a pressure&driven south china sea warm...

15
A pressure-driven South China Sea Warm Current Y. Hsueh Department of Oceanography, Florida State University, Tallahassee, Florida, USA Liejun Zhong Horn Point Laboratory, University of Maryland Center for Environmental Science, Cambridge, Maryland, USA Received 9 March 2004; revised 26 May 2004; accepted 25 June 2004; published 29 September 2004. [1] Imposed pressure along the continental shelf break is shown to generate, in accordance with the arrested topographic wave (ATW) theory, a steady state flow that resembles the South China Sea Warm Current (SCSWC), flowing northeastward year- round along the outer continental shelf off the southern coast of China. The requisite shelf- break pressure distribution arises from the collision with the continental slope near the Dongsha Islands at about 116E and 21N, of the Kuroshio that has intruded into the South China Sea through the southern part of the Luzon Strait. The flow deflection following the collision creates the South China Sea Branch of Kuroshio (SCSBK) as a splinter current to the southwest, while the mainstream of the Kuroshio veers to the northeast and eventually exits the South China Sea through the northern part of the Luzon Strait. A channel flow model driven by an inflow-outflow condition at one end and closed at the other suggests the SCSBK may be feeding the SCSWC all along the shelf break through a weak onshore flow driven by the gradual drop in pressure in the SCSBK due to bottom friction. The SCSWC and SCSBK flow combination is also found in a Bryan-Cox model of the northern South China Sea. The SCSWC in the Bryan-Cox model thrives in summer and survives only near the shelf break during the winter when the northeast monsoon wind dominates. INDEX TERMS: 4512 Oceanography: Physical: Currents; 4576 Oceanography: Physical: Western boundary currents; KEYWORDS: ocean currents, continental shelf currents, pressure-driven flows Citation: Hsueh, Y., and L. Zhong (2004), A pressure-driven South China Sea Warm Current, J. Geophys. Res., 109, C09014, doi:10.1029/2004JC002374. 1. Introduction [2] The South China Sea, extending from the equator to 23°N and from 99°E to 120°E, is a semienclosed marginal sea in the western Pacific Ocean (see Figure 1). It is connected with the Pacific Ocean via the Luzon Strait (sill depth, 2500 m) through which the Kuroshio often intrudes [Wyrtki, 1961; Nitani, 1972; Shaw, 1989, 1991; Su et al., 1990; Qu et al., 2000]. The intruded Kuroshio moves westward and often reaches the continental slope, in the neighborhood of the Dongsha Islands, a few hundred kilo- meters west of the Luzon Strait. (Liang et al. [2003] show a vertical section of currents along 21°N from a 10-year composite of Shipboard Acoustic Doppler Current Profiler measurements that seems to indicate that the intrusion is limited to east of 119°E; see later in conclusions and discussion.) The contact with the continental slope causes the Kuroshio to bifurcate. The mainstream turns right to move eastward and leaves the South China Sea through the northern portion of the Luzon Strait. It forms, with the intruding Kuroshio in the southern portion of the Luzon Strait, an anticyclonic current loop [Nitani, 1972; Su et al., 1990]. The splinter current turns left and moves along the upper continental slope to the southwest. [3] The southwestward splinter current has been found through both direct velocity measurement and hydrographic data, and is named the South China Sea Branch of Kuroshio (SCSBK) due to its Kuroshio origin [Guan, 1985; Qiu et al., 1984; Guo et al., 1985]. (More recently, Qu [2000] noted from dynamic height differences a Kuroshio intrusion current along the continental slope south of China.) Just shoreward of the SCSBK is the South China Sea Warm Current (SCSWC) in waters 200–400 m deep. This narrow warm current is directed to the northeast and is thought to extend from the area just southeast of Hainan Island (111°E, 19°N) to the southern end of the Taiwan Strait [Guan, 1985]. The SCSWC has been documented in almost all months of the year, even in winter when the winds are to the southwest, although it is not certain how persistent the warm current is due to the lack of observations of long duration [Guan, 1985; Zhong, 1990]. [4] Spatially, the SCSWC seems to consist of two distinct portions. The eastern portion, found off the Fujian coast east of the Dongsha Islands, flows steadily north- eastward; the western portion found off the Guangdong coast, west of the Dongsha Islands is seasonal in terms of the flow path, width, salinity and flow velocity [Guan, JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 109, C09014, doi:10.1029/2004JC002374, 2004 Copyright 2004 by the American Geophysical Union. 0148-0227/04/2004JC002374 C09014 1 of 15

Upload: danganh

Post on 24-Jun-2018

212 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: A pressure&driven South China Sea Warm Currentshoni2.princeton.edu/ftp/lyo/journals/HsuehZhong-SCSWC...break pressure distribution arises from the collision with the continental slope

A pressure-driven South China Sea Warm Current

Y. HsuehDepartment of Oceanography, Florida State University, Tallahassee, Florida, USA

Liejun ZhongHorn Point Laboratory, University of Maryland Center for Environmental Science, Cambridge, Maryland, USA

Received 9 March 2004; revised 26 May 2004; accepted 25 June 2004; published 29 September 2004.

[1] Imposed pressure along the continental shelf break is shown to generate, inaccordance with the arrested topographic wave (ATW) theory, a steady state flow thatresembles the South China Sea Warm Current (SCSWC), flowing northeastward year-round along the outer continental shelf off the southern coast of China. The requisite shelf-break pressure distribution arises from the collision with the continental slope near theDongsha Islands at about 116E and 21N, of the Kuroshio that has intruded into the SouthChina Sea through the southern part of the Luzon Strait. The flow deflection followingthe collision creates the South China Sea Branch of Kuroshio (SCSBK) as a splintercurrent to the southwest, while the mainstream of the Kuroshio veers to the northeast andeventually exits the South China Sea through the northern part of the Luzon Strait. Achannel flow model driven by an inflow-outflow condition at one end and closed atthe other suggests the SCSBK may be feeding the SCSWC all along the shelf breakthrough a weak onshore flow driven by the gradual drop in pressure in the SCSBK due tobottom friction. The SCSWC and SCSBK flow combination is also found in a Bryan-Coxmodel of the northern South China Sea. The SCSWC in the Bryan-Cox model thrives insummer and survives only near the shelf break during the winter when the northeastmonsoon wind dominates. INDEX TERMS: 4512 Oceanography: Physical: Currents; 4576

Oceanography: Physical: Western boundary currents; KEYWORDS: ocean currents, continental shelf currents,

pressure-driven flows

Citation: Hsueh, Y., and L. Zhong (2004), A pressure-driven South China Sea Warm Current, J. Geophys. Res., 109, C09014,

doi:10.1029/2004JC002374.

