Surface circulation in the South China Sea

Surface circulation in the South China Sea

Deep-Sea Reseanch 1, Vol. 41, No. 11/12, pp. 1663-1683, 1994 Pergamon Copyright © 1994 Elsevier Science Ltd Printed in Great Britain. All rights res...

1MB Sizes 2 Downloads 85 Views

Deep-Sea Reseanch 1, Vol. 41, No. 11/12, pp. 1663-1683, 1994

Pergamon

Copyright © 1994 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0967-0637/94 $7.00+0.00

0967--0637(94)00026-3

Surface circulation in the South China Sea PING-TUNG SHAW* a n d SHENN-YU CHAOt

(Received 5 November 1993; in revised form 27 April 1994; accepted 13 May 1994) Abstraet--A three-dimensional, primitive-equation model with a free surface is used to simulate the monthly circulation in the South China Sea. The model has a resolution of 0.4° in the horizontal and 21 layers in the vertical in a region from 2°N to 24°N and from 99~E to 124°E. Inflow and outflow in the Kuroshio, through the Taiwan Strait, and between the Sunda Shelf and the Java Sea are prescribed bimonthly. At the sea surface, the model is forced by monthly-averaged climatological winds and temperature and seasonally-averaged salinity. Several important features are reproduced in the model simulation. First, a strong coastal jet is present at the western boundary. The current is southward along the continental margin from China to southern Vietnam in winter. In summer, the current is northward and separates from the coast between ll°N and 14°N. The transition in September begins as a southward undercurrent, which is remotely forced by the northeast monsoon in the northern reaches of the South China Sea. The undercurrent extends to the surface in about a month. Second, inflow through the Luzon Strait from October to February transports the Kuroshio water in the top of 300 m of the water column westward along the continental slope south of China. In summer, eastward flow in the Luzon Strait transports surface water west of Luzon to the region east of Taiwan. Finally, a subsurface current, which is opposite to the surface current, exists over the Sunda Shelf and is driven by a pressure gradient set up by monsoon winds. These simulated currents are in qualitative agreement with the circulation inferred from the available observations.

INTRODUCTION THE S o u t h C h i n a S e a is th e largest m a r g i n a l sea in S o u t h e a s t A s i a with a m a x i m u m d e p t h r e a c h i n g 5000 m (Fig. 1). It e x t e n d s f r o m t h e e q u a t o r to 23°N an d f r o m 99°E to 121°E. T h e d e e p basin in t h e c e n t e r is b o r d e r e d by t w o b r o a d sh el v es s h a l l o w e r t h a n 100 m. T h e n o r t h e r n o n e , e x t e n d i n g f r o m T a i w a n s o u t h w e s t w a r d to 13°N, consists o f t h e shelf so u t h o f C h i n a an d t h e G u l f o f T o n k i n . E x c h a n g e o f s h e l f w a t e r s b e t w e e n t h e S o u t h C h i n a Sea an d t he E a s t C h i n a S ea occurs t h r o u g h t h e T a i w a n Strait at a sill d e p t h o f 60 m. T h e s o u t h e r n shelf consists o f t h e G u l f o f T h a i l a n d to t h e west an d t h e S u n d a S h e l f b e t w e e n M a l a y P e n i n s u l a a n d B o r n e o . T h e S u n d a S h e l f is o p e n to t h e J a v a S e a ( o u t s i d e t h e m a p in Fig. 1) to t h e s o u t h an d is c o n n e c t e d to t h e I n d i a n O c e a n t h r o u g h t h e Strait o f M a l a c c a . O n t h e east side, t h e P h i l i p p i n e s a n d P a l a w a n s e p a r a t e t h e S o u t h C h i n a S ea f r o m t h e Pacific

*Department of Marine, Earth and Atmospheric Sciences, North Carolina State University, P.O. Box 8208, Raleigh, NC 27695, U.S.A. tHorn-Point Environmental Laboratory, University of Maryland, P.O. Box 775, Cambridge, MD 21613, U.S.A. 1663

1664

PING-TUNG SHAWand SHENN-YU CHAO

Taiwon Stroit

I00 ° E

A

I IO°E

120°E

Fig. 1. Map of the South China Sea showing the domain of simulation. The small insert at the upper-left corner showsthe locationof the studyregion relative to the PacificOcean and the Indian Ocean. Bottom topographyis illustrated by the 100, 200, 1000, 2000, 3000 and 4000-m isobaths. Open ocean boundaries are indicated by thick dashed lines (A, B, C and D). Ocean. The continental slope is steep and with practically no continental shelf on the east side of the basin. There are three openings along the eastern boundary. The widest and deepest one is the Luzon Strait with a sill depth of about 2000 m. Two narrow, shallow passages to the north and south of Palawan Island connect the South China Sea to the Sulu Sea. The major circulation in the South China Sea is driven by the monsoon winds (WYRTKI, 1961). The annual cycle of the wind stress field (HELLERMANand ROSENSTEIN,1983) is illustrated in Fig. 2. Winds prior to September are dominated by the southwest monsoon. In September, the northeast monsoon begins to appear in the seas north of 20°N. South of that latitude, the southwest monsoon still prevails. The northeast monsoon, expanding southward against the diminishing southwest monsoon in October, reaches its maximum strength and covers the entire South China Sea in December. April marks the end of the winter monsoon. The southwest monsoon first appears in the central basin in May and expands over the entire basin in July and August. The most significant ocean circulation in response to the changing winds is the current off the coast of Vietnam. In Wyrtki's surface current charts, a strong southward current develops along the coast of Vietnam from October to February during the northeast monsoon; the current becomes northeastward during the southwest monsoon. Another circulation pattern, which is significant only along the northern boundary of the basin, is the Kuroshio intrusion through the Luzon Strait in winter. A m o n g the three passages along the eastern boundary, the Luzon Strait is the only opening where there is

Surface circulationin the South China Sea

25"N l(a)Sep ' ' '~ Windstre~,

I

(ciDec Wind stress. ~:~-,~F'_'f~f 0.3 Pa :i

5

°

N

~

(b) Oet

t (el Jun Wind st. . r m ~ . (13 Pa . : ~ ' ~

/Z

..... _ _ ~ .'.

