Continental Shelf Research 25 (2005) 2075–2083
Intra-seasonal variation in the velocity ﬁeld of the northeastern
South China Sea
, T.Y. Tangb
, S.F. Linc
Department of Earth Sciences, National Taiwan Normal University, Taipei, Taiwan, ROC
Institute of Oceanography, National Taiwan University, Taipei, Taiwan, ROC
Energy & Resources Laboratories, Industrial Technology Research Institute, Hsinchu, Taiwan, ROC Available online 26 September 2005
Two subsurface Acoustic Doppler Current Proﬁlers (ADCP) were deployed in the northeastern South China Sea to study the circulation structure in the area as well as the path and process of the Kuroshio intrusion. The 48 h low-pass ﬁltered data, for the ﬁrst time, reveal signiﬁcant intra-seasonal variations in the velocity ﬁeld. Flow is alternately cyclonic or anticyclonic even within a single month. Local wind forcing dominated by monsoon winds fails to address the phenomena. The present study suggests that intruding current patterns are likely triggered by the strong wind stress curls. Strong negative wind stress curls off the southern tip of Taiwan introduce negative vorticity to form anticyclonic circulation in the intruding current. With diminishing wind stress curls, the intruding current weakens, forming a cyclonic circulation pattern.
r2005 Elsevier Ltd. All rights reserved.
Keywords: Intra-seasonal variation; South China Sea; Eddies; Wind stress curl
The circulation structure in the northern South China Sea is complicated because of the complex topography and forcing by seasonally reversing monsoons. In addition, this region is the conﬂuence of currents in the northern South China Sea, the Taiwan Strait and the Kuroshio. The Kuroshio is the strongest current in the vicinity of the Luzon Strait.
The Paciﬁc western boundary current, the Kuroshio, ﬂows northward east of Luzon and
Taiwan (Nitani, 1972). Similar to the Loop Current in the Gulf of Mexico, the Kuroshio water has been reported to intrude into the Luzon Strait, a deep gap in the western boundary (Nitani, 1972; Shaw, 1991). Since the Luzon Strait is the only deep passage of the South China Sea, the intrusion of water from the Kuroshio is important to the salt budget in the basin. Seasonal variations of Kuroshio intrusion based on water mass distribu-tion have been reported extensively in a wealth of the existing literature (e.g. Wyrtki, 1961; Nitani, 1972;Shaw, 1991). Although the intrusion of waters from the Kuroshio to the northeastern South China Sea is known, the circulation pattern associated with the intrusion is unclear. Wang and Chern (1987) suggested that intrusion is in the form of
0278-4343/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.csr.2005.03.005
Corresponding author. Tel.: 886 2 2934 7120ex.16; fax: 886 2 2933 3315.
an anticyclonic (clockwise) eddy at the onset of the northeast monsoon.Shaw (1989)suggested that the intruding water is passively advected by the basin-wide cyclonic circulation in the northern South China Sea in winter. Recent moored current-meter and ship-board ADCP data showed that westward ﬂow persisted in the region southwest of Taiwan year round (Liang et al., 2003).
In this study, data from two mooring stations in the region were examined to study the circulation pattern. The data consist of ﬂow velocity and water properties. Thus, the effect of the Kuroshio intru-sion can be inferred. The blended QSCAT/NCEP (NASA Quick Scatterometer/National Centers for Environmental Prediction) wind stress ﬁelds at a resolution of 0.51 0.51 (Milliff et al., 1999) were used to infer the driving mechanisms.
2. Mooring data
Two subsurface moorings (Wand E) were deployed in continental margin southwest of Taiwan from October 2000 to April 2001. Fig. 1
shows the mooring locations and the surrounding
bathymetry. Each mooring includes an upward-looking, 150 kHz, self-contained ADCP mounted on a 4500diameter spherical syntactic foam buoy and
a SEACAT conductivity–temperature–depth (CTD) mounted 5 m beneath the ADCP. The water depths at Wand E are 1014 and 968 m, respectively.
Table 1lists the locations and local water depths for each mooring, the depths of instruments, the duration of deployments, and the vertical range of current proﬁle measured by the ADCP. The bin length of ADCP was 8 m. The hourly current velocity was averaged over 240 pings for broad-band ADCP. The standard deviations of velocity were 1.2 cm s1. Errors caused by vertical excursion of the instrument and sound speed variation in the ADCP data were corrected by the tempera-ture and depth obtained by CTD. The data were also adjusted for local magnetic deviation. The velocity in the vertical proﬁle was then linearly interpolated and resampled at 10 m intervals. The time series of current velocity discussed in the following sections were low-pass ﬁltered to remove the ﬂuctuations for frequencies higher than 0.5 cycles per day.