1. Introduction

[2] The South China Sea, extending from the equator to23�N and from 99�E to 120�E, is a semienclosed marginalsea in the western Pacific Ocean (see Figure 1). It isconnected with the Pacific Ocean via the Luzon Strait (silldepth, 2500 m) through which the Kuroshio often intrudes[Wyrtki, 1961; Nitani, 1972; Shaw, 1989, 1991; Su et al.,1990; Qu et al., 2000]. The intruded Kuroshio moveswestward and often reaches the continental slope, in theneighborhood of the Dongsha Islands, a few hundred kilo-meters west of the Luzon Strait. (Liang et al. [2003] show avertical section of currents along 21�N from a 10-yearcomposite of Shipboard Acoustic Doppler Current Profilermeasurements that seems to indicate that the intrusion islimited to east of 119�E; see later in conclusions anddiscussion.) The contact with the continental slope causesthe Kuroshio to bifurcate. The mainstream turns right tomove eastward and leaves the South China Sea through thenorthern portion of the Luzon Strait. It forms, with theintruding Kuroshio in the southern portion of the LuzonStrait, an anticyclonic current loop [Nitani, 1972; Su et al.,

1990]. The splinter current turns left and moves along theupper continental slope to the southwest.[3] The southwestward splinter current has been found

through both direct velocity measurement and hydrographicdata, and is named the South China Sea Branch of Kuroshio(SCSBK) due to its Kuroshio origin [Guan, 1985; Qiu et al.,1984; Guo et al., 1985]. (More recently, Qu [2000] notedfrom dynamic height differences a Kuroshio intrusioncurrent along the continental slope south of China.) Justshoreward of the SCSBK is the South China Sea WarmCurrent (SCSWC) in waters 200–400 m deep. This narrowwarm current is directed to the northeast and is thought toextend from the area just southeast of Hainan Island (111�E,19�N) to the southern end of the Taiwan Strait [Guan,1985]. The SCSWC has been documented in almost allmonths of the year, even in winter when the winds are to thesouthwest, although it is not certain how persistent thewarm current is due to the lack of observations of longduration [Guan, 1985; Zhong, 1990].[4] Spatially, the SCSWC seems to consist of two

distinct portions. The eastern portion, found off the Fujiancoast east of the Dongsha Islands, flows steadily north-eastward; the western portion found off the Guangdongcoast, west of the Dongsha Islands is seasonal in terms ofthe flow path, width, salinity and flow velocity [Guan,

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 109, C09014, doi:10.1029/2004JC002374, 2004

Copyright 2004 by the American Geophysical Union.0148-0227/04/2004JC002374

C09014 1 of 15

Page 2: A pressure&driven South China Sea Warm Currentshoni2.princeton.edu/ftp/lyo/journals/HsuehZhong-SCSWC...break pressure distribution arises from the collision with the continental slope

1985]. The eastern portion is stronger, wider and deeperthan the western one. In addition, the northeastward flowwas observed in deeper waters east of the DongshaIslands [Guo et al., 1985], and was also interpreted as apart of the SCSWC in some literature. (Xue et al. [2004]find in a numerical model an on-shelf part of this flowdiagnosable from cross-shelf pressure gradients.) In viewof the existence of the anticyclonic Kuroshio loop in thisregion, this deep current is probably some combination ofthe SCSWC and the return part of the anticyclonic loop.In the present work, SCSWC will refer only to thewestern portion.[5] Of particular significance in the western portion is the

existence of a tongue-like patch of high-pressure (and high-temperature) found between the SCSWC and the SCSBK,extending southwestward with decreasing strength from theDongsha Islands area [Zhong, 1990]. Associated with thistongue of high dynamic pressure is temperature as much as4�C higher than that of the surrounding water [Zhong, 1990](plots of the high temperature patch are reproduced in thework of Zhong [2001]). These facts suggest that the SCSBKmight be feeding the SCSWC along the high-pressuretongue, tapering in strength southwestward. The transportof the two currents also seems to support this hypothesis.The transports of SCSWC and SCSBK estimated by Guan[1985] and Zhong [1990] all increase from west to east.Following the flow, the transport of the SCSBK thusdecreases and that of the SCSWC increases. A dynamicallysimilar situation apparently exists off the west Florida shelf,where the contact of the Gulf of Mexico Loop Current withthe shelf gives rise to a high-pressure tongue along the shelfbreak, supporting a combination of a southward shelf-breakjet and a northward splinter current just offshore [Hetland etal., 2001].

[6] Although the existence of the SCSWC has beenknown for over thirty years, there is not yet a commonlyaccepted interpretation. Zeng et al. [1989] and Li et al.[1993] use a 1=2� barotropic model to simulate the circula-tion of the South China Sea. They propose that, in winter, astrong along shelf sea surface slope is the main force todrive the northeastward SCSWC, the along shelf slopebeing maintained by water pile-up east of Hainan Islandunder the northeast monsoon. They acknowledge that theopening up of the Luzon Strait and allowing the Kuroshio tointrude greatly enhances the SCSWC formation year-round.Chao et al. [1995] suggest that it is the relaxation of thenortheast monsoon winds that gives rise to the SCSWC inwinter, as the piled-up water off Hainan Island surges back.While the pile-up and its relaxation appear to contribute tothe SCSWC formation, they do not lead to the concentrationof the current over the shelf/slope junction as observed[Guan, 1985; Chao et al., 1995].[7] In this work, an alternate explanation of the formation