~.:

!

I

(d)May

:~

.~

i!!i 1 IOOE

;~;t I.....

I

if)Aug

"i~'~ : -

100°E

'#

;;;',;;;;' ,

25*N

1665

120°E

1000E

~ .

'"

~.

r / / / / / / P /

'

'i ......... 1 IO°E

1200E

1000E

1 IOOE

1200E

Fig. 2. Windstress distribution from HELLERMANand ROSENSTEIN(1983). Months are chosen to show the development of the winter and summer monsoons. (a) September, (b) October, (c) December, (d) May, (e) June and (f) August. The scale vector is 0.3 Pa for all plots.

significant water exchange between the South China Sea and other basins below 100 m. Water originating from the Pacific at depths between 1500 and 2000 m has been known to enter the Luzon Strait and to form the deep water in the South China Sea (NITANI, 1972). At shallower depths, the Kuroshio water has been observed along the continental slope south of China from October to February (SHAW, 1991). Some of the intruding water flows through the Taiwan Strait and enters the East China Sea in spring (SHAW, 1992). The maximum transport through the Luzon Strait is about 3.0 × 106 m 3 s -1, which is comparable to the transport associated with the basin-wide circulation in the South China Sea (WVRTRI, 1961). The intrusion coincides with the appearance of the northeast monsoon. Therefore, it may be part of the wind-driven circulation in the South China Sea, but processes in the Kuroshio may also contribute to the intrusion. The inflow water through the Luzon Strait is practically the only salinity source for surface waters in the basin. Thus, variations in the Kuroshio intrusion may affect the seasonal circulation in surface layers through modification of water characteristics in the northern South China Sea. The relatively small amount of mass exchange between the South China Sea and the surrounding waters may explain why the basin has received little attention from modelers. Even in the most advanced high-resolution global circulation model (SEMTNER and CHERVIN, 1988), the South China Sea is excluded. A barotropic model (ZENG et al., 1989) does exist but has the obvious drawback of failing to resolve the three-dimensional circulation. It nevertheless provides an interesting comparison. The South China Sea is

1666

PING-TUNGSHAWand SHENN-YUCHAO

included in the reduced-gravity model of INOUE and WELSH (1993) for the tropical Pacific. Besides the issue of vertical resolution, their emphasis is on how the South China Sea modified the Kuroshio transport instead of the South China Sea circulation itself. Of more relevance is the three-dimensional, primitive-equation model of POHLMANN (1987), who computed the circulation in the South China Sea in January and July, using climatological stratification and forcing. The boundary conditions in his model require vanishing normal derivatives of the normal velocity components on open boundaries. Although stabilizing the computation, these open boundary conditions limit the prognostic skill of the model to less than a month. Longer simulations must account for the water mass exchanges through open boundaries in a more realistic manner. In the present paper, a three-dimensional, primitive-equation model is used to continue the existing modeling efforts and to investigate the seasonal variation of the basin circulation driven by monthly varying winds and inflow/outflow through open boundaries. MODEL DESCRIPTION The model solves the three-dimensional Navier-Stokes equations with the Boussinesq and hydrostatic approximations. The general circulation model is similar to that of SEMTNER(1974) except for the addition of a free surface. The choice of using vertical grids at constant depths over terrain-following sigma coordinates is to reduce truncation errors associated with the horizontal pressure gradient force. This force becomes significant in sigma-coordinate models when steep topography is encountered (HANEV, 1991). The integration domain is from 2°N to 24°N and from 99°E to 124°E. The horizontal grid size is 0.4 ° . The study region covered the entire South China Sea except for the shelf area south of 2°N. The model includes the southern half of the Taiwan Strait and the Kuroshio Current east of Taiwan and Luzon. Bottom topography in a global 1/12 degree data set obtained from the Marine Geology and Geophysics Division of the National Geophysical Data Center is smoothed to fit the 0.4 ° resolution. In the vertical direction, the model has 21 layers with z = 0 m at the surface. Variables are evaluated at the center of each layer at z = - 2 . 5 , - 10, - 2 0 , - 3 0 , - 5 0 , - 7 5 , - 100, - 125, - 150, - 2 0 0 , - 3 0 0 , - 5 0 0 , - 7 0 0 , - 9 0 0 , -1200, -1500, -2000, -2500, -3000, - 4 0 0 0 and - 5 0 0 0 m. These depths are chosen to be as close to the N O D C standard depths as possible for easy implementation of initial temperature and salinity. Horizontal mixing is achieved by using a Laplacian mixing coefficient of 5 × 103 m 2 sin the momentum, temperature and salinity equations. The vertical viscosity and diffusivity are calculated from the Richardson number according to the formulae of PacaNowsKi and PHXLANDER(1981). Wind forcing is applied to the surface of the first layer as shear stress. Subsequent vertical mixing, following PACANOWSKI and PHILANDER'S (1981) parameterization scheme, produces an Ekman layer involving the top four layers, about 35 m deep. The surface temperature and salinity are prescribed. At the bottom, heat and salt fluxes through the boundary vanish, and a quadratic drag law with a dimensionless coefficient of 0.001 is used for bottom friction. The climatological temperature and salinity distributions for January (LEVlTUS, 1982) are used in the initial conditions, The model ocean is spun up from rest by monthly wind stress fields (HELLERMANand ROSENSTEIN,1983). In the model, temperature, salinity and the vertical shear of the horizontal velocity are resolved with an internal-mode time step of 2160 s, while the sea level elevation and the vertically averaged horizontal velocity

Surface circulationin the South China Sea

1667

Table 1. Bimonthly transports (in 106m3 s-1) across the open boundaries shown in Fig. 1. Positive and negative values represent inflow and outflow, respectively

