Content uploaded by Pengfei Xue
Author content
All content in this area was uploaded by Pengfei Xue
Content may be subject to copyright.
Physical mechanisms for the offshore detachment of the Changjiang
Diluted Water in the East China Sea
Changsheng Chen,
1,2
Pengfei Xue,
1
Pingxing Ding,
2
R. C. Beardsley,
3
Qichun Xu,
1
Xianmou Mao,
4
Guoping Gao,
1,5
Jiahua Qi,
1
Chunyan Li,
6
Huichan Lin,
7
Geoffrey Cowles,
1
and Maochong Shi
5
Received 27 October 2006; revised 2 April 2007; accepted 8 August 2007; published 2 February 2008.
[1]Physical mechanisms for the summertime offshore detachment of the Changjiang
Diluted Water (CDW) into the East China Sea are examined using the high-resolution,
unstructured-grid, Finite-Volume Coastal Ocean Model (FVCOM). The model results
suggest that isolated low salinity water lens detected west of Cheju Island can be formed
by (1) a large-scale adjustment of the flow field to the Changjiang discharge and (2) the
detachment of anticyclonic eddies as a result of baroclinic instability of the CDW front.
Adding the Changjiang discharge intensifies the clockwise vorticity of the subsurface
current (originating from the Taiwan Warm Current) flowing along the 50-m isobath and
thus drives the low-salinity water in the northern coastal area of the Changjiang mouth
offshore over a submerged plateau that extends toward Cheju Island. Given a model
horizontal resolution of less than 1.0 km, the CDW front becomes baroclinically unstable
and forms a chain of anticyclonic and cyclonic eddies. The offshore detachment of
anticyclonic eddies can carry the CDW offshore. This process is enhanced under
northward winds as a result of the spatially nonuniform interaction of wind-induced
Ekman flow and eddy-generated frontal density currents. Characteristics of the
model-predicted eddy field are consistent with previous theoretical studies of baroclinic
instability of buoyancy-driven coastal density currents and existing satellite imagery. The
plume stability is controlled by the horizontal Ekman number. In the Changjiang, this
number is much smaller than the criterion suggested by a theoretical analysis.
Citation: Chen, C., et al. (2008), Physical mechanisms for the offshore detachment of the Changjiang Diluted Water in the East
China Sea, J. Geophys. Res.,113 , C02002, doi:10.1029/2006JC003994.
1. Introduction
[2] In 1986, an international collaboration among Amer-
ican, Chinese, and Korean scientists made two regional
hydrographic surveys in the East China/Yellow Seas
(Figure 1) during 8 January to 1 February and 4 –20 July
[Chen et al., 1994]. The summer survey clearly showed an
isolated low-salinity lens in the region near 32.5°N, 124°E
between the Changjiang and Cheju Island (Figure 2). This
unique summertime distribution of salinity was detected in
earlier field measurements off the Changjiang [Beardsley
et al., 1985] and recent field data collected east of Cheju
Island [Lie et al., 2003]. It is commonly thought that the
low-salinity water lens originates from the Changjiang
Diluted Water (CDW). During the summer high river
discharge season, this water can be detached eastward
from the Changjiang plume as a result of either vortex
stretching [Beardsley et al., 1985] or summer monsoon-
driven offshore advection [Isobe et al., 2002; Chang and
Isobe, 2003].
[3] The eastward movement of CDW was also suggested
in the tracks of two surface drifters deployed in July 1986
[Beardsley et al., 1992]. Drifter 6986, which was launched
with a drogue at 10 m below the surface in the CDW plume
northeast to the mouth of the Changjiang, rotated anti-
cyclonically at a mean speed of 13 cm/s in the first 2 weeks
and then moved quickly toward Cheju Island at a speed of
14–20 cm/s (Figure 3). This drifter flowed through Cheju
Strait and eventually entered the Japan/East Sea (JES)
through the western channel of the Tsushima/Korea Strait.
The eastward flow in Cheju Strait was also evident in the
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, C02002, doi:10.1029/2006JC003994, 2008
1
Department of Fisheries Oceanography, Intercampus Graduate School
for Marine Science and Technology, University of Massachusetts, New
Bedford, Massachusetts, USA.
2
State Key Laboratory for Estuarine and Coastal Research, East China
Normal University, Shanghai, China.