of the SCSWC is proposed, namely that of an imposed step-like alongshore pressure distribution over the shelf/slopejunction, giving rise to a steady state, barotropic (shelf-break) jet with the proper flow concentration. The requisitepressure distribution arises from the collision of the intrudedKuroshio with the continental slope that causes a flowbifurcation into two branches [Hsueh and Zhong, 2003].The pressure is low to the right and high to the left of thestagnation point, leading to a step-like pressure distribution.The friction dominated steady state flow on the continentalshelf in balance with such an imposed pressure distributionalong the shelf/slope junction is then solved for, leading to ashelf-break jet that resembles the SCSWC. The feeding ofthe SCSWC by the SCSBK in deep water is demonstrated ina channel flow model with a continental slope topographyusing the SCRUM (S coordinate Rutgers University Model)code. In addition, the output from an Asian marginal seasmodel that include the northern South China Sea [Hsueh,2000] will be presented to show snapshots in which theintrusion of the Kuroshio leads to the impingement and therequisite pressure distribution along the upper slope. It canbe seen that the model SCSWC is broad and strong insummer when wind is to the northeast, but weak and shelf-break confined in winter when wind blows to the southwest.[8] The paper is organized as follows. Section 2 gives the

shelf-break jet solution forced by the imposition of apressure along the offshore side that is similar to theexpected wall-pressure distribution resulting from the im-pingement of the Kuroshio upon the continental slope.Section 3 presents the channel flow solution showing thefeeding of SCSWC by the SCSBK. Section 4 presentsresults from the Asian marginal seas model. Conclusionsand discussion are found in section 5.

2. SCSWC as a Shelf-Break Jet

2.1. Pressure Along the Shelf Break

[9] Evidence of a step-like along shelf break pressuredistribution set up by the intruding Kuroshio in the northernSouth China Sea is found first in a numerical model of theAsian marginal seas [Hsueh, 2000]. The Kuroshio inflow-outflow driven model is based on the Bryan-Cox code [Cox,1984] and is run for an additional three years with the

Figure 1. Map of the South China Sea. Thin curves are the100 and 1000 m isobaths.

C09014 HSUEH AND ZHONG: SOUTH CHINA SEA WARM CURRENT

2 of 15

C09014

Page 3: A pressure&driven South China Sea Warm Currentshoni2.princeton.edu/ftp/lyo/journals/HsuehZhong-SCSWC...break pressure distribution arises from the collision with the continental slope

inclusion of monthly mean wind stresses of Hellerman andRosenstein [1983] beyond the quasi-equilibrium statereported in the work of Hsueh [2000]. Figure 2 shows thesea surface elevation field superimposed with the barotropicvelocity field at Y6D283.50 (i.e., day 283.50 in the 7th yearof model integration). At this early winter (October) situa-tion, wind already blows to the southwest. (The monthlymean field (not shown) for October is similar.)[10] In the sea-surface elevation field a tongue-shaped

patch of high values extends roughly from where theKuroshio runs into the continental slope (116�E, 21�N) tothe southwest along the shelf break (just seaward of the100 m isobath represented by the heavy curve). Flanking thehigh-pressure tongue are the SCSBK in deep water andthe SCSWC in shallow water, consistent with observationsnoted before and confirming the possibility that the SCSBKis feeding the SCSWC. Similar surface pressure tongue-likefeature also exists in summer time when wind stress is to thenortheast (see later in Results from the Asian marginal seasmodel). The sea-surface elevation along the 100 m isobathfor Y4D000, just before the wind stress is applied,Y6D226.50, a time of wind stress to the northeast, andY6D283.50 is plotted in Figure 3. The common feature ofthe curves is that, facing the coast, the sea level rises sharplyfrom the right and levels off to the left of the stagnationpoint, irrespective of the wind stress applied, leading to astep-like variation in pressure.[11] Theoretical studies of flow bifurcation following the

incidence at a shelf-fronted coast of a baroclinic current inan inviscid, reduced-gravity model ocean have found thatvortex tube foreshortening due to the uplifting forced by thestep yields pressure at the coast low in the right branch

current and high in the left branch current, facing the coast[Hsueh and Zhong, 2003]. In a more realistic setting such asthat found in the Asian marginal seas model, the approachcurrent never penetrates much beyond the shelf break andthe pressure differential similar to that found at the coast inthe theoretical study is found instead along the shelf-slopejunction (Figure 2). The flow on the south China shelf,particularly the SCSWC, must thus be understood in part asa frictional flow in balance with the pressure along the shelfbreak imposed by the incidence of the Kuroshio.

2.2. ATW Solution

[12] It is well known that the barotropic flow on a slopedcontinental shelf obeys the so-called Arrested TopographicWave (ATW) equation [Csanady, 1978; Hetland et al.,1999]. In terms of pressure (sea surface height, z), theequation reads

@2z@x2

þ fhx

r

@z@y

¼ 0: ð1Þ

[13] Here the y axis coincides with the straight coast andthe x axis points offshore. Water depth h = h(x) varies onlyin x and hx represents the bottom slope. The bottomfrictional stress, only present in the y direction, is linearin y velocity with a coefficient of r, taking to be 0.1 cm/sec.The Coriolis parameter, f, is constant set to 10�4 sec�1.[14] Equation (1) is similar to the one-dimensional heat

equation with �y playing the role of time. Any pressureimposed at the shelf break (x = ‘) thus appears at succes-sively large �y as it diffuses onshore at a rate inverselyproportional to hx, the bottom slope. Thus across the shelf

Figure 2. Sea surface elevation superimposed with the barotropic velocity at Y6D283.50 from theAsian marginal seas model. The heavy arrow indicates the wind stress (0.8 N/m2) near the coast. The100 m isobath is represented by the heavy curve with solid squares marking the location where values ofsea level elevation are extracted for a later plot of along-isobath pressure distribution.

C09014 HSUEH AND ZHONG: SOUTH CHINA SEA WARM CURRENT

3 of 15

C09014

Page 4: A pressure&driven South China Sea Warm Currentshoni2.princeton.edu/ftp/lyo/journals/HsuehZhong-SCSWC...break pressure distribution arises from the collision with the continental slope

imposed with an offshore step-like pressure with the low tothe right (in the y direction), the pressure decreases onshore,leading to a geostrophic current in the y direction. Becausethe diffusivity is low over large hx, the cross-shelf pressuregradient, thus the alongshore current, is the greatest near theshelf break. For the south China shelf, the imposed shelf-break pressure is low to the northeast, and thus the shelfbreak jet is to the northeast, resembling the SCSWC.[15] Figure 4 shows the setup for calculating z from

equation (1) for the south China shelf (180 km � x � 0,�800 km� y� 0). The top panel shows the shelf depth, h =h(x), and the lower panel shows the distribution of thepressure imposed along the shelf break at x = ‘ = 180 km.The boundary conditions used in the calculation are thus