Open boundaries Month February April June August October December

A

B

C

D

-3.0 0 3.0 3.0 - 1.0 -3.5

27.0 37.5 32.0 28.0 27.5 24.5

-24.5 -37.5 -34.0 -30.5 -27.0 -21.5

0.5 0 -1.0 -0.5 0.5 0.5

components are calculated with a much smaller time step of 21.6 s. This mode-splitting technique essentially follows that of MELLOR (1993). The monthly distribution of surface temperature and the seasonal distribution of surface salinity are obtained from LEvrrus (1982). Transports through the open boundaries in the model ocean are treated as follows. The Pacific Ocean east of the Kuroshio is excluded by placing a meridional vertical wall at 124°E. The two narrow passages to the Sulu Sea between the Philippines and Palawan and between Palawan and Borneo are also closed by vertical walls. These two channels are too narrow to be resolved by the present model. If left open in a coarse-grid model, these artificially widened passages produce the undesirable effect of weakening the alongshore currents in the South China Sea. There remain four open boundaries, marked by thick dashed lines in Fig. 1, in the model. Transports at open boundaries A and D account for inflow/outflow across 2°N and through the Taiwan Strait, respectively, while the difference in transports at B and C allows inflow and outflow through the Luzon Strait. Bimonthly transports estimates across these open boundaries are obtained from WYRXKI (1961) but are slightly adjusted to compensate for the closing of the narrow passages to the Sulu Sea (Table 1). Inflow and outflow conditions are applied to the values of velocity, temperature and salinity at the open boundaries. Let (U, V) be the vertically averaged velocity on the boundary with the coordinate axes parallel and normal to the boundary and (u, v) be the deviation from the vertical mean. The normal component V is specified by V = Vb, where Vb is obtained from the transport values in Table 1 divided by the bottom depth. For v and the sea level elevation, the normal derivatives vanish. Other variables, U, u, temperature and salinity, are assumed to be advected by the normal velocity, V + v. In the case of inflow, U and u are set to zero, and the climatological temperature and salinity values are specified at points one grid space outside the boundary. This set of boundary conditions is stable for outflow but often becomes unstable under inflow conditions. The instability seems to be associated with the Arakawa B-grid employed in this model, since similar boundary conditions in the Arakawa C-grid are generally stable for inflow (MELLOR, 1993). This instability is suppressed by placing a sponge layer over either three or four grid points next to the boundary in the barotropic velocity field. Within the sponge layer, U and V are replaced by (1 - e)U and (1 - e ) V + eVb, respectively. In the model computation, e is 0.05.

1668

PING-TUNG SHAW and SHENN-YU CHAO

Climatological forcing is applied at the beginning of each month without nudging, and the model is integrated for three years. However, the flow field in the third year is visually indistinguishable from that in the first year. The result is consistent with that of POHLMANN (1987), who found that a steady-state response to wind forcing can be reached in about 15 days. In the latter part of this paper, results on the fifteenth of each month in the third year are analysed to provide a description of the circulation in response to the monthly climatological conditions. MODEL RESULTS The simulated currents in the South China Sea are mostly Consequently, the discussion is focused on the circulation in column. Currents over the continental slope south of China outflow through the Luzon Strait, and circulation on the features to be examined in this paper.

confined in the upper ocean. the upper 500 m of the water and off Vietnam, inflow and Sunda Shelf are the major

Ekman transport Typical summer (August) and winter (December) velocity fields in the top layer centered at z = - 2.5 m are shown in Fig. 3. In August, velocity vectors north of 6°N generally veer to the right of the wind stress vectors given in Fig. 2f, suggesting that the Ekman drift dominates the flow field at this depth. The Ekman drift is proportional to the intensity of the wind stress in the northern basin. In the southern basin, the surface currents also agree with the Ekman flow except in the proximity of the equator. The eastward Ekman drift is maximum at a latitude south of the maximum northward wind, as one would expect for the Coriolis parameter approaching zero near the equator. The Ekman transport accumulates water along the eastern boundary in summer (Fig. 4a). The effect is the strongest near Borneo and Palawan. On the west side of the basin, surface waters in the Gulf of Tonkin and in the Gulf of Thailand are transported toward the deep water. South of 6°N on the Sunda Shelf, the current is mostly downwind, indicating the lack of Ekman flow. In December, velocity vectors show similar veering under the winter monsoon (Fig. 3b). Since the maximum wind in winter is in the northern South China Sea (Fig. 2c), the surface flow is also strong there. Waters are transported off the coast of Luzon and Palawan by the Ekman drift. On the west side, the surface Ekman drift is into the Gulf of Tonkin and the Gulf of Thailand. On the Suna Shelf, the current is mostly downwind as in August; the surface water is transported into the Java Sea farther south. Ekman flow generally dominates the velocity field to a depth of about 35 m in the model. Since the remaining portion of this paper deals with the interior flow beneath the Ekman layer, the Ekman layer will be excluded from the vertical sections of velocity. Furthermore, currents at a depth of 50 m will be used to represent the surface circulation.

The surface elevation field and surface currents Following the onset of the southwest monsoon over the central basin in May, the winter circulation gradually changes to the summer one. By August, the summer circulation is fully developed (Fig. 4). The region of higher sea level is distributed along the southeast side of the basin, mainly along Luzon, Palawan and Borneo; lower sea level is along the

1669

Surface circulation in the South China Sea 250N

Io, Velocity at 2.5m

~:~

15°N

~ :

..." -"':~..

-

:,,..:~.." ": . . . . . . . .

~,

~



~

.

.

.

.--" '~.

.

.

.

.

.

,

_ _ .

.

.

.

.

.

-

A,

I

5*N .": .:~

~,

3!

/

/ ." ..

,

100°E 25°N

110°E

(b) Dec

~

Velocity at 2 . 5 m ----

~

15ON -

"

~ ~ ~ ~ - - - ~ - ~.~. .

~

0.8 m/s

L ~ ~. ~ .

' ~ ~~

"-~"-------~

.

. . . . . . . . . . . . . .

,t."%J

_ "~

5°N-

~

~

~

Fig. 3.

120°E

~

.~]IP-.~,, -'-''-'--'~

~

,

~

~

~

J

" ' ~~

~

.