3
Department of Physical Oceanography, Woods Hole Oceanographic
Institution, Woods Hole, Massachusetts, USA.
4
State Key Laboratory of Satellite Ocean Environment Dynamics,
Second Institution of Oceanography, State Oceanic Administration,
Hangzhou, China.
5
College of Physical and Environmental Oceanography, Ocean
University of China, Qingdao, China.
6
Department of Oceanography and Coastal Sciences, School of the
Coast and Environment, Louisiana State University, Baton Rouge,
Louisiana, USA.
7
Marine Extension Service, University of Georgia, Athens, Georgia,
USA.
Copyright 2008 by the American Geophysical Union.
0148-0227/08/2006JC003994
C02002 1of17
trajectory of drifter 6988 deployed with a drogue at 10 m
below the surface west of Cheju Island in July (Figure 3).
That drifter moved eastward at a speed of about 20–28 cm/s
through Cheju Strait and then entered Tsushima/Korea
Strait.
[4] The past and recent hydrographic and drifter measure-
ments raise a fundamental question regarding the physical
mechanisms for the formation of the isolated low-salinity
lens west of Cheju Island. It is clear that CDW can be
advected eastward by the southerly dominated summer
monsoon wind but unclear how the wind interacts with
the buoyant CDW plume to cause the offshore detachment
of CDW in a large shallow region bounded by 50-m isobath
between the Changjiang mouth and Cheju Island. Can
CDW detach from the main body of the plume without
wind forcing? If so, what other process is important?
[5] To address these questions, we must resolve the
spatial/temporal structure of the Changjiang plume in the
East China Sea (ECS). The Changjiang river mouth has
complex coastal geometry and abruptly varying bathymetry
(Figure 4). It is divided into southern and northern branches
by Chongming Island. Two smaller islands are located in
the mouth of the southern branch, which divides this branch
into two channels. At the mouth of the river, the bottom
topography features four ‘‘deep’’ passages connected to the
southern branch. There are also two relatively deep passages
linked to the northern branch. In the summer flood season,
the Changjiang freshwater water splits into two streams at
the northern tip of Chongming Island which flow out onto
the inner shelf of the ECS through southern and northern
branches [Kong et al., 2004]. In the winter dry season, the
Changjiang freshwater water flows into the ECS primarily
through the southern branch and a saltwater return flow is
frequently observed in the northern branch [He et al., 2006].
In view of the plume dynamics, the Changjiang discharge is
like a multiple river/estuarine system in which the offshore
Figure 1. Bathymetry (in meters) and locations of the East China Sea, the Yellow Sea, the Bohai Sea,
and the Japan/East Sea.
C02002 CHEN ET AL.: RIVER PLUME BAROCLINIC INSTABILITY
2of17
C02002
detachment of plume water may be caused by eddy forma-
tion due to baroclinic instability of the plume [Qiu et al.,
1988] or by nonlinear interaction between upwelling-favor-
able wind and buoyancy flow at the low-salinity front [Chen
et al., 1999; Chen, 2000].
[6] A few modeling studies have explored the offshore
eastward transport of CDW in the ECS [e.g., Isobe et al.,
2002; Chang and Isobe, 2003]. These models captured
wind-driven eastward transport in the ECS, but realistic
plume dynamics were restricted due to difficulties in re-
solving the complex bathymetry of the Changjiang rive
mouth region. In previous structured-grid, finite-difference
models of the ECS, the Changjiang discharges were either
specified as a point source or an idealized short channel
flow. Ignoring the actual along-river surface pressure gra-
dient in the Changjiang can cause an overestimation of the
offshore advection and failure to resolve the nature of
multiple outflows. The horizontal resolution used in these
models was too coarse to resolve the abruptly varying
bottom topography in the river channels and thus the
models were unable to simulate the mesoscale structure of
the Changjiang plume over the inner shelf of the ECS. As a
result, the buoyancy contribution to the water transport,
particularly for the offshore eastward detachment of CDW,
was significantly underestimated.