@z@x

þ fh0

r

@z@y

¼ fF

gr; x ¼ 0; ð2Þ

z ¼ P yð Þ x ¼ ‘ ð3Þ

z ¼ 0 y ¼ 0 ð4Þ

Here, F is the y direction wind stress, which is set to zero, h0the coastal depth, and P(y) represents the step-like imposedpressure at the shelf break. Equation (4) sets the inflow tozero across the backward section (in the sense of continentalshelf wave propagation) and assures that the solution is allforced by the imposed pressure.[16] Figure 5 shows the solution in plots in the horizontal

plane of sea-surface height z and velocity vectors. The

expected spread of pressure and the concentration of flowat the shelf break are clearly demonstrated. The strongestcurrent in the jet to the northeast appears around �200 km,just as the imposed pressure reaches the high, consistentwith the fact that in Figure 2 where the strongest SCSWC isfound just inshore of the stagnation point near the fourthmarker from the east on the 100 m isobath. The inclusion ofa wind stress in the y direction leaves largely intact the shelfbreak jet, but renders currents reversed near the coast (notshown).[17] Not represented in P(y) used in Figure 4 is the gentle

decline in sea-surface height to the left of the stagnationpoint as shown in Figure 3. Although the inclusion of thisdecrease in the imposed pressure does not change theformation of the shelf break jet (not shown), it raises thepossibility that the SCSWC is fed by offshore flows(SCSBK) through a weak cross shelf onshore transportproportional to the along shelf pressure decline. The feedingof a shelf break jet by offshore flows has been demonstratedin a barotropic channel flow setting [Hetland et al., 2001].Below, a similar attempt is made in a stratified model toexplain the connection of the SCSWC and the SCSBKacross the high-pressure tongue first noted in Figure 2.

3. SCSBK-SCSWC Connection

[18] To demonstrate that indeed a current like the SCSBKcan leak water to a companion current like the SCSWCacross the shelf break, an S coordinate Rutgers UniversityModel (SCRUM) of the flow in a zonal channel with across-channel topography similar to that of the continentalmargin off south China is constructed (see Hedstrom [1997]

Figure 3. Sea-surface elevation along the 100 m isobath calculated from the Asian marginal seas model.The markers on the curve mark the positions of the solid squares found in Figure 2.

C09014 HSUEH AND ZHONG: SOUTH CHINA SEA WARM CURRENT

4 of 15

C09014

Page 5: A pressure&driven South China Sea Warm Currentshoni2.princeton.edu/ftp/lyo/journals/HsuehZhong-SCSWC...break pressure distribution arises from the collision with the continental slope

Figure 4. (top) Depth for the south China shelf and (bottom) the distribution of pressure, P(y), imposedalong the shelf break at an offshore distance of 180 km.

Figure 5. Arrested topographic wave (ATW) solution for the south China shelf. The z-field incentimeters is to the left, and the velocity field to the right.

C09014 HSUEH AND ZHONG: SOUTH CHINA SEA WARM CURRENT

5 of 15

C09014

Page 6: A pressure&driven South China Sea Warm Currentshoni2.princeton.edu/ftp/lyo/journals/HsuehZhong-SCSWC...break pressure distribution arises from the collision with the continental slope

for the SCRUM code). Figure 6 shows the plane view of themodel channel with the z axis pointing upward from thesurface. The inflow (SCSBK) is imposed in the deep-waterhalf of the open boundary and the outflow (SCSWC) isimposed in the shallow-water half. The other end of thechannel is closed. Thus in the model, the SCSBK com-pletely turns around to become SCSWC, whereas in morerealistic models, the SCSBK encounters no end wall[Hsueh, 2000; Xue et al., 2004]. The point of the channelmodel is however that the SCSBK leaks to the SCSWC allalong the shelf break before any end wall is reached. So, thepresence of the closed end does not detract from the mainpurpose of the model of showing the subtle dynamicconnection between the SCSBK and the SCSWC indepen-dent of the end wall.[19] Initially, the channel water is at rest and has a density

with the anomaly given by

r Tð Þ ¼ R0 þ Tcoef T � T0ð Þ; ð5Þ

where T = T(z) = 16 + 10 exp [z/100], z vertical coordinate inmeters,R0 = 31.3795 kg/m

2,Tcoef=� 2.8 � 10�1 kg/m3/C, andT0 = 0.[20] The initial temperature profile is exponential in z

coordinate with the surface temperature at 26�C and reach-ing a near constant of 16�C at about 600 m. There are 10 Slevels. The horizontal resolution is 10 km.[21] The S coordinate is terrain-following and has a

geometric advantage over the z coordinate in representingsteep continental margin topography. However, it doesintroduce quantitative (hopefully, not qualitative) errors inpressure gradients [Chu and Fan, 2003]. (Doubling thenumber of S levels leads to a reduction in pressure gradients(see later in Sea surface height and velocity across thechannel).) The physical quantities in the model include thethree-dimensional velocity, sea surface height, density, andtemperature. The equation of state is chosen to be linearsince temperature is the only active tracer in the experiment.The initial temperature field only depends on the depth, sodoes the initial density.[22] Tracer fluxes are set to zero at both the sea surface

and bottom. No wind stress is applied. At the seafloor, alinear drag law is specified with a coefficient of 1 cm s�1.At the solid lateral boundary tracer flux vanishes and no-slipcondition is used for velocities. Radiation boundary condi-tion [Orlanski, 1976] is applied for tracer and velocity at the

eastern open boundary. Gradient condition is implementedfor sea surface heights. The normal component of barotropicvelocity is calculated from the stream function along theopen boundary and the tangential component is set to zero.The stream function distribution along the boundary is fixedin time and shown in Figure 7. The total inflow velocity isthe sum of the barotropic and baroclinic parts, the lattercalculated from the thermal wind relation.[23] Horizontal mixing of momentum and temperature is

accomplished through the use of a Laplacian operator. Thehorizontal mixing coefficient for momentum is 1 103 m2/sand that for the temperature is 5 102 m2/s. Vertical mixingand viscosity are represented with the coefficient values of1 10�4 m2/s, 1 10�5 m2/s for momentum andtemperature, respectively.[24] The model starts from a state of rest, i.e., flat sea

surface and zero interior velocity. The model integration iscarried out to day 1000. The volume-averaged kineticenergy and potential energy (not shown) indicate that themodel has reached a near steady state. (The temperaturefield is still slowly evolving, due to the fact that the heatflux is not exactly in balance.) Following analyses are basedon the output in this near steady state.