~z~.j~.~

~

. . . . .

~

.__#. " ".'.~

i

/

; P'

/

I

I

I

100°E

110°E

120°E

Velocity vectors derived from model simulation at a depth of 2.5 m. The velocity scale is 0.8 m s -1. (a) August and (b) December.

northwest side from Taiwan to Vietnam. Sea level is the lowest in the Gulf of Tonkin and in the Gulf of Thailand. The sea level difference is largely caused by the Ekman transport described earlier. Sea level over the Sunda Shelf is generally higher in the southeast and lower in the northwest; this trend continues well into the Gulf of Thailand. North of 6°N,

1670

PING-TUNG ShAW and SHENN-Yu CHAO

25°N Sea level (m)

.......

~-.""

V.

.(,i; ,,i :9+ ,,: / %,.:. '.

;

,

15ON -

<:...i ~

S°N

-~,:

,i:.~i~,.'"..-" O.

-

::i.,: 25°N]

15°N-[

°°+

i

I

lO0°E

1 lO°E

Aug

_I

i

120°E ) ....

:>:"

.

/<" it,.+-

.~

..

l l v ........

\ ' % ' % \ \ \ X --.. . . . . .

I

IO0°E Fig. 4.

I

~.'-.~

I

1 IO°E

120°E

(a) Contours of constant sea level elevation in August. t h e contour interval is I).04 m.

Dashed contours represent negative values. (b) Velocity field at 50 m in August. The velocity scale is 0.4 m s I t h e r e are two regions w h e r e sea level d r o p s s h a r p l y t o w a r d the west; i.e. in the L u z o n Strait a n d off the s o u t h e a s t coast of V i e t n a m . A t 50 m, s t r o n g c u r r e n t s a l o n g c o n t o u r s of c o n s t a n t sea level a r e p r e s e n t in t h e s e two r e g i o n s (Fig. 4b). S o u t h e a s t o f V i e t n a m , t h e r e is a s t r o n g n o r t h e a s t w a r d b o u n d a r y current. T h e c u r r e n t leaves the coast b e t w e e n 11 a n d t4°N a n d

Surface circulationin the South China Sea

1671

diffuses toward northeast. In the Luzon Strait, the flow is toward the Pacific. Both currents are in geostrophic balance with the sea level gradients. Over the Sunda Shelf, the current at a depth of 50 m is forced by the sea level gradient toward northwest, entering the Gulf of Thailand to compensate for the outflow near the surface. The beginning of the northeast monsoon in September changes the surface elevation pattern in the northern South China Sea. E k m a n transport is now toward west. In the October sea level distribution, the highest sea level is along the coast between Hong Kong and Hainan (Fig. 5a). The sea level in the deep central basin is low. Associated with the sea level change is the development of a current on the continental slope from Hong Kong southward to about 7°N (Fig. 5b). There is a weak return flow along the eastern boundary from Borneo to Luzon. In the meantime, the flow in the Luzon Strait becomes northwestward, indicating intrusion of the Kuroshio into the South China Sea. The current at 50 m over the Sunda Shelf is weak in October. The development of the winter circulation continues from October to December. The high sea level off the coast of China expands southward along the coast of Vietnam, following the southward movement of the northeast monsoon. By D e c e m b e r (Fig. 6a), the sea level is high along the entire western boundary. The highest sea level is in the Gulf of Thailand. Low sea level extends southwestward in a tongue from Taiwan and the Luzon Strait. Current vectors show that the flow off the coast of China is still present, and the current in the southern basin is strengthened (Fig. 6b). The southward current is the strongest southeast of Vietnam, returning northeastward off the coasts of Borneo and Palawan. In the Luzon Strait, flow at 50 m is toward the South China Sea as in October. On the Sunda Shelf, a strong eastward current appears. The shelf circulation is consistent with a flow driven by a pressure-gradient force set up by the higher sea level in the Gulf of Thailand, indicating that the Coriolis force is negligible at these latitudes. Inside the Gulf of Thailand, the surface inflow returns seaward at a depth of 50 m. The two-layer circulation is opposite to that in August. The sea level distribution and currents at 50 m in February show the later stage of winter circulation. The highest sea level is still in the Gulf of Thailand and the lowest sea level remains in the Luzon Strait (Fig. 7a). This sea level distribution suggests accumulation of water on the southwest side of the basin by the current along the western boundary. In the velocity field at 50 m, the current south of China is weakened and so are the currents in the southern basin. Nevertheless, the general circulation pattern in February is similar to that in December. After February, the strength of the currents decreases rapidly, and the winter circulation pattern can no longer by recognized in April. The summer circulation pattern similar to that in August sets up rapidly in May. CURRENT SYSTEMS IN THE SOUTH CHINA SEA

Reversal of the alongshore current off Vietnam The structure of the current system off the coast of Vietnam is demonstrated in a zonal vertical section of the meridional velocity along 12.2°N. In August, there is a welldeveloped northward coastal jet in the top 500 m of the water column (Fig. 8a). The maximum speed is 0.1 m s -1 at the surface and decays both offshore and toward deeper waters. A southward current begins to appear below 50 m in September (Fig. 8b). In the meantime, the maximum northward flow is weakened and is displaced offshore. By

1672

PING-TUNG SttAW and SHENN-Yu CHAO

25ONSea7eC;,(m>

",i~

15oN-

SON-

100°E

T"

I

110°E

120°E

25°N ¢elocltyOCt at 50m ~ 0.4m/s

..fl: ~ '

15°N

"

~

i

'

i

~i

-

/

~

~iii ~ ~ i i i i i i i. ~ i~ ;, .

.

.

.

.

.

t

-

' ~,~:.

::3 i l .........

%

"~.~

Z SON-

#.

:],

lO0°E

,

,

.

.

.

.

.

.

.

.

.

.

1lO°E Fig. 5.

_2

1200E

Same as Fig. 4 but for October.

