July and became mature in August (Figures 1b and d). In the meantime, the eddy merged into and was reinforced by the gyre. The eddy's radius, EKE, and ΔSLA were all enhanced, reaching their peaks around early August. In the early fall, winter monsoons gradually invaded the SCS from the north, and a winter cyclonic gyre in the whole basin gradually formed (Liu et al., 2002; Fang et al., 2012). As a result, the eddy was strengthened by the winter gyre and finally merged into it in October.
The maintenance of the eddy was closely related to the cyclonic gyre east of Vietnam. The formation of this gyre and its inter-annual variation have been found to be associated with the recirculation caused by the baroclinic instability of the Vietnam offshore jet (Hwang and Chen, 2000; Hu et al., 2011; Fang et al., 2012). As shown in Figure 1d, the offshore jet extended remarkably eastward in the summer of 2011. The intensification of the jet induced a large cyclonic gyre, by which the eddy was strengthened around July and August (Figure 1b).
The footprint of the relatively cold water is consistent with the gyre and the eddy, detected using the SST image from 31 August (Figure 1e). Influenced by the offshore jet, the colder water was transported eastward from the coast to the deep basin, and then extended cyclonically around the eddy. This result further suggests that the gyre induced by the recirculation originated from the offshore jet and resulted in the eddy being embedded into the gyre.
Current and thermohaline properties response to the eddy
The eddy's development was recorded at different depths by the moored array of instruments. The current profiles recorded by the moored ADCP are shown in Figures 2a and b. The current mainly travels southwestward in winter and northeastward in summer. The current directions are generally uniform over the entire water column, although they move slowly in the deeper water. The northeastward currents usually take place during July and September. Modulated by the eddy, the current's direction was completely reversed and became strong from the surface to a depth of about 1300 m during July 2011. Even in August and September, the currents above 500 m still kept moving southwestward.
The reversed boundary current in 2011 is further confirmed by the current field, measured by the shipboard ADCP (Figures 2c and d). The cyclonic current field clearly shows the hydrodynamic characteristics in the northwestern part of the eddy. The southwestward current appears, not just locally around the mooring line, but in a wide region located in the northwestern SCS as well.
The changes in temperature and salinity of the water caused by the eddy were recorded by the moored instruments (Figures 3c and 3d). In the middle of July, a significant decrease of temperature (∼1°C) and increase of salinity (∼0.05) were observed in the water column from a depth of 660 m to 950 m. After the eddy settled in the Xisha region, the responses diminished and the temperature and salinity returned back to their original state. During the episode, the uplifted ranges of isotherms and isohalines were as high as ∼100 m. In comparison, the isotherms and isohalines along the 18°N section also show a remarkable uplift at about 112°E (Figures 3e and f). It is noteworthy that the thermohaline response in deep water near the boundary region is double that of the subsurface water of the adjacent open ocean.
The significant thermohaline response in deep water near the boundary region appeared not induced by the pumping effect because the mooring locates at the edge area of the eddy (Figure 1c). Instead, it could be attributed to two causes. First, the eddy brings the colder and saltier water from the basin to the boundary area during its northwestward propagation. The thermohaline property of the eddy in its formation stage was recorded by an Argo buoy released southeast of it (Figures 3a and b). The buoy moved northeastward under the current associated with the eddy and finally escaped from it (Figure 1c). During its first month, the eddy's isotherms and isohalines generally decreased, indicating that the eddy contained the colder and saltier water, which is consistent with the general features of cyclonic eddies in the SCS (Xiu et al., 2010; Chen et al., 2011). Second, resulting from the strong interaction of the southwestward eddy current and steep topography (Figure 1c), the eddy-related upwelling was induced in the deep layers. The thermohaline responses began diminishing in the middle of August (Figures 3c and d). At the same time, the deep ocean currents changed their direction from southwestward to northeastward (Figure 2b), a time period consistent with the hypothesis.
Recurrence of the phenomenon
The above analysis has shown that the presence of a cyclonic eddy in the Xisha region can distort the boundary current and play an important role in regulating the hydrological structure of the entire water column. Thus, the frequency of these phenomena taking place might have implications for the hydrodynamics of the area. Aiming to determine the repeatability of the episode, we used the eddy detection algorithm (Chaigneau et al., 2009; Chaigneau et al., 2011) to identify cyclonic eddies in the same region where the mooring was deployed (solid-line box of 15–18°N, 110–113°E, in the map in Figure 1a) from 1993 to 2011. Only those eddies whose centers remained inside the region during summer (June–August) were chosen.
Results show cyclonic eddies dominating the region in six of the 19 summers. Table 1 lists their starting and ending dates, duration, ΔSLA, radius, EKE, and the relative EKE compared to the eddy studied in this article, sorted by decreasing EKE. The mean duration of the selected episodes is 7.8 weeks per summer, with a standard deviation of 2.9 weeks. The mean radius is 118 km with a standard deviation of 19 km. The eddy studied in this article was the longest recorded, spanning 11 weeks, and had the largest radius. However, the eddy was just the fifth most intense of the selected episodes. The eddy with the highest ΔSLA and EKE occurred in the summer of 2009. The mooring data also show the southwestward current during a large part of the summer. The northeastward current only appears in August (Figure 2a), when the cyclonic eddy moved southeastward and became far away from the mooring (figure not shown).
|Start .||End .||Duration (weeks) .||ΔSLA (m) .||Radius (km) .||EKE (cm2 s−2) .||Percentage (EKE/EKEOur) .|
|Start .||End .||Duration (weeks) .||ΔSLA (m) .||Radius (km) .||EKE (cm2 s−2) .||Percentage (EKE/EKEOur) .|
Most of the episodes (5) take place during the El Niño development (2002, 2004 and 2009) and/or the subsequent summers of La Niña (2006, 2009 and 2011). None of the episodes took place in a summer following an El Niño event. This fact is related to the intensification of the southwesterly monsoon during the summers of El Niño development and/or the subsequent summers of the La Niña years (Wang et al., 2009), which intensified the off-shore jet and the gyre east of Vietnam.
During summer the boundary current mainly flows northeastwards in the northwestern SCS (Hwang and Chen, 2000; Fang et al., 2012; see also Figure 2a), driven by the southwesterly monsoon. Associated with the current, the Xisha region is usually dominated by anticyclonic eddies or meanders (Nan et al., 2011; Chu et al., 2013). In contrast, our results indicate that cyclonic eddies associated with the southwestward boundary current are also common phenomena in the northwestern SCS.
In 2011, the northwestern SCS was dominated by a cyclonic eddy for the entire summer. The eddy was maintained by the gyre, which originated from the intensive offshore jet east of Vietnam. The eddy modified the thermohaline characteristics and the currents of the entire water column, especially the along-slope boundary currents, which were completely reversed from the surface to a depth of about 1300 m. Furthermore, the thermohaline property in the water column of deeper layers suggests a significant response to the eddy, with the isotherms and isohalines uplifted for up to 100 m.
The recurrence of the cyclonic eddies in the northwestern SCS during the summer was further determined to be related to the ENSO events. Most selected episodes took place during the El Niño development and/or the subsequent summer of La Niña, when the offshore jet was significantly intensified.
The authors thank the crew of Shiyan 3 and all the cruise participants for help with field work.
This study was supported by Strategic Priority Research Program of the Chinese Academy of Sciences (XDA11010203 and XDA11020201), the Knowledge Innovation Program of the Chinese Academy of Sciences (KZCX2-EW-204, SQ201101, LTOZZ1202 and LTOZZ1304) and National Natural Science Foundation of China (U1033003, 41276022, 41206010, 41130855 and 41406039).