[7] A research team from the University of Massachu-
setts-Dartmouth (UMASS-D), East China Normal Univer-
sity (ECNU), and Woods Hole Oceanographic Institution
(WHOI) led by C. Chen (UMASS-D), P. Ding (ECNU), and
R. C. Beardsley (WHOI) have applied the unstructured-grid,
Finite-Volume Coastal Ocean Model (FVCOM) to study the
physical mechanisms for the offshore detachment of CDW
in the ECS. FVCOM was originally developed by Chen et
al. [2003] and improved by a team of UMASS-D and
WHOI researchers [Chen et al., 2004, 2006a, 2006b]. The
unstructured triangular grid used in FVCOM allows us to
Figure 2. The distribution of salinity at a depth of 4 m
below the surface measured on a large-scale survey in July
1986. The dots are the locations of the CTD stations and
dashed line is the 200-m isobath. This figure was digitized
from Chen et al. [1994].
Figure 3. Trajectories of drifters 6986 and 6988 with drogues at 5 m below the surface. The drifters
were released in summer 1986. Dots indicate the drifter locations every 5 d from launch. The data used in
this figure were digitized from Beardsley et al. [1992, Figure 4].
C02002 CHEN ET AL.: RIVER PLUME BAROCLINIC INSTABILITY
3of17
C02002
include realistic coastline and bathymetry of the Changjiang
and place higher horizontal resolution in the plume area.
The finite-volume approach used in FVCOM guarantees
local mass, momentum, heat, and tracer conservation,
making it possible to trace the spread of CDW during a
long-term integration.
[8] This paper describes the initial development of the
ECS-FVCOM model system and its application to examine
the role of the Changjiang discharge in the summertime
adjustment of the ECS circulation and the contribution of
plume instability to the offshore detachment of CDW. A
series of idealized process-oriented numerical experiments
were conducted and the model results compared with buoy-
ant plume theory and remote-sensing data to develop
physical insight into the low-salinity plume dynamics in
the ECS.
[9] The remaining part of this paper is organized as
follows. ECS-FVCOM and the design of numerical experi-
ments are described in section 2. The process-oriented
numerical model results under different physical conditions
are presented in section 3. Section 4 presents a discussion
and the conclusions summarized in section 5.
2. ECS-FVCOM and Design of Numerical
Experiments
[10] ECS-FVCOM is a spherical coordinate version of the
unstructured-grid finite-volume, three-dimensional (3-D)
primitive equation coastal ocean model developed originally
by Chen et al. [2003] and upgraded by the UMASS-D/
WHOI model development team [Chen et al. 2006a,
2006c]. In common with other free-surface coastal models,
FVCOM uses the modified Mellor and Yamada level 2.5
(MY-2.5) and Smagorinsky turbulent closure schemes as
default setups for vertical and horizontal mixing, respec-
tively [Mellor and Yamada, 1982; Galperin et al., 1988;
Smagorinsky, 1963]. Similar to popular structured-grid
models like the Princeton Ocean Model (POM) and Re-
gional Ocean Model system (ROMs), FVCOM is numeri-
cally solved using a split-mode method. The external mode
is composed of vertically integrated transport equations in
which the water elevation is solved explicitly using a shorter
time step constrained by the ratio of the horizontal resolu-
tion to the phase speed of the long surface gravity wave.
The internal mode consists of fully 3-D governing equations
and is solved using a longer time step constrained by the
phase speed of the lowest mode internal wave. Linkage
between external and internal modes is through the water
elevation, with mode adjustments based on the vertically
integrated volume transport at each internal time step.
Unlike structured-grid finite-difference and unstructured-
grid finite-element models, FVCOM solves the flux form
of the governing equations in unstructured triangular vol-
umes by a second-order accurate discrete flux scheme. The
finite-volume approach used in FVCOM takes advantage of
finite-element methods in geometric flexibility and finite-
difference methods in computational efficiency and pro-
vides an accurate representation of mass, momentum, heat,
and salt conservation. (See http://fvcom.smast.umassd.edu/
for a full description of FVCOM, recent additions to its
capabilities, and related publications.)
Figure 4. Three-dimensional bathymetry of the Changjiang and its adjacent shelf region.