N o v e m b e r , the southward current extends to the surface and reaches a m a x i m u m speed of 0.2 m s - l , while a weak n o r t h w a r d flow spans most of the basin interior (Fig. 8c). T h e strength of the coastal jet decreases afterward. In F e b r u a r y and March, the southward current extends eastward over the entire basin but is shallower. T h e m a x i m u m speeds are reduced to 0.09 and 0.03 m s -1 , respectively (Fig. 8d,e). T h e r e is still a w e a k return flow at

Surface circulation in the South China Sea

1673

25°N

.........:

Dec

,.i..~'';

.

~

/,'",,'" ~'

,,,"'"

,

15ON -

S°N

-

100°E

110°E

120°E

25"N Velocity at 50m

........... . . . ~ . .

) ~,.,

,

15°N

~-':.."..~,:.;..

..i~" ~ ~

.L -~,

5"N

..... ' 2:

\

,-

~:

7/~,

~

~

. . . . . . .

, . . . . . . .

,h "j ,-

-.g~ ~ -..-.,,--~----~

. . . . . .

,

I

100°E

110"E

120°E

Fig. 6. Same as Fig. 4 but for December. depth in the central basin. In May, the northward current begins to develop along the coast of Vietnam and reaches a speed of 0.07 m s -1 (Fig. 80. The coastal jet spreads into the central basin and reaches the m a x i m u m penetration in August. The velocity field shows that variations in currents off Vietnam are dominated by a coastal jet, which is northward from May to August and is southward from October to

1674

PIN6-TuN6SHAWand SHENN-YuCHAO 25°N Sea level (m)

~ .]"

.-"" . . . .

'. .

.

.

.

'

,"

o

....

15ON -

S°N -

IO0°E

110°E

120°E

25°N Velocity at 5 0 m

0.4m/s

.... . : , ~ ~

,

~

, , ,

-~.

[ ; : : .......

15°N

9! .............

:~,: ,f.. # /.-.¢ •

'

~

~::. .:Jl l ~ . . . . . . . . . . . ~ , : . ~ z J ' . . . . . . D

"~~ . ~~-~

5oN

i

IO0°E Fig. 7.

I

I

110°E

120°E

Same as Fig. 4 but for February.

February. The transition in September is rapid. It begins at depth and progresses from north to south. Since the current does not reach the surface, it is not driven by local winds. Instead, the propagation is likely due to southward propagation of coastally trapped waves along the coast of Vietnam. As noted earlier, the northeast monsoon develops from the

1675

Surface circulation in the South China Sea

__•_..;-•°

-,. ;/"

0

^

I,,iJ @

o

v

I.u

E~

o

~

>~ o

o

o

o

!

m

I

o

0

g

o -

!

6 o

6o

!

!

-

I

1

°o oo ~"

tO

!

$

z=

(~u)z

0

o

f._.--------

"EO'O~

~

o

A

o

~-::-'::-'- ...... 0------~ lljjjijJiijiiiiii.::iii;i~ii.-:-~-----:::-::--. %:- -::. ;:.. ~~A~.~..-' - , . . . ........~-~:~:::-.:,---.>



o

!

04

3v

m

0

!

o

o 0

6 0

o

!

!

!

!

6

oo 0

0

0

6

8

i

!

!

i

0

(w)z ,.,

-~=~

/

ov ~°

--

~

-

,0 "0=:i_.. 0

>(~J

6 .o

o !

i

8 o I

<[

6 o8 I

o

&

0 -

I

(w)z

6

~

!

!

0

6

o 0 !

~T

~ ~ N~

1676

PING-TUNG SHAW and SHENN-Yu CHAO

~

0

- I00

m

- 200 - 300-

- 400-

T

12.S?.° N ~ 3 4 - 5 - -

I

12.2°N Aug

(a)Aucj - 500-

E

i

0

N

- I00 -200

S

- 300-

-400-12.2°N (b) Feb

T

-

--34.5 --

-[

15

12.2*N Feb

IO105*E I IO*E 115"E 120=E 105"E IIO='E 115*E 120" E Fig. 9. Salinity(S, left panels) and temperature (T, right panels) along the velocitysection shown in Fig. 8. The contour intervals are 0.1 for S and I°C for T in (a) August and (b) February. -

500

,

northern South China Sea in September, and the local sea level rise in response to wind forcing produces an alongshore pressure gradient (Fig. 5a). The pressure gradient forces a southward current, which expands southward along the Vietnamese coast. The mechanism is similar to an arrested topographic wave described by CSANAOV (1978). This nearly barotropic current weakens the surface-intensified northward coastal jet and produces a southward undercurrent in deeper waters. In about a month, the southward undercurrent surfaces and replaces the northward summer coastal jet. The reversal from a southward current in winter to a northward current in summer is slower because of weak winds in March and April. Associated with the development of the current off the coast of Vietnam is the vertical movement of isotherms. In August, both the temperature and salinity fields show upwelling along the coast of Vietnam and downwelling at the eastern boundary (Fig. 9a). The lowest salinity is located at the surface near the western boundary because of river runoff from Vietnam in summer. Although the present simulation does not include river forcing explicitly, specifying the surface temperature and salinity by the climatological values ensures its presence to some degree. In February under the northeast monsoon, there is upwelling (downwelling) at the eastern (western) boundary (Fig. 9b). The salinity is still the lowest at the surface near Vietnam because of river run-off. Upwelling and downwelling off the coast of Vietnam and off Palawan are mostly in the top 200 m of the water column and are in agreement with the numerical predictions of POHLMANN (1987).

1677

Surface circulation in the South China Sea 0

/

-I00

,/

0

°tl /

-I00

<0

\

- 200-

/

U

114°E )Aug

-300 E

E

O-

N

/ i \i()i)"-i"~F --[j -I00-

o

!

- 200-

.

i:?':~"

"100

-200-- t

,,

oS

U 0

"300 II°N

-3OO I4°N

17"N

20°N

23aN

II*N

\

l)Jun 14°N

17°N

20°N

23°N

Fig. 10. Zonal velocity in a meridional vertical section from Borneo (left) to Hong Kong (right) along 114°E. The depth is between 0 and 300 m. The contour interval is 0.01 m s-1 . Solid (dotted) lines represent eastward (westward) velocity. The surface E k m a n flow above 35 m is not plotted for clarity. (a) August, (b) October, (c) March and (d) June.