3.1. Sea Surface Height and Velocity Across theChannel

[25] The contours of sea surface height are shown inFigure 8. A high-pressure tongue, like that found in theBryan-Cox model output (see Figure 2), extends over theshelf break in the direction of the inflow. The solid line inFigure 9 shows the sea surface height along the channel aty = 200 km, showing a gradual drop to the west at a ratethat is nearly constant.[26] The along-channel drop in sea-surface height toward

the western wall generates an across-channel (onshore)water movement. The onshore flow, particularly strong overthe shelf break, is shown in Figure 10 at two sections, onein the middle of the channel and the other 400 km awayfrom the eastern open boundary. The contour plots looksimilar, indicating the onshore movement is probablypresent all along the channel. The along channel flow isdominant (u velocity). The inflow (westward flow) ismostly in deep-water region and is much weaker than theeastward outflow over the shelf due to mass conservation.In contrast to the SCSWC, which is concentrated over theshelf break, the core of the model eastward flow shiftstoward the coast. This discrepancy does not detract from thefact that the eastward model flow, thus the SCSWC, is fedby water offshore through the onshore water movement.[27] This onshore movement across the shelf break is

apparently associated with the along-channel pressure drop,which is caused by bottom friction. To examine the role ofbottom friction, another identical SCRUM run is madeexcept with a much smaller linear drag coefficient. Thesea surface height along the channel from this run is plottedas dashed curve in Figure 9. The drop in the case of thesmaller friction coefficient is much more concentratedtoward the western wall, perhaps within a deformationradius, the length scale of an inviscid flow, approximately

R ¼ffiffiffiffiffigh

p

f�

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi10ms�2 102m

p

10�4s�1� 300 km:

Figure 6. Domain of a SCRUM channel. An inflow-outflow transport condition is imposed at the open east end.

C09014 HSUEH AND ZHONG: SOUTH CHINA SEA WARM CURRENT

6 of 15

C09014

Page 7: A pressure&driven South China Sea Warm Currentshoni2.princeton.edu/ftp/lyo/journals/HsuehZhong-SCSWC...break pressure distribution arises from the collision with the continental slope

When bottom friction is important, the length scale isgreatly increased. The equation for the barotropic, along-channel velocity is

@u

@tþ u

@u

@xþ v

@u

@y� f v ¼ �g

@z@x

� gub

D:

Here, g is the drag coefficient, ub the velocity at the bottomand D the water depth. In steady state, the balance ismaintained mainly between the pressure and bottom friction,

g@z@x

� gub

D:

Thus the length scale is

x � gzDgub

� 10m s�2 0:2m 100m

10�2 m s�1 10�2 m s�1� 2000 km:

This length scale is practically the length of the modelchannel. So the onshore movement occurs all along theshelf-break, not just near the end wall to the west, confirmingthe likeness of the two contour plots of the cross-channelvelocity component at two sections that are 800 km apart.Model run with 20 S levels produces a much milder declinein sea level heights and the difference between the two drag

Figure 7. (top) Cross-channel bathymetry and (bottom) imposed stream function at the east openboundary in the SCRUM run. P1–P6 mark the position in a section 400 km from the eastern end ofparticles to be followed for a study of the trajectories and vorticity.

Figure 8. Contours of sea-surface height in the near steady state in the SCRUM channel.

C09014 HSUEH AND ZHONG: SOUTH CHINA SEA WARM CURRENT

7 of 15

C09014

Page 8: A pressure&driven South China Sea Warm Currentshoni2.princeton.edu/ftp/lyo/journals/HsuehZhong-SCSWC...break pressure distribution arises from the collision with the continental slope

Figure 9. Sea surface height along the shelf break (y = 200 km) in the SCRUM channel with bottomdrag coefficients of 1 cm/s (solid curve) and 10�3 cm/s (dashed curve).

Figure 10. Contour plots of horizontal velocities in the SCRUM channel at two sections: (left) x =1200 km and (right) x = 2000 km.

C09014 HSUEH AND ZHONG: SOUTH CHINA SEA WARM CURRENT

8 of 15

C09014

Page 9: A pressure&driven South China Sea Warm Currentshoni2.princeton.edu/ftp/lyo/journals/HsuehZhong-SCSWC...break pressure distribution arises from the collision with the continental slope

coefficient cases is much reduced. Apparently, the highresolution in the vertical suppresses numerical friction sothat the total frictional effect is reduced. However, thequalitative picture of a gradual length-wise drop in sea levelheight enhanced by bottom friction remains unchanged.

3.2. Particle Trajectories and Vorticity

[28] The near steady state velocity and temperature fieldsat the last time step of the SCRUM run with the largefriction coefficient form the basis for obtaining trajectoriesand vorticity following particles P1, P2, P3, P4, P5, P6 in asection 400 km from the eastern end of the SCRUM channel(see Figure 7 for the initial particle positions in the section).The trajectories are obtained by integrating forward intime the three-dimensional velocity. The integration stopswhen the particle goes out of the domain, when it touchesthe bottom topography, or when it reaches the sea surface.At each point along the trajectory, potential vorticity andrelative vorticity are also calculated or interpolated.[29] There are a total of six particles followed. These

particles are arranged on a cross-channel section in a ‘‘T’’pattern. Four particles (PI, P2, P3 and P4) are all 200 mdeep but 20 km apart in cross channel direction fromshallow to deep water. Particles P5 and P6 align with P2vertically at depth of 300 m and 400 m, respectively. Thetrajectories are plotted in Figure 11.[30] Particles PI, P2, P3 and P4 move with the inflow and

veer from deep-water region to shallow region. All of themturn around across the shelf break and eventually leave withthe outflow through the eastern open boundary. The trajec-tory calculation of the other two, P5 and P6, is terminatedprematurely since they touch the bottom. As an example forthe range of movement, P3 initially moves westward in thedeep region following the inflow current; meanwhile, it

sinks. It is at its greatest depth as it reaches the shelf break.It crosses the shelf break and continues to move onshore toget into the shelf region where it is caught in the outflowand moves to the east.[31] All the particles move downward in the over the