C02002 CHEN ET AL.: RIVER PLUME BAROCLINIC INSTABILITY
4of17
C02002
[11] The computational domain of ECS-FVCOM covers
the entire ECS, Yellow Sea (YS), Bohai Sea (BS), and
Japan/East Sea (JES) (Figure 5). The domain is bounded by
four open boundaries: Taiwan Strait (A), a line starting at
the eastern coast of Taiwan (B), running off the slope of the
ECS and then crossing onto the eastern shelf of Japan (C),
Tsugaru Strait (D), and Soya Strait (E). The unstructured
triangular grid has a horizontal resolution varying from 0.1
to 0.5 km in the Changjiang, 0.5–1.5 km in the river mouth
area and along the coast, 3-km in the path of the Kuroshio,
and 10–15 km in the interior of the ECS/YS/BS/JES and
Pacific Ocean. An enlarged view of the triangular grid in the
Changjiang area is shown in Figure 6. In the vertical, the
domain is divided into 30 uniform s-layers, which corre-
sponded to a vertical resolution of 1.7 m or less in the
Changjiang and coastal region where the water depth is
shallower than 50 m. To reduce the model spin-up time, the
model bathymetry off the slope in the western Pacific Ocean
is cut at 800 m, a roughly e-folding depth of the Kuroshio
Current.
[12] In this study, ECS-FVCOM is forced by (1) the
Changjiang freshwater discharge at the upstream end of
the Changjiang model domain and (2) a constant uniform
wind stress. The model is initialized using 10-km resolution
3-D August climatology hydrographic fields and a specified
water transport of the Kuroshio at the open boundary. The
hydrographic fields were constructed using 10-km box
averaging of historical water temperature and salinity data
collected in this region over the last 40 a. The inflow
transports are (1) 2.1 Sv for the Taiwan Warm Current
(TWC) through A and (2) 27.5 Sv for the Kuroshio east of
Taiwan through B. These are balanced by outflow transports
of (1) 26.9 Sv for the Kuroshio southeast of Japan (C), (2)
1.6 Sv through the Tsugaru Strait (D), and (3) 1.1 Sv through
the Soya Strait (E). The outflow transport on C was adjusted
to balance an additional inflow from the Changjiiang when
the river discharge was added. The inflow and outflow
transport values used here are based on the summertime
climatology conditions. The Kuroshio transport was esti-
mated from current profiles in the upper 800 m, which is
1.0–2.0 Sv smaller than the total Kuroshio transport. Verti-
cal profiles of the inflow and outflow Kuroshio transports are
specified based on the observed structure of the Kuroshio
from our previous ADCP measurements [Chen et al. 1992].
[13] Numerical experiments were conducted in three
steps: day 0–30 spin-up with the climatology fields; day
31–60 spin-up with the Changjiang discharge; and day 61 –
90 run with (1) river discharge only (case A) and (2) river
discharge and wind (case B). In particular, the model was
first spun up with the August climatology hydrographic
fields and fixed inflow and outflow transports for 60 model
days. Since the Kuroshio and shelf circulation reached
quasi-steady state after 30 model days, the river discharge
was started on model day 31 by adding a freshwater
discharge of 60,000 m
3
/s (a typical summertime value) at
the upstream end grid of the Changjiang domain. The
Changjiang plume was established by model day 60. For
the river discharge only case A, the model was run for an
additional 30 d. For case B with river and wind forcing, the
Figure 5. Unstructured grid of the East China Sea
FVCOM. Labels A, B, C, D, and E indicate sections of
the open boundary where water transports were specified.
Figure 6. An enlarged view of the unstructured grid in the
Changjiang area.
C02002 CHEN ET AL.: RIVER PLUME BAROCLINIC INSTABILITY
5of17
C02002
model was run for an additional 30 d followed by 7 d with
no wind (to simulate a wind relaxation).
3. Model Results
3.1. Adjustment of the ECS Circulation to the
Changjiang Discharge
[14] By model day 30, the Kuroshio and ECS, YS, BS,
and JES circulation reach a quasi-steady state (Figure 7).
The Kuroshio, with a mean velocity of 100 cm/s, forms
over the continental slope, entering the ECS east of Taiwan,
flows along the 200-m isobath on the slope of the ECS,
turns southeastward to go through the Tokara Strait and then
moves northeastward along the Japanese coast to leave the
computational domain. The TWC flows northeastward
parallel to the 50-m isobath, enters a submerged river valley
off the Changjiang and then turns anticyclonically along the
50-isobath to move onto the interior of the ECS continental
shelf. The Yellow Sea Coastal Current (YSCC), Yellow Sea
Warm Current (YSWC), Tsushima Warm Current, and
Korean Coastal Current (KCC) are all resolved in this
climatology-based flow field. The simulated circulation
patterns agree well with previous field measurements sum-
marized by Qiu and Imasato [1990], Lie et al. [1998], and
Ichikawa and Beardsley [2002].