Fig. 1. The study area with locations of mooring stations (squares) and Tung-Chi Island (asterisk). Current distribution at 30 m is adopted fromLiang et al. (2003).
3. Results and discussion 3.1. Low-pass filtered velocity
Fig. 2 shows the 48-h low-pass ﬁltered eastward (U) and northward (V) velocity components at W. In general, the ﬂow is barotropic in the upper 120 m with a slight decrease in magnitude. Both U and V show signiﬁcant intra-seasonal variations. In No-vember 2000, U is weak and alternates directions between west and east. Flow strengthens and turns eastward in December, reaching its maximum strength at the end of December. At the beginning of January, the eastward current suddenly reverses the direction toward west. The westward persists until mid-March. Another burst of strong eastward ﬂow appears afterward and reaches its maximum strength at the end of the month. The V component
velocity shows a similar pattern of bursts of northward ﬂow at the beginning of December 2000 and March 2001, but its phase leads U by about half a month.
Similar toFig. 2,Fig. 3shows U and V of low-pass ﬁltered velocity at location E. Signiﬁcant intra-seasonal variations are also evidenced at E. Flow is generally toward southeast except in January and after mid-March. Strong eastward ﬂow appears in mid-October 2000, late November 2000 and late February 2001 when current speeds reach more than 50 cm s1. Two periods of weak westward currents are present in January and middle March 2001. The strong eastward ﬂow is accompanied by strong southward ﬂow in V as shown in the bottom panel. The stick plots of depth averaged velocity at W (20–130 m) and E (20–180 m) are shown in Fig. 4. Bursts of northeastward ﬂow at Wand those of
Fig. 2. Vertical sections of 48-hr low-pass current velocity observed at W. Contour interval is 10 cm s1.
Mooring locations; water, ADCP and CTD depths; durations; and data distribution depths (interval of 10 m) at Wand E, respectively Station Longitude Latitude Water
depth (m) ADCP depth (m) CTD depth (m) Duration Measurement range (m) W1181 440E 221N 1014 160 165 2000/10/13–2001/04/18 20–130 E 1201 100E 221N 968 202 207 2000/10/04–2001/04/18 20–180
southeastward ﬂow at E seem to correlate well with changes at E lead those at Waround 10–15 days. The phenomenon deserves to further verify and will be examined in detail later in the next section.
In Fig. 4, northward ﬂow prevails at Wand southeastward ﬂow exists at E prior to the middle of December 2000. The feature infers that this region is
dominated by a conceptual clockwise circulation pattern. Currents at Wreverse southwestward and meanwhile currents at E turn either northwestward or northeastward during the period from middle December to middle February. The reversal phe-nomena suggest that current pattern alters from clockwise circulation to counterclockwise in the
study region. Clockwise circulation dominates again during the period from middle February to middle March. Note that the currents are always stronger during clockwise circulation. After March, weak counterclockwise circulation plays the ﬁnale. Furthermore, as mentioned in the introduction, currents perform an anticyclone or a cyclone might directly inﬂuence the geographic distribution of the water masses in the region. Without a correct description of circulation pattern, it is not possible to identify the water masses and to verify the intrusion process.
3.2. Water masses and circulation pattern
To describe the distribution of various water masses and the path of Kuroshio intrusion, the mooring CTD data at both Wand E were plotted in
Fig. 5. Also they join with velocity vectors to separate different episodes of clockwise and coun-terclockwise circulation. In the ﬁgure, the red color represents circulation pattern is clockwise while the blue represents there is counterclockwise circula-tion. Temperature and salinity of Wand E were measured at depth of around 160 and 205 m, respectively. Although the absolute values of CTD data are probably problematic, the relative value can still be applied to identify where the water masses come from. Since not only temperature but salinity of E are generally lower than those of W, the measured depth at E is deeper could not account
for the phenomena singly. Rather, there must be colder and fresher water entering and mixing with the local water around regions of E. Based on long-term mooring observations around sea southwest of Taiwan, mean current always ﬂows southeastward during winter (Chern, 1982). With mixing cold and fresh coastal water from Taiwan Strait, it makes sense that temperature and salinity at E should be lower than those at W.
Furthermore, concerning the temporal variability in each station, both stations show that temperature is higher whenever clockwise circulation dominates in the region. The temperature contrast is more distinct at Wthan that at E; the latter location is affected by cold coastal water (Fig. 5). The higher values of temperature were presented during two periods in December 2000 and in March 2001. The results are reasonable and can be explained by water movement. Currents generally intensify while clock-wise circulation performs as mentioned earlier so that stronger intrusion current brings in warmer water from Kuroshio front and then increases water temperature at W. On the other hand, during counterclockwise circulation episode (e.g. in the middle of January), the current is weak and ﬂows westward continuously along its northern boundary in Taiwan Strait. By gradually mixing with cold front from Taiwan Strait, the water temperature decreases.
To further study alternation of circulation pattern and relation between Wand E, we rotated the
current velocity to its principal axis. The principal axis of Wis 451 clockwise-rotated Cartesian (north-east–southwest) and E is 1351 clockwise-rotated Cartesian (southeast–northwest). Figs. 6a and 6b
show time series of principal axis depth-average current velocity at Wand E, respectively. Graphi-cally, both Wand E evidence two signiﬁcant peaks. However, the timing is not consistent with each other. Two peaks at Ware during the periods December and March while at E are from middle November to middle December and from middle February to middle March. The results demonstrate that E leads Wand an interval of 315 h might properly describe time lag of W. InFig. 6c, E with a time lag of 315 h is overlapped to W. Peak to peak comparison indicates that the variations of Wand those in time lag of E are frequently in phase, suggesting E leads Wby 10–15 d indeed. The corresponding advection velocity is around 0.12–0.18 m s1 which is the same order of the propagation speed of a baroclinic Rossby wave in the area (0.1 m s1). However, the region is not only with complex topography but also bounded in the east by the Taiwan Island. The limited observa-tions cannot describe how a baroclinc Rossby wave propagates in the study region. More observations or numerical modeling efforts are needed for further investigation. The trend that E leads Wby 10–15 d is always valid whether it is during episode of
clockwise or counterclockwise circulation. The fact that E leads Wduring clockwise episode is probably against intuitive. The two stations might not be located in the same anticyclonic structure.
What causes the intra-seasonal variation in the area deserves further investigation. Intuitively, currents in this area, especially those near the surface, should be driven by the strong monsoon during winter when northeasterly winds prevail. However, there is almost no correlation between the current and local wind forcing from weather station at Tung-Chi Island (ﬁgure not shown). On the other hand, the relation of the wind stress curl and the Kuroshio intrusion at the South China Sea has been investigated byMetzger and Hurlburt (1996, 2001). Also, based on model simulations in the South China Sea, Wu et al. (1998) suggested that the formation of a gyre, whether anticyclonic or cyclonic, is determined by the wind stress curl.
Shaw et al. (1999) also found that the ﬁrst two empirical orthogonal functions (EOF) modes of the wind stress curl agree well with those in the corresponding altimeter sea level modes, further supporting the wind stress curl forcing scenario. However, it is not clear whether the wind stress curl is important to meso-scale circulation as in the
Fig. 5. T–S diagram at Wand E. Velocity vectors are also presented to separate different episodes of clockwise (red) and counterclockwise (blue) circulation.
present study. Since the scale is quite small in the present study, high-resolution winds are needed. We adopted 6-hourly maps of 10 m zonal and meridio-nal wind components at a resolution of 0.51 0.51, which is derived from a space and time blend of QSCAT-DIRTH satellite scatterometer observa-tions and NCEP analyses (Milliff et al., 1999). The blended data set is one of the most up-to-date high-resolution data of ocean surface winds in the present time.
In this section, wind stress curls are calculated from the blended wind stress ﬁelds.Fig. 7shows the monthly mean of the wind stress curl during the 6-month period. The days in February and March 2001 have been slightly adjusted to ﬁt the intra-seasonal variation in the velocity ﬁeld. Main features in the intra-seasonal variation of the wind stress curl are captured in the sequence of plots from (a)–(f). In October 2000 (Fig. 7a), a dipole structure with positive values to the north appeared in the
region. Note that the south portion is what we should be more concerned about since its position approaches the Kuroshio intrusion. Off the south-ern tip of Taiwan, the large and negative curl extends southwestward in October that marks the onset of the winter monsoon. The pattern persists well into November, gradually increasing its strength in December. In the following month, the southern pattern weakens (Fig. 7d). The curl increases again and reaches its maximum strength during the period from February 10 to March 10 (Fig. 7e) and decays afterwards (Fig. 7f).
The variations of the southern pattern are corresponding to those shown in the velocity ﬁeld and the curl is capable of explaining the intra-seasonal variations in the velocity ﬁeld. While the southern pattern appeared and developed in the position, the intruded current from Kuroshio would form anticyclonic eddies clockwise because the curl provides negative vorticities to the current. Note
Fig. 6. Time series of principal axis depth-average current velocity at Wand E, respectively. E with time lag of 315 h is overlapped to W shown in the bottom panel as well.