slope but upward in the shelf region (not shown). Suchvertical movement is consistent with bottom Ekman layerpumping [Hsueh, 1968]. The vertical velocity is negative indeep water over the continental slope due to a divergence inbottom Ekman transport (seaward under the westwardinflow and shoreward under the eastward return flow). Overthe continental shelf, onshore flow (see Figure 10) over theshoaling sea bottom causes upwelling.[32] Both potential and relative vorticity are calculated

along the trajectories. Relative vorticity is @v/@x � @u/@y,and potential vorticity is defined as �r�1(@r/@z)(f + vx �uy), where f, the planetary vorticity, is a constant in themodel. Figure 12 shows potential and relative vorticity ofthe particles along their trajectories. Since the particles turnin a clockwise sense as they cross the shelf break, it isexpected that they will gain negative relative vorticity toconserve potential vorticity. This is apparently the caseaccording to Figure 12. For instance, P2 (positions markedby the cross, ‘‘x’’) has a relative vorticity value close to�1 10�6 s�1 initially. Its relative vorticity steadilydecreases to �6 10�6 s�1 along its trajectory, while itspotential vorticity remains within a relatively narrow rangeof 1.0 10�10 m�1 s�1 to 1.5 10�10 m�1 s�1.[33] Taking it all together, the SCRUM channel flow

suggests that the SCSWC appears fed by the onshoreleakage of the deep water (SCSBK), driven by a gradual,bottom friction induced drop in pressure all along the uppercontinental slope. The negative vorticity generated by thecrossing from deep to shallow water regions facilitates the

Figure 11. SCRUM channel particle trajectories. The legend gives the coordinates of the initialpositions.

C09014 HSUEH AND ZHONG: SOUTH CHINA SEA WARM CURRENT

9 of 15

C09014

Page 10: A pressure&driven South China Sea Warm Currentshoni2.princeton.edu/ftp/lyo/journals/HsuehZhong-SCSWC...break pressure distribution arises from the collision with the continental slope

joining into the SCSWC of the deep water, which alsosupplies the heat, making the northeast flowing currentwarmer than otherwise.

4. Results From the Asian Marginal Seas Model

[34] As mentioned earlier, the flow combination of theSCSWC and SCSBK is prevalent in the output of the Asianmarginal seas model that includes the Luzon Strait andnorthern South China Sea [Hsueh, 2000]. Briefly, the modelis an inflow-outflow experiment with the Kuroshio admittedthrough the southern boundary at 15�N and let out south ofKyushu at 140�E. The model domain is covered with a 1/6�grid with 30 vertical levels. The model bottom flattens outat 5000 m. The width of the pressure-driven SCSWC canbe estimated from equation (1) to be about 100 km with a1000 km alongshore scale. So, the model resolves theSCSWC. Figure 13 shows a snapshot of the impingementon the continental slope south of China of the Kuroshio justbefore the wind stress is applied at Y4D000 (i.e., the end ofthe 4th year of model integration). A tongue-shaped patchof high sea-surface elevation is seen, extending along theshelf break from the region of Kuroshio impingement 116Eand 21N. Offshore of the high-pressure tongue, the modelflow is to the southwest, representing the SCSBK. Inshore,the model flow is to the northeast and with velocity highnear the shelf break, as expected from the ATW theory.Figure 14 shows a similar snapshot at Y6D226.50, a typicalsummertime situation.[35] It appears that in the model the SCSWC and SCSBK

flow combination is always present across the high-pressuretongue along the upper continental slope. In summer

(Figure 14), the SCSWC is broad and strong. In like-wintersituation with a northeast wind (Figure 2), the SCSWC isconfined to the shelf break and considerably weaker. Inci-dentally, monthly mean fields do not differ much from thesesnapshots.[36] Figure 15 shows two model salinity sections along

the straight line in Figure 13, one at Y6D283.50 and theother at Y6D325.50 when the SCSWC is not present. It isclear that with the SCSWC present, the high salinity core(representing the SCSBK water) is moved up the slope andonto the continental shelf, indicating a substantial onshoreflow. The SCSWC appears to coincide with the patch ofsaline (and warm) water that has crept onto the shelf justinshore of the 100 m isobath.

5. Conclusions and Discussion

[37] The pressure distribution alongshore brought aboutby the impingement with the continental margin south ofChina of the Kuroshio that has intruded through the LuzonStrait appears to drive a shelf break jet that resembles theSCSWC. The shelf break jet arises as an ATW solution tothe flow on the continental shelf. As such, it is a steadystate. The intrusion of the Kuroshio is however a time-dependent process. Figure 16 shows a distance-time plot ofsea-surface height anomaly along the 1000 m isobath (from112E to 119E) in the northern South China Sea fromJanuary 1993 to December 1998. The plot in upper panelis made with TOPEX/Poseidon altimeter data relative to athree-year mean (January 1993 to December 1995; seehttp://ibis.grdl.noaa.gov/SAT/hist/tp_products/topex.html)[Cheney et al., 1994]. Monthly mean anomaly values are

Figure 12. (top) Potential and (bottom) relative vorticity along trajectories of particles P1–P6.

C09014 HSUEH AND ZHONG: SOUTH CHINA SEA WARM CURRENT

10 of 15

C09014

Page 11: A pressure&driven South China Sea Warm Currentshoni2.princeton.edu/ftp/lyo/journals/HsuehZhong-SCSWC...break pressure distribution arises from the collision with the continental slope

Figure 13. Sea surface elevation superimposed with the barotropic velocity at Y4D000 from the Asianmarginal seas model. There is no wind stress forcing. The 100 m isobath is represented by the heavycurve with crosses marking the location where values of sea level elevation are extracted for the along-isobath pressure plot in Figure 3. The straight line with circles mark the cross-isobath section along whichsalinity distributions are later shown to demonstrate the onshore movement from SCSBK to SCSWC.

Figure 14. Sea surface elevation superimposed with the barotropic velocity at Y6D226.50 from theAsian marginal seas model. The wind stress (0.2 N/m2) is southerly. The 100 m isobath is represented bythe heavy curve with circles marking the location where values of sea level elevation are extracted for thealong-isobath pressure plot in Figure 3.

C09014 HSUEH AND ZHONG: SOUTH CHINA SEA WARM CURRENT

11 of 15

C09014

Page 12: A pressure&driven South China Sea Warm Currentshoni2.princeton.edu/ftp/lyo/journals/HsuehZhong-SCSWC...break pressure distribution arises from the collision with the continental slope

then plotted. The plot in the lower panel is made similarlywith the Asian marginal seas model output. (The three-yearperiod for the model mean is Y004 to Y006 when theclimatological wind stress is applied.) The Kuroshio intru-sion is signaled by the appearance during winter of highanomaly first near the northeast end (top in each panel) ofthe stretch of the 1000 m isobath considered. The season-ality of the phenomenon is consistent with that expected inliterature [Qu, 2000]. The fact that the model outputessentially reproduces the wintertime-limited intrusion pat-terns once the climatological wind stress is applied supportsthe argument that the northeast monsoon probably promotesin some way the intrusion [e.g., Farris and Wimbush, 1996].The highest sea surface height anomaly values (orange-colored) appear at about the 600 km point, near theDongsha Islands. There is a general movement of thesignal to the southwest, at a speed somewhat greater thanthat for the b-induce movement of a baroclinic eddy [Nof,1981]. (The orange-colored patch from the model indeedrepresents the movement of a current ring. Another oneshows up at the end of the integration (not shown).) Thehigh, however, appears to linger near the Dongsha Islandsarea. It thus appears that the timescale of the Kuroshioimpingement is on the order of months, much greater thanthe setup time of the shelf circulation in response to anoffshore pressure distribution. The latter can be easilyshown to be several days [Zhong, 2001]. Thus the imposedpressure by the Kuroshio impingement can be consideredquasi-steady and the interpretation of the SCSWC as anATW solution seems valid. The corollary is then that in

reality the SCSWC must indeed at least possess the sameseasonality that is found in the Kuroshio intrusion.[38] There is also the issue of the pressure gradient in the

momentum balance of the SCSWC. For the ATW theory toapply, the alongshore gradient of pressure must be indis-pensable. Figure 17 shows a plot of the main terms in thevertically integrated momentum balance across the SCRUMchannel flow that features the SCSWC and the SCSBK. It isquite clear in this section 600 km from the east openboundary that the alongshore pressure gradient is quiteprominent in the momentum balance. The cross-channelbalance is mainly geostrophic. Given the fact that substan-tial gradient in sea surface height anomaly is indicated alongthe continental slope (Figure 16), it is expected that thealongshore pressure gradient must be important in reality aswell. It is certainly important in the Asian marginal seasmodel (Figure 3).[39] Finally, it should be mentioned that the Asian mar-

ginal seas model probably exaggerates the extent of Kur-oshio intrusion as the Luzon Strait islands are not fullyresolved. The inclusion of these islands has been shownto reduce the extent of modeled intrusion [Metzger andHurlburt, 2001]. On the other hand, the presence of Kur-oshio water along the continental slope of the northernSouth China Sea [Qu, 2000] seems to indicate that theintruded Kuroshio does at times reach the slope area beyond119�E. The composite velocity plot in the work of Liang etal. [2003] that seems to point to this longitudinal limit mightbe biased as historically there are few cruises in winterwhen the intrusion is at its strongest. (In fact, most of the

Figure 15. Salinity sections (left panels) along the straight line in Figure 13, one for (top left)Y6D283.50 and the other for (bottom left) Y6D325.50. On the right panel are barotropic velocitycomponents normal to the section, solid curve for Y6D283.50 and dashed curve for Y6D325.50.

C09014 HSUEH AND ZHONG: SOUTH CHINA SEA WARM CURRENT

12 of 15

C09014

Page 13: A pressure&driven South China Sea Warm Currentshoni2.princeton.edu/ftp/lyo/journals/HsuehZhong-SCSWC...break pressure distribution arises from the collision with the continental slope

Figure 16. Distance-time plots of sea surface height anomaly along the 1000 m isobath in the northernSouth China Sea. The monthly mean anomaly values are plotted in the upper panel and are based onTOPEX/Poseidon altimeter data relative to the January 1993 to December 1995 mean. Those plotted inthe lower panel are based on the output from the Asian marginal seas model relative to the Y004–Y006mean when the climatological wind stress is applied.

C09014 HSUEH AND ZHONG: SOUTH CHINA SEA WARM CURRENT

13 of 15

C09014

Page 14: A pressure&driven South China Sea Warm Currentshoni2.princeton.edu/ftp/lyo/journals/HsuehZhong-SCSWC...break pressure distribution arises from the collision with the continental slope

cruises are conducted in summer and only one-quarter of thecruises take place in winter (T. Y. Tang, personal commu-nication, 2004).) At any rate, the aim of showing resultsfrom the Asian marginal seas model is to demonstrate thatwhen the Kuroshio reaches the slope area, an alongshorepressure gradient is set up that could drive a SCSWC-likeflow. The change in the extent of the intrusion probablyaffects a quantitative, not qualitative, change in the SCSWCdynamics put forth.[40] It thus appears that the SCSWC is a quasi-steady

continental shelf flow feature that is driven by the along-shore pressure gradient arising from the impingment of theKuroshio that has intruded into the northern South ChinaSea through the Luzon Strait. As such, it inherits thetemporal character of the intrusion, which is at leastseasonal, weak in summer and strong in winter. Owing tothe prevalence of northeast winds in winter, the SCSWCthus manifests itself particularly well as a shelf break jet,which is what it is as an ATW solution.

[41] Acknowledgments. The research is supported by a grant fromthe Physical Oceanography Program of the Office of Naval Research(N00014-00-1-0406). The constructive comments of two reviewers aregratefully noted.

ReferencesChao, S.-Y., P.-T. Shaw, and J. Wang (1995), Wind relaxation as a possiblecause of the South China Warm Current, J. Oceanogr., 51, 111–132.

Cheney, B., L. Miller, R. Agreen, N. Doyle, and J. Lillibridge (1994),TOPEX/Poseidon: The 2-cm solution, J. Geophys. Res., 99, 24,555–24,564.

Chu, P. C., and C. Fan (2003), Hydrostatic correction for sigma coordinateocean models, J. Geophys. Res., 108(C6), 3206, doi:10.1029/2002JC001668.

Cox, M. D. (1984), A primitive equation three-dimensional model of theocean, GFDL Ocean Group Tech. Rep. 1, 144 pp., Geophys. Fluid Dyn.Lab., Princeton Univ., Princeton, N. J.

Csanady, G. T. (1978), The arrested topographic wave, J. Phys. Oceanogr.,8, 47–62.

Farris, A., and M. Wimbush (1996), Wind-induced Kuroshio intrusion intothe South China Sea, J. Oceanogr., 52, 771–784.