[15] The addition of the Changjiang discharge changes
the circulation pattern in the shelf region between the river
mouth and the Tsushima/Korea Strait (Figure 8). CDW
flows out from the Changjiang from both southern and
northern branchs, turning anticyclonically in the river mouth
area shallower than 50 m before flowing southward along
the coast. Correspondingly, the TWC, YSCC, YSWC, and
Tsushima Warm Current are all intensified, even though the
path and intensity of the Kuroshio change little. As a result,
the northward transport of saline water is enhanced in the
submerged river valley off the Changjiang and west of
Cheju Island, and in turn a large portion of the low-salinity
water (30–32 PSU) is advected onto the shallow plateau
bounded by the 50-m isobath by the enhanced YSCC and
flows through the Cheju Strait and then Tsushima/Korea
Strait as part of an intensified Tsushima Current. The local
adjustment of the circulation off the Changjiang can be
viewed more clearly in Figure 9, which shows that the less
saline water patch detected over the submerged plateau
between the Changjiang and Cheju Island can be advected
from the western coast of the YS as a result of the current
adjustment to the Changjiang discharge.
[16] The intensification of the TWC is evident in a
comparison of the flow field at a depth of 20 m below the
surface between the cases without and with the Changjiang
discharge (Figure 10). With no river outflow, the relatively
weak northward subsurface current has a magnitude of
about 3–5 cm/s in and around the submerged river valley.
After the CDW plume forms in the upper water layer, the
northward current at the depth of 20 m is significantly larger
(10 cm/s or larger) and drives a relatively strong offshore
flow along the 50-m isobath near the tip of this valley. This
subsurface offshore flow can act like an ‘‘interior stress’’ to
push the near-surface less-saline water eastward onto the
plateau area east of the submerge valley.
Figure 7. Distribution of near-surface current and salinity
in the East China Sea on day 30 for the model run started
with August climatology temperature and salinity fields. To
make the vector fields more viewable, current vectors were
selected in a search radius of 20 km.
Figure 8. Distribution of near-surface current and salinity
in the East China Sea and Changjiang estuary on day 30
after the Changjiang discharge was added. The river
discharge was specified as a constant 60,000 m
3
/s and
added into the model after the current field (with the
climatology hydrography) reached a quasi-steady state
(Figure 7). The current vectors were selected in a search
radius of 20 km.
C02002 CHEN ET AL.: RIVER PLUME BAROCLINIC INSTABILITY
6of17
C02002
[17] This simple adjustment experiment suggests that the
summertime circulation in the ECS and YS as well as in
the Tsushima/Korea Strait is significantly influenced by the
Changjiang discharge. The summertime intensification of
the TWC, YSWC, YSCC, and Tsushima Warm Current can
be in part a result of the broad-scale adjustment of the
circulation to an increase in the Changjiang discharge. As a
result of this adjustment, the less-saline (CDW) water can
be carried to the submerged plateau area between the
Changjiang and Cheju Island without any wind forcing.
3.2. Baroclinic Instability of the CDW Plume
[18] The model-predicted summertime CDW plume has a
complex salinity and current pattern (Figure 9). After 30 d
of Changjiang discharge, the CDW is vertically well-mixed
in the shallow area (depth < 20 m) and forms a surface-
intensified salinity front or transition zone between the
CDW core and the shelf water around the 50-m isobath.
This is a typical density front which is baroclinically
unstable and creates anticyclonic and cyclonic eddies. With
a continuous supply of CDW, anticyclonic eddies grow
quickly and extend offshore. In addition to the main
southward current, the eddy-induced unstable waves prop-
agate northward at a speed of 5 cm/s (Figure 11). After
60 d of Changjiang discharge, five well-defined anticy-
clonic eddies are found at the edge of the front, with
accompanying cyclonic eddies (Figure 11). The size of an
anticyclonic eddy (measured by the diameter) varies in a
range of 30–60 km, which is about three times larger than
a cyclonic eddy.
[19] A significant amount of CDW is carried eastward
toward Cheju Island as the anticyclonic eddies first elongate
and then separate from the CDW plume. This detachment
occurs when an opposite current shear forms between each
cyclonic eddy and the main body of the CDW plume. This
experiment demonstrates that the eastward advection of
these detached baroclinic eddies can form the isolated
low-salinity lens observed between the Changjiang and
Cheju Island.