Kuroshio intrusion current south of China Another significant current system in the basin is the westward flow over the continental slope south of China in winter. This current appears only during the northeast monsoon and virtually vanishes in summer as shown in the velocity field at 50 m. The seasonal variation of this current is demonstrated in a meridional vertical section of the zonal velocity along 114°E (Fig. 10). In August, the eastward flow south of 17°N is a continuation of the coastal jet off Vietnam, and the current over the continental slope south of China is weak. A westward current develops following the beginning of the northeast monsoon in September and reaches 0.15 m s - ] in October (Fig. 10b). The current is the strongest over the northern continental slope. It decreases in strength afterward, and the maximum current is less than 0.07 m s -1 in March (Fig. 10c). In June, the westward flow is replaced by a weak eastward flow of 0.07 m s - t after the onset of the southwest monsoon (Fig. 10d). The temperature and salinity distributions shown in Fig. 11 demonstrate seasonal variations of water characteristics along the same meridional section. In August, water above 200 m in the central basin is warmer and fresher than that along the continental slope south of China. The warm, fresh water is associated with the northeastward flow leaving the coast of Vietnam in the central basin (Fig. 4b). From August to February, there are downward movements of both the 22°C isotherm and the 34.3 isohaline north of 17°N. The water above 100 m over the continental slope also becomes nearly uniform with tempera-

1678

PING-TUNG SHAW and SHENN-Yu CHAO

~4"0~

2 4 - -

"100, 22~-20

-30o

,

+--~

+1 Aug

E

-200-

i~:6-~~ I

oE

--

~0~-------I

8

p

- 3oo-

](b) eb

II°N 14°N 17°N 20°N

m

II°N 14°N 17°N 20~N

Fig. 11. Same as Fig. 9 except for the meridional section in Fig. 10. (a) August and (b) February. ture between 22 and 23°C and salinity between 34.2 and 34.3. The relatively warm, salty water is supplied by the Kuroshio intrusion. However, because the surface temperature and salinity are fixed by climatology in the model, the dynamics of the intrusion process may not be adequately simulated. The westward current in October and D e c e m b e r begins on the continental slope west of the Taiwan Strait, reaches the highest velocity east of Hainan, and joins the coastal jet off Vietnam to the south of Hainan (Figs 5 and 6). The disappearance of the current along the southwest coast of Taiwan is probably caused by the gap in the Taiwan Strait. The much weaker current in summer is due to the fact that the southwest monsoon is mostly in the central basin while the winter monsoon extends farther to the north (Fig. 2). The current south of China is probably wind-driven, but other mechanisms may govern the inflow of warm, salty water through the Luzon Strait and in the region southwest of Taiwan. Current structure in the Luzon Strait

Typical outflow and inflow velocity structures through the Luzon Strait are shown in a meridional section of the zonal velocity between 0 and 1500 m at 120.8°E (Fig. 12). Water exchange between the South China Sea and the Pacific Ocean is mostly concentrated in the top 300 m of the water column. In summer, flow in the surface layer is toward the Pacific across the entire strait (Fig. 12a). The maximum velocity is 0.04 m s -1 . Below 300 m, there is a weak outflow on the north side of the strait and a weak inflow on the south side. In winter, the flow is into the South China Sea in the surface layer with a m a x i m u m speed of

Surface circulationin the South China Sea

1679

u 120.8" E (b) Nov

(a) Aug

0 50O

-500

- 1

I: 5

0

P

-I000- / I -1500 ~0

~

20*N 22"N Fig. 12. Same as Fig. 10 except for a section from Luzon (left) to Taiwan (right) along 120.8°E. The depth is from 0 to 1500 m. (a) August and (b) November. 20°N

22"N

0.05 m s - 1 (Fig. 12b). The flow below the surface current is weak and is toward the Pacific on both north and south sides of the strait. At the center of the strait, there is a weak inflow extending to the bottom. The seasonal variations of inflow and outflow at the surface in the Luzon Strait agree with the current field at 50 m. In summer, the current follows contours of sea surface elevation closely (Fig. 4). Therefore, the outflow in shallow layers is a geostrophic current produced by a higher sea level off Luzon than off the coast of China. The geostrophic contours in Fig. 4a show that the northeastward flow from Vietnam to Luzon prevents the Kuroshio water from entering the South China Sea in summer. Further, this current enters the region between the east coast of Taiwan and the Kuroshio front and pushes the Kuroshio front away from the coast of Taiwan. The inflow of the Pacific water in surface layers in winter, on the other hand, is not balanced by a sea level gradient (Fig. 6a). This inflow is likely an inertial current governed by nonlinear processes. Although the total transport in the Luzon Strait is given by the boundary conditions at B and C, the baroclinic current structure probably depends on the dynamics in the Kuroshio and the basin-wide circulation. COMPARISON WITH OBSERVATIONS AND OTHER MODEL RESULTS Because of the lack of data in the South China Sea, a quantitative comparison with observations is not possible. Nevertheless, it is feasible to compare the major features obtained from this simulation with observations qualitatively. Several data sets are available. Monthly surface velocity vectors were compiled from ship drift data by LEVITUS (1982). WYRTKI(1961) described the circulation and the wind system based on data before 1960. Using historical hydrographic station data and recent CTD observations, SHAW (1989, 1991) examined the circulation associated with the Kuroshio intrusion in the northern South China Sea. Surface current vectors inferred from ship drift were given monthly by LEVITUS(1982) and bimonthly by WYRTrI (1961). However, there are some discrepancies between Wyrtki's and Levitus' charts. Levitus' summer circulation, represented by the chart in

1680

PING-TUNG SHAWand SHENN-Yu CHAO 25oN

. ~-7,.

_,,.

-

0.5m/s

,

, t

~,.,

.~.

• ,

WI / /

.:.~ ' ~j, : : : : :. :. ~ / t

t/

..~--..~.'/ll

"."" :~,.



~. ~ , :. . ~ , : , : , ; , , ~ ,

...

/.

....'

ZZ/.

p~'i///_~_<~-_

100"1::

- -

_

-

,.,,

~. '

110*E

.

f

:~

120*E

Fig. 13. Surface velocityderived from ship drift (LEV1TUS,1982)in (a) August and (b) December. The scale of the velocityvector is (I.5 m s i.

August (Fig. 13a), shows strong eastward flow in the southern half of the South China Sea and northward flow in the Luzon Strait• The surface flow is weak off the coast of China. On the southern part of the Sunda Shelf the flow is northward. In the Gulf of Thailand, flow is northeastward in the southern portion and is southeastward in the northern reaches. In Wyrtki's chart, the currents are similar except along the coast of Vietnam. It is northward in June and August in Wyrtki's chart but is eastward in Levitus' chart. The eastward flow is probably due to E k m a n drift at the surface. Levitus' ship drift chart in D e c e m b e r is shown in Fig. 13b and is representative from N o v e m b e r to February. The major features include westward drift in the northern South China Sea, southwestward flow off the coast of Vietnam and on the Sunda Shelf, and offshore flow off Borneo• The surface current in the Gulf of Thailand is variable, westward in the upper reaches and southward near the mouth. WVRTKrS (1961) chart in D e c e m b e r is similar off the coast of Vietnam and in the northern South China Sea. However, his northward flow in the central basin from the coast of Borneo to about 15°N is missing in Levitus' chart. Examination of Wyrtki's charts shows that the current vectors are sparse; the discrepancy may be due to scant data. Before comparing the ship drift data with the results of the simulation, it should be pointed out that ship drift is affected by surface wind and E k m a n drift and may not adequately represent the interior flow below the E k m a n layer. As shown in the last section, the surface E k m a n flow and the interior flow may differ in directions, especially on the Sunda Shelf near the equator. Therefore, the flow field at 2.5 m shown in Fig. 3 is more suitable for comparison with the ship drift than at 50 m. The simulated summer circulation reproduces the weak flow off China, strong eastward flow in the southern basin

Surfacecirculationin the South China Sea

1681

and northward flow on the Sunda Shelf. The effect of Ekman drift is shown by the veering of the simulated currents to the right of the ship drift velocity. The most significant discrepancy occurs in the Luzon Strait. The strong northward flow is not present in the model because of the coarse grid separation, which cannot resolve the strong Kuroshio current. West of Luzon, ship drift indicates a northward flow, which is different from the simulated eastward current in the Ekman layer. However, a northward flow agrees with the simulated current at 50 m (Fig. 4b). In December, the simulated surface velocity agrees with the ship drift in the deep basin: strong westward flow in the Luzon Strait and in the deeper part of the basin and southeastward flow off the coast of south Vietnam and on the Sunda Shelf. Discrepancies occur mostly in shallow waters; i.e. in the Gulf of Tonkin and Gulf of Thailand. The offshore flow adjacent to Borneo and Palawan is also missing in the simulation. The latter may be produced by errors in wind estimates. Differences between shift drift and the simulated currents at 2.5 m are mainly due to the fact that PACANOWSKI and PHILANDERS (1981) parameterization of vertical mixing underestimates the Ekman layer depth and consequently strengthens the surface Ekman drift. The strong Ekman flow may overwhelm the interior flow at the surface and may result in the discrepancies. Overall, the discrepancies are too local to invalidate the simulation. The existence of a strong current east of Vietnam has been reported in literature. WYRTKI (1961) estimated a southward transport of 5 x 106 m 3 s -1 in winter and a northward transport of 3 × 106 m 3 s- 1in summer. The numerical simulation of POHLMANN (1987) also suggests the existence of a southward current in winter and a northward current in summer off Vietnam. Similar currents are present in the barotopic model of ZENG et al. (1989). Interestingly, the coastal current in the barotopic model separates from the coast in a pattern very similar to that in Fig. 4b in August. The similar behavior in a barotopic model suggests that the jet separation from the coast is produced by the wind field. Note that the axis of the maximum northeastward wind separates from the coast at ll°N and is directed northeastward toward the Luzon Strait in August. North of ll°N, winds west of this axis weaken markedly and become perpendicular to the coastline and isobaths because of the sudden change in the coastal orientation. Consequently, the August wind field can drive a northward coastal jet only south of ll°N. North of this latitude, the jet separates from the coast and diffuses northeastward. This mechanism apparently dominates whether the model ocean is barotopic or baroclinic. Both the interior flow and Ekman drift contribute to the inflow and outflow in the Luzon Strait. The total Ekman transport can be estimated from the meridional component of wind stress integrated across the Luzon Strait along 120.8°E. From June to August, the flow in the Ekman layer is toward the Pacific with a maximum transport of 0.28 x 106 s -1 in August. The transport is negligible compared to the total outflow of 2.5 x 106 m s- 1, which is the difference of the transports between boundaries B and C in Table 1. The Ekman transport is into the South China Sea for the rest of the year, mostly from November to January, with a maximum transport of 1.1 x 106 m s -1, or about one third of the inflow in the Luzon Strait (Table 1). Therefore, flow beneath the Ekman layer is the major contributor to inflow and outflow in the Luzon Strait. The simulated infow and outflow beneath the Ekman layer is consistent with the recent observations southwest of Taiwan. SHAW(1989) observed that remnant warm, salty water from the Kuroshio intrusion in the previous season can be found southwest of Taiwan in May but not in August. The intrusion water reaches a depth below 150 m. Later, SHAW (1991) analysed the temperature and salinity characteristics of water masses from

1682

PING-TUNG SHAWand SHENN-Yu CHAO

historical hydrographic station data in the northern South China Sea. He concluded that the Kuroshio intrusion occurs to a depth of 250 m between October and February. Using the same data set, SHAW (1992) suggested that the cold, fresh water in the South China Sea enters the region between the east coast of Taiwan and the Kuroshio front in summer. The model simulation clearly shows the development of a westward current on the continental slope south of China from September to February. The depth of the current agrees with that observed. By April, the current is weakened and is gradually replaced by the northeastward flow from the interior South China Sea. The northeastward flow transports cold, fresh water west of Luzon to the western edge of the Kuroshio front east of Taiwan. In addition to the velocity field, the rise and fall of sea level in the region west of the Luzon Strait in summer and winter, respectively, agree with the sea level distribution estimated by WYRTKI (1961). DISCUSSION AND CONCLUSIONS This paper presents results obtained from a three-dimensional model simulation of the circulation in the South China Sea. The model is forced by climatological distributions of wind stress, surface temperature and salinity as well as the transport estimates across open boundaries. The model ocean includes the Kuroshio current, which is excluded in the earlier model of POHLMANN (1987). The present approach seems to give a better agreement with observations than earlier model simulations. An unrealistic cyclonic gyre in the northern South China Sea appears in summer in Pohlmann's simulation, but it is no longer present in the present model. Although WYRTKI'S (1961) transport estimates are subject to uncertainties, the simulation based on these estimates seems to be able to account for the major features revealed by limited observations available to date. Specifying sea surface temperature and salinity at climatological values with one-degree resolution inevitably suppresses mesoscale dynamics associated with fronts. The model therefore produces only large scale features that are primarily wind-driven. The problem may not be serious, since the South China Sea circulation is dominated by the monsoons. The coarse-grid model, being strongly constrained by climatology, also tends to diffuse the Kuroshio as in all large-scale models with coarse resolution. At present, mesoscale processes are poorly known in the South China Sea. The present model is commensurate with the level of understanding borne out by observations. Numerical simulation in this paper reveals several key features in the South China Sea. First, a southward coastal jet along the western boundary appears in September when the northeast monsoon begins in the northern South China Sea. The current expands quickly southward and reaches the southern tip of Vietnam by October. At any given zonal sections off the coast of Vietnam, the southward flow initially appears as an undercurrent beneath the northward surface current, extending to the surface and replacing the surface northward flow in about a month. The current weakens in March and is replaced by a weak northward coastal current along the southern coast of Vietnam in May. The northward current does not follow the coastline north of 1 I°N but flows into the interior toward the northeast. The summer pattern is fully developed in August and changes to the winter condition in September. The second feature, which is confirmed by observations, is the inflow of the Kuroshio water from October to February. The current advects warm, salty water westward along the continental slope south of China. The simulated current agrees both in location and timing with the observed intrusion. Third, the outflow through the

Surface circulation in the South China Sea

1683

Luzon Strait is from the sea west of Luzon to the region east of Taiwan in summer. This current helps explain the existence of fresh water of South China Sea origin between the Kuroshio front and Taiwan. Fourth, there is a counter current below the surface layer on the Sunda Shelf, especially near the entrance of the Gulf of Thailand. The current is opposite to the direction of the surface flow and can be explained as flow directly forced by pressure gradients without Coriolis effects. Acknowledgements--We are indebted to D. A. Willey for her efforts in programming and data analysis, to J. M. Klinck and K. K. Liu for providing background materials that helped initiate this study, and to L. Salzillo for drafting the figures. Partial support of this research is from National Science Foundation Grant OCE-9101995 to North Carolina State University.

REFERENCES CSANADYG. Z. (1978) The arrested topographic wave. Journal of Physical Oceanography, 8, 47-62. HANEY R. L. (1991) On the pressure gradient force over steep topography in sigma-coordinate ocean models. Journal of Physical Oceanography, 21,610-619. HEELERMANS. and M. ROSENSTEIN(1983) Normal monthly wind stress over the world ocean with error estimates. Journal of Physical Oceanography, 13, 1093-1104. INOUEM. and S. E. WELSH (1993) Modeling seasonal variability in the wind-driven upper-layer circulation in the Indo-Pacific region. Journal of Physical Oceanography, 23, 1411-1436. LEvrrus S. (1982) Climatological atlas of the World Ocean. NOAA Professional Paper No. 13, U.S. Government Printing Office, Washington, DC, 173 pp. MELLOR G. L. (1993) User's guide for a three-dimensional, primitive equation, numerical ocean model. Atmospheric and Oceanic Sciences Program, Princeton University, 35 pp. NITANI H. (1972) Beginning of the Kuroshio. In: Kuroshio, H. STOMMELand K. YOSntDA, editors, University of Washington Press, Seattle, pp. 129-163. PACANOWSKIR. C. and S. G. H. PHILANDER(1981) Parameterization of vertical mixing in numerical models of tropical oceans. Journal of Physical Oceanography, 11, 1443-1451. POHLMANN Z. (1987) A three-dimensional circulation model of the South China Sea. In: Three-dimensional models of marine and estuarine dynamics, J. J. NIHOULand B. M. JAMART,editors, Elsevier, New York, pp. 245-268. SEMPER A. J. (1974) An oceanic general circulation model with bottom topography. Technical Report 9, Department of Meteorology, University of California, Los Angeles, 99 pp. SEr*ITNERA. J. and R. M. CHERVIN(1988) A simulation of global ocean circulation with resolved eddies. Journal of Geophysical Research, 93, 15,502-15,522, 15,767-15,775. SHAW P.-T. (1989) The intrusion of water masses into the sea southwest of Taiwan. Journal of Geophysical Research, 94, 18,213-18,226. 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. (1992) Circulation off the southeast coast of China. Reviews in Aquatic Sciences, 6, 1-28. WYRTK1K. (1961) Physical oceanography of the Southeast Asian waters. N A G A Report Vol. 2, Scientific Results of Marine Investigations of the South China Sea and the Gulf of Thailand, Scripps Institution of Oceanography, La Jolla, California, 195 pp. ZENG Q., R. LI, Z. Jl, Z. GAN and P. KE (1989) Calculation of the monthly mean circulation in the South China Sea. Scientia Atmospherica Sinica (in Chinese), 13, 127-168.