Guan, B. (1985), Some features of the temporal and spatial distributions ofthe ‘counter-wind’ current in northern South China Sea in winter (inChinese), Oceanol. Limnol. Sin., 16(6), 429–438.

Guo, Z., T. Yang, and D. Qiu (1985), The South China Sea Warm Currentand the southwestward current on its right side in winter: Tropic (inChinese), Oceanology, 4, 1–9.

Hedstrom, K. S. (1997), User’s manual for an s-coordinate primitive equa-tion ocean circulation model (SCRUM) version 3.0, Inst. of Mar. andCoastal Sci., Rutgers Univ., New Brunswick, N. J.

Hellerman, S., and M. Rosenstein (1983), Normal monthly wind stress overthe world ocean with error estimates, J. Phys. Oceanogr., 13, 1093–1104.

Hetland, R. D., Y. Hsueh, R. R. Leben, and P. P. Niiler (1999), A loopcurrent-induced jet along the edge of the west Florida shelf, Geophys.Res. Lett., 26(15), 2239–2242.

Hetland, R. D., Y. Hsueh, and D. Yuan (2001), On the decay of a baroclinicjet flowing along a continental slope, J. Geophys. Res., 106, 19,797–19,807.

Hsueh, Y. (1968), Viscous fluid flow over a corrugated bottom in a stronglyrotating system, Phys. Fluids, 11, 940–944.

Hsueh, Y. (2000), The Kuroshio in the East China Sea, J. Mar. Syst., 24,131–139.

Figure 17. The near balance of (top) vertically integrated along-channel and (bottom) cross-channelmomentum balance in a cross-channel section 600 km from the east open boundary in the SCRUMchannel flow. Included are the Coriolis (circle) (�f

R 0

�hvdz, f

R 0

�hudz for upper and lower panels

respectively), pressure gradient (x) (R 0

�h@p@xdz,

R 0

�h@p@ydz), bottom stress (star)(ru

h, rv

h), and diffusion

(cross) terms. Here h is the water depth and r = 1 cm/s the bottom drag coefficient.

C09014 HSUEH AND ZHONG: SOUTH CHINA SEA WARM CURRENT

14 of 15

C09014

Page 15: A pressure&driven South China Sea Warm Currentshoni2.princeton.edu/ftp/lyo/journals/HsuehZhong-SCSWC...break pressure distribution arises from the collision with the continental slope

Hsueh, Y., and L. Zhong (2003), A note on the deflection of a barocliniccurrent by a continental shelf, Geophys. Astrophys. Fluid Dyn., 97(5),393–415.

Li, R. F., Q. C. Zeng, Z. J. Gan, and W. Z. Wang (1993), Numericalsimulation of the South China Sea Warm Current and the Taiwan Straitflow in winter, Progr. Nat. Sci., 3(2), 123–129.

Liang, W.-D., T. Y. Tang, Y. J. Yang, M. T. Ko, and W.-S. Chuang (2003),Upper-ocean currents around Taiwan, Deep Sea Res. Part II, 50, 1085–1105.

Metzger, E. J., and H. E. Hurlburt (2001), The importance of high hori-zontal resolution and accurate coastline geometry in modeling SouthChina Sea inflow, Geophys. Res. Lett., 28(6), 1059–1062.

Nitani, H. (1972), Beginning of the Kuroshio, in Kuroshio: PhysicalAspects of the Japan Current Kuroshio, edited by H. Stommel andK. Yoshida, pp. 129–163, Univ. of Wash. Press, Seattle.

Nof, D. (1981), On the b-induced movement of isolated baroclinic eddies,J. Phys. Oceanogr., 11, 1662–1672.

Orlanski, I. (1976), A simple boundary condition for unbounded hyperbolicflows, J. Camp. Phys., 21, 251–269.

Qiu, D., T. Yang, and Z. Guo (1984), A westward flow in the northernSouth China Sea in summer (in Chinese), Tropic Oceanol., 3, 65–73.

Qu, T. (2000), Upper layer circulation in the South China Sea, J. Phys.Oceanogr., 30, 1450–1460.

Qu, T., H. Mitsudera, and T. Yamagata (2000), Intrusion of the NorthPacific waters into the South China Sea, J. Geophys. Res., 105(C3),7415–7424.

Shaw, P.-T. (1989), The intrusion of water masses into the sea southwest ofTaiwan, J. Geophys. Res., 94, 18,213–18,226.

Shaw, P.-T. (1991), The seasonal variation of the intrusion of the PhilippineSea water into the South China Sea, J. Geophys. Res., 96, 821–827.

Su, J., B. Guan, and J. Jiang (1990), The Kuroshio. part I. Physical features,Oceanogr. Mar. Biol. Annu. Rev., 28, 11–71.

Wyrtki, K. (1961), Physical Oceanography of the Southeast Asian Waters,NAGA Rep., vol. 2, Scientific Results of Marine Investigations of theSouth China Sea and the Gulf of Thailand, 195 pp., Scripps Inst. ofOceanogr., La Jolla, Calif.

Xue, H., F. Chai, N. Pettigrew, D. Xu, M. Shi, and J. Xu (2004), Kuroshiointrusion and the circulation in the South China Sea, J. Geophys. Res.,109, C02017, doi:10.1029/2002JC001724.

Zeng, Q., R. Li, Z. Ji, Z. Gan, and P. Ke (1989), Calculations of the monthlymean circulation in the South China Sea (in Chinese), Sci. Atmos. Sin.,13, 127–168.

Zhong, H. (1990), Density-related current structures (in Chinese), in Reportof 10-year Hydrographic Sections Surveys of the Northern South ChinaSea Continental Shelf Region and Adjacent Water, edited by Y. Ma,pp. 215–239, China Ocean Press, Beijing.

Zhong, L. (2001), The circulation in the northern South China Sea, Ph. D.dissertation, 116 pp., Fla. State Univ., Tallahassee.

�����������������������Y. Hsueh, Department of Oceanography, Florida State University,

Tallahassee, FL 32306-4320, USA. ([email protected])L. Zhong, Horn Point Laboratory, University of Maryland Center for

Environmental Science, 2020 Horn Point Road, Cambridge, MD 21613-0775, USA.

C09014 HSUEH AND ZHONG: SOUTH CHINA SEA WARM CURRENT

15 of 15

C09014