Figure 9. An enlarged view of the near-surface CDW
plume current and salinity off the Changjiang at the same
time as that in Figure 8. The current vectors drawn in this
figure were selected in a search radius of 5 km.
Figure 10. Currents at a depth of 20 m below the surface around the 50-m isobath off the Changjiang at
the end of day 30 for the climatology spin-up model run (left) and at the end of day 30 after the
Changjiang discharge was added (right). The current vectors were selected in a search radius of 5 km.
C02002 CHEN ET AL.: RIVER PLUME BAROCLINIC INSTABILITY
7of17
C02002
[20] The CDW plume instability found in our numerical
experiment is consistent with the previous theoretical study
of baroclinic instability of a buoyancy-driven coastal den-
sity current by Qiu et al. [1988]. They considered an
idealized 200-m deep shelf with a straight coastline and
found that when the horizontal Ekman number was smaller
than a critical value, the density front generated by a
sequence of uniformly distributed freshwater sources along
the coast became baroclinically unstable. The salinity front
first exhibited a wave pattern (with a wavelength of 100 km
in the along-shelf direction) which then evolved into anticy-
clonic eddies in the wave crests and cyclonic eddies in the
wave troughs. The unstable wave crests tended to propagate
upstream (in the opposite direction to the main stream of the
density current) and low-salinity water was transported to the
outer shelf by the detached eddies. In this idealized case,
baroclinic instability of the front occurs when
Eh¼Ahf
g0Qe
ðÞ
2=3<Ehc 0:34 0:57 ð1Þ
where E
h
is the horizontal Ekman number, E
hc
is the critical
Ekman number, Q
e
is the river discharge per unit length
along the coast, A
h
is the horizontal eddy viscosity, fis the
Coriolis parameter, and g
0
is the reduced gravity constant.
[21] In our model simulation, CDW flows onto the shelf
through multiple passages with an along-shelf extend of
80 km. With
Qe60000 m3=s
8104m;
f10
4
,
g0¼r2r1
ro
g0:015 9:8m
2=s¼0:15 m=s2;
and A
h
20 m
2
/s given by the Smagorinsky turbulent
closure scheme based on the horizontal resolution within the
front, the horizontal Ekman number is
Eh0:086 Ehc:
This indicates that the model Changjiang plume is
baroclinically unstable as suggested by Qiu et al. [1988].
[22] The model-predicted multiple eddy feature of the
CDW plume can be seen in the recent 1-km resolution
MODIS (Moderate Resolution Imaging Spectroradiometer)
sea surface temperature (SST) image in the ECS/Changjiang
received at 1350:26 GMT, 14 September 2005 (Figure 12).
This image shows four major outflow passages of CDW and
a complex eddy field at the CWD front and in Hangzhou
Bay. This observed eddy field is similar in shape and size to
that predicted by ECS-FVCOM, supporting the reality of
the model-predicted CDW plume in the ECS.
3.3. Effects of Wind Forcing
[23] Wind forcing was added 30 d after initiation of the
Changjiang discharge.We ran the model with various wind
directions (southerly, southwesterly, southeasterly, westerly,
and easterly) and speeds (3, 5, 8, and 10 m/s) typically of
those observed during the summer monsoon season in the
ECS. A southerly or easterly wind slows the southward
along-coast transport in the CDW plume. Owing to the large
Figure 11. Near-surface CDW plume currents and salinity
off the Changjiang at day 60 after the Changjiang discharge
was added for the case with only river discharge. The
current vectors were drawn with a search radius of 5 km.
Figure 12. MODIS-derived sea surface temperature image
at 1350:26 GMT 14 September 2005 in the western East
China Sea area off the Changjiang mouth.
C02002 CHEN ET AL.: RIVER PLUME BAROCLINIC INSTABILITY
8of17
C02002
pressure gradient built by the CDW outflow, however, the
summer winds have little influence on the CDW current
inside the Changjiang. The interaction between wind and
CDW plume is strongest near the edge of the salinity front
where unstable waves occur. Here the wind tends to speed
up and enhance the offshore detachment of CDW. Examples
for the southerly wind are described below.
[24] In the case with a constant southerly wind, