Based on satellite altimeter data, it was found that the mesoscale eddy propagates along the continental slope in northern South China Sea. When the mesoscale eddy passed the Dongsha Islands, the propagation speed clearly slowed down. Additionally, the radius increased and strength decreased when the mesoscale eddy touched the Dongsha Islands. Using the analytic expression for the speed of the center of an isolated eddy, it was found that the sudden slowdown of the zonal propagation speed was accompanied by the sudden decrease of the meridional topography gradient, which was the main reason for the slowdown of the propagation speed.
South China Sea (SCS) is a semi-enclosed marginal sea in the western Pacific Ocean, which has a broad shelf and slope in its northern part. The Dongsha Islands are located about 200 km offshore on a plateau over the continental slope.
The upper-layer circulation in SCS is forced by both the Asian monsoon and the Kuroshio intrusion through the Luzon Strait (Qu, 2000; Dong et al., 2010; Wang et al., 2011; Shu et al., 2011; Wang et al., 2013; Xu and Oey, 2014, 2015). In winter, the northeast monsoon prevails over the SCS and the Kuroshio intrusion is strong, and the reverse is true in summer (Wrykti, 1961; Qu, 2000; Fang et al., 2002; Yang et al., 2002; Gan et al., 2006; Zhang et al., 2015). Correspondingly, the upper-layer SCS circulation presents a cyclonic and anticyclonic pattern in winter and summer, respectively (Wyrtki, 1961; Qu, 2000; Zeng et al., 2009).
In the SCS, many eddies are observed from in situ observations (Xu et al., 1982; Chu et al., 1998; Li et al., 1998; Hu et al., 2011; Zhang et al., 2013, 2016) and satellite sea-level anomaly data (Li et al., 2002; Su et al., 2002; Yuan et al., 2007; Chow et al., 2008; Wang et al., 2008). Wang et al. (2003) counted the SCS mesoscale eddies using a merged satellite altimeter data set and found that eddies can be mainly grouped into four geographic zones in the SCS. Xiu et al. (2010) used model output and found that about 32.9±2.4 eddies are predicted each year in the SCS and the radii of these eddies range from about 46.5 to 223.5 km, with a mean value of 87.4 km. Zhuang et al. (2010) analyzed the strong intraseasonal variability (associated with mesoscale eddy energy) regions. These are on the northern continental slope, in the Gulf of Thailand, and along two bands in the deep basin, with the northern band located west of the Luzon Strait and the southern one southeast of Vietnam.
The causes of mesoscale eddies may be revealed to be the bathymetry anomaly (Sangra et al., 2005) or shedding from strong currents such as the Gulf Stream and the Kuroshio (Hurlburt and Thompson, 1980; Li et al., 1998; Hetland et al., 1999; Zu et al., 2013). When the Kuroshio flows over the Luzon Strait, a meander sometimes forms and the mesoscale eddies may be shed from the unstable meander and propagate westward into the SCS (Li et al., 1998; Wang et al., 2000; Yang et al., 2000; Li et al., 2002; Su et al., 2002; Jia and Liu, 2004; Yuan et al., 2006; Zhang et al., 2013). The mesoscale eddy shedding event can occur all year round (Yuan, 2006).
Wang et al. (2015) pointed out that there are mainly two paths of the mesoscale eddy's westward propagation: one is the along the North South China Sea (NSCS) continental slope, and the other is across the mid-basin from the Philippines moving westward. Along the NSCS continental slope, both the topography and the slope current contribute to the mesoscale eddy propagation (Chow et al., 2008; Wang et al., 2008; Zu et al., 2013). Much of the research has discussed the interaction between the mesoscale eddy and the topography based on ideal numerical experiments (Smith and O'Brien, 1983; Jacob et al., 2002; Sutyrin et al., 2003; Wei and Wang, 2009). The Dongsha Islands protrude over the NSCS continental slope toward the deep sea, and mesoscale eddy frequently occurs east of the Dongsha Islands (Wang et al., 2003; Chen et al., 2011). The effects of the Dongsha Islands on the mesoscale eddy propagation have not been discussed in detail prior to this which is the focus of this article.