that the strongest southern pattern is conﬁned and limited near E. Therefore, a tight and stronger anticyclone might occur in close vicinity of E merely. The loose clockwise vortexes might shed away from E while the corresponding Reynolds number is large enough. The large value of curl also explains part of the reason why the current is accelerated during clockwise circulation dominated
episode as mentioned earlier. For example, the strongest southward current at E with maximum speed of 100 cm s1shown on March 1, 2001 is well corresponding to the strongest southern pattern in
Fig. 7e. Although the wind stress curl is not energetic enough to give rise to a 100 cm s1current, the well consistency between the wind stress curls and the observational currents still imply that the
Fig. 7. Monthly mean of the wind stress curl during the six-month period. Contour interval is 0.1 107N m3. Regions of negative curls
curls might be one of the major factors to trigger the currents. As shown in Centurioni et al. (2004)that near surface velocities are well below the observed drifter speeds, the strong currents would be the baroclinic currents associated with the density gradients within the Kuroshio. On the other hand, while the southern pattern weakens or loosens (Figs. 7d, 7f), the intruded current overwhelms the curl and continues westward along the shelf break, forming a weak cyclonic (counterclockwise) eddy. Weaker currents in these two counterclockwise episodes are also noted in the previous section. 4. Summary and conclusions
Although the East Asian monsoons dominate, an intra-seasonal instead of seasonal variation in the velocity ﬁeld has been found in the region. The well consistency between wind stress curls and observed velocity ﬁelds suggests that the curls could be one of the major factors to trigger the intra-seasonal variations. The intensity of wind stress curl off southern tip of Taiwan might determine whether clockwise or counterclockwise circulation pattern performs in the region. While the southern pattern enhanced, the intruded current strengthens by getting large vorticity from the curl and forms an anticyclonic eddy afterwards. The mooring velocity data indeed evidence that currents during clockwise circulation are much stronger than those during counterclockwise. Temperature is also higher while it is clockwise since the stronger intrusion current will bring in Kuroshio warmer water and increase local water temperature.
Gracious thanks are extended to Drs. Y.J. Yang and W.–D. Liang at Chinese Naval Academy for useful discussions and Mr. Y. C. Tsai for processing the mooring data. The invaluable comments given by two reviewers greatly improved both the presentation and discussion. This research was
supported by the National Science Council, Taiwan, ROC, under Grants NSC 92-2611-M-003-004 (CRW) and NSC 89-2611-M-002-031-OP2 (TYT).
Centurioni, L.R., Niiler, P.P., Lee, D.–K., 2004. Observations of inﬂow of Philippine Sea surface water into the South China Sea through the Luzon Strait. Journal of Physical Oceano-graphy 34, 113–121.
Chern, C.S., 1982. Current and wave measurement in the vicinity of Hsien-Da-Kang and Tung- Kang (continued). Institution of Oceanography, National Taiwan University, Taiwan (Special publication 39).
Liang, W.-D., Tang, T.Y., Yang, Y.J., Ko, M.T., Chuang, W.-S., 2003. Upper-ocean current around Taiwan. Deep-Sea Re-search II 50, 1085–1105.
Metzger, E.J., Hurlburt, H.E., 1996. Coupled dynamics of the South China Sea, the Sulu Sea, and the Paciﬁc Ocean. Journal of Geophysical Research 101, 12331–12352.
Metzger, E.J., Hurlburt, H.E., 2001. The nondeterministic nature of Kuroshio penetration and eddy shedding in the South China Sea. Journal of Physical Oceanography 31, 1712–1732. Milliff, R.F., Large, W.G., Morzel, J., Danabasoglu, G., Chin, T.M., 1999. Ocean general circulation model sensitivity to forcing from scatterometer winds. Journal of Geophysical Research 104, 11337–11358.
Nitani, H., Beginning of the Kuroshio, In: Kuroshio, 1972. University of Washington Press, Seattle, Washington. pp. 129–163.
Shaw, P.-T., 1989. The intrusion of water masses into the sea southwest of Taiwan. Journal of Geophysical Research 94, 18213–18226.
Shaw, P.–T., 1991. The seasonal variation of the intrusion of the Philippine Sea water into the South China Sea. Journal of Geophysical Research 96, 821–827.
Shaw, P.-T., Chao, S.-Y., Fu, L.-L., 1999. Sea surface height variations in the South China Sea from satellite altimetry. Oceanologica Acta 22, 1–17.
Wang, J., Chern, C.S., 1987. The warm-core eddy in the northern South China Sea, Preliminary observations on the warm-core eddy. Acta Oceanographica Taiwanica 18, 92–103.
Wu, C.-R., Shaw, P.-T., Chao, S.-Y., 1998. Seasonal and interannual variations in the velocity ﬁeld of the South China Sea. Journal of Oceanography 54, 361–372.
Wyrtki, K., 1961. Physical oceanography of the southeast Asian waters, Scientiﬁc Results of Marine Investigation of the South China Sea and the Gulf of Thailand, NAGA report, Vol. 2 Scripps Institution of Oceanography, La Jolla, CA, p. 195