Data and methods
The merged sea surface height anomalies (SSHA) are from TOPEX/Poseidon, Jason 1, and ERS (European Research Satellite), provided by the French Archiving, Validation Interpretation of Satellite Oceanographic Data (AVISO) project. The data are plotted on a grid with 1/3 resolution and averaged over 7 days (weekly).
The ocean surface current data is from the Ocean Surface Currents Analyses Real-time (OSCAR) (Bonjean and Lagerloef, 2002) project, which is an endeavor to calculate ocean surface velocities from satellite fields. The dataset contains near-surface ocean current estimates, derived using quasilinear and steady flow momentum equations. The horizontal velocity is estimated from SSH, surface wind, and sea surface temperature; these data were collected from various satellites and in situ instruments. The model formulation combines geostrophic, Ekman, and Stommel shear dynamics, as well as a complementary term from the surface buoyancy gradient. Surface currents are provided on a global 1/3 grid with a 5-day resolution dating from 1992 to present. Compared with the satellite sea surface geostrophic current, the OSCAR data contains more dynamic processes and more closely represents the real ocean current.
ETOPO2 (Marks and Smith, 2006) is used to describe the bathymetry in the Northern South China Sea.
In this article, an eddy detection scheme based on velocity geometry (Nencioli et al., 2010) is used. The scheme has been verified (Nencioli et al. 2010) and successfully applied to statistics of the mesoscale eddy properties (Dong et al., 2012; Liu et al., 2012). Compared with the Okubo–Weiss method and the winding angle method, the vector geometry method can get a higher success and a lower excess of detection rate (Nencioli et al., 2010).
Influence of Dongsha Islands on the mesoscale eddy properties
Using the eddy detection scheme (Nencioli et al., 2010), we detected the mesoscale eddy in the SCS from 1992 to 2013 based on the SSHA data set. All the mesoscale eddy propagation speeds are averaged in a 0.5 × 0.5 gird, and shown in Figure 1a. In order to focus on the NSCS continental slope, the grids with mean depths of less than 200 m or greater than 2500 m have been removed. In Figure 1, the shading indicates the propagation speed, and the unit arrow indicates the propagation direction. It can be found that the mesoscale eddy propagates along the shelf to a large extent. The main forcing factors that control the mesoscale eddy propagation are the planetary β effect (i.e., planetary vorticity changes with latitudes), topographical β effect, and the background large scale currents (Smith and O'Brien, 1983; Wang et al., 2008). Based on numerous SCS observational data, Yuan et al. (2008) have pointed out that the topographical β effect is superior to the β effect in the SCS. Follows the Smith and Reid (1982), the ratio of topographic and planetary betas is estimated by. Hence, the topographic betas effect is far larger than the planetary betas. Figure 1b shows the background currents calculated from the OSCAR dataset. The direction of the mesoscale eddy is similar to the background currents. However, the biggest difference is around 118°E, where the shelf touches the Dongsha Islands and turns toward the southwest. The background currents exhibit an obvious climbing movement towards the northwest around 118°E. The mesoscale eddy follows the topography and turns to the southwest to a large extent. This difference indicates that the topographical β effect is more significant than the background currents around 118°E.
The most interesting thing is that the mesoscale eddy propagation speed is suddenly slowed down when touching east of the Dongsha Islands from the eastern shelf. For convenience, a black dashed line has been utilized to separate the shelf into two domains southwest from where the speed becomes suddenly smaller. Along the isobaths, we selected 7 stations (S1–S7) to show the propagation speed variations. Compared to the mesoscale eddy propagation speed, there is no such sudden slowdown phenomenon of the background current speeds, and the background current speed is much smaller than the mesoscale eddy propagation speed. This confirms that the topographical β effect is more significant than the background currents around the 118°E.
Figure 2 shows the climatological mesoscale eddy radius. It can be found that both the cyclone and anticyclone mesoscale eddy radii become larger when crossing the black dashed line from the east. Based on numerical experiments and theoretical study, a number of studies (Flierl, 1979; McWilliams and Flierl, 1979; Smith and Reid, 1982) of the free isolated mesoscale eddy point out that when the mesoscale eddy radius increases, the meridional speed will decrease and the zonal speed will enlarge (the zonal speed will decrease when the radius is larger than the first baroclinic radius of deformation). Around the Dongsha Islands, the first baroclinic radius of deformation is about 25 km (Gan and Cai, 2001). At this point, both the meridional and zonal speed decrease with the enlargement of the mesoscale eddy radius. However, it cannot be distinguished whether the mesoscale eddy radius changes the propagation speed distribution or vice versa. We can hypothesize two different processes: First, the interaction between the topography and mesoscale eddy influences the eddy radius and then slows down the propagation speed. Second, the topography slows down the propagation speed and then the mesoscale eddy radius increases. Those hypotheses will not be discussed in this article. However, the rule found in the free isolated mesoscale eddy can also work with the influence of the topography.
Matching the mesoscale eddy centers and the associated SSHA, which indicates the mesoscale eddy strength, the climatologic SSHA at the mesoscale eddy center within a 0.5 × 0.5 grid is shown in Figure 3. It can be seen that both the cyclonic and anticyclonic mesoscale eddy strength decrease when crossing the black dashed line. When the mesoscale eddy touches the Dongsha Islands, the lateral and bottom friction caused by these Islands may be responsible for the decay of eddies.
From the above discussions, it can be seen that the mesoscale eddy propagation speed, radius, and strength will all be modulated by the Dongsha Islands. We will now focus on how the topography contributes to the propagation speed variation.
Topography effect on mesoscale eddy propagation
Term a enhances or decreases the long Rossby wave speed, depending on the meridional slope of the topography. Term b reflects the interaction between β, the meridional slope of topography, and the interface displacement. Term c adjusts the long Rossby wave speed for the topography. Term d in the meriondial speed analytic expression is equivalent to term a in the zonal speed expression. Term e in the meriondial speed analytic expression is equivalent to term b in the zonal speed expression. (Jacob et al., 2002).
From the typical scale given in Equations (3), it can be found that the leading scale of the propagation speed is 4 ms−1, which is much larger than the observation. The typical scale of the topography around the Dongsha Islands is 10−2, which is higher than the commonly value by two orders (Jacob et al., 2002). The complex topography, including small islands and an irregular small scale topography gradient, may block the propagation of the mesoscale eddy. However, Equations (4) and (5) also control the propagation of the mesoscale eddy.
Figure 4 shows the propagation speed and the topography gradient in the selected stations, marked in Figure 1. The zonal propagation speed is negative, which is related to the negative meridional topography (see Equation (4)), as is the meridional propagation speed. The consistency between the observed and analytic expressions confirms that the topography gradient controls the mesoscale eddy propagation. It can be found that the zonal propagation speed suddenly decreases when crossing the black dashed line from the east. However, the meridional propagation speed is not as obvious as the zonal speed. Comparison with the associated topography gradient indicates that there is a sudden decrease of the meridional topography gradient, which can be regarded as the main reason for the zonal propagation speed slow down. The root mean square of the propagation speed and the topography gradient also behave consistently. Therefore, it can be found that the slowdown of the propagation speed is mainly caused by the sudden decrease of the meridional topography gradient, which reduces the zonal propagation speed.
The continent shelf of the Northern South China Sea is one of the main mesoscale eddy propagation paths, and the propagation speed is closely related to the topography. Using the satellite altimetry data and the analytic expression for the speed of the center of an isolated eddy induced by the topography, (Jacob et al., 2002), some conclusions can be made.
Firstly, we found that the mesoscale eddy propagation speed is sudden slowed down when touching the east of the Dongsha Islands from the eastern shelf. Both the cyclonic and anticyclonic mesoscale eddy radii also suddenly increase when arriving at the Dongsha Islands. Under the restriction of the potential vorticity conservation, the strength of the mesoscale eddy also decreases.
Secondly, the associated background current is not the key factor that controls the slowdown of the propagation speed. Using the analytical expression for the eddy propagation speed confirms that the slowdown of the propagation speed is mainly caused by the sudden decrease of the meridional topography gradient, which reduces the zonal propagation speed.
This study is supported by the National Natural Science Foundation of China (No. 41406038, No. 41376026 and 41476014), and by State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology Chinese Academy of Sciences (No. LTO1512) by the Strategic Priority Research Program of the Chinese Academy of Sciences (No. XDA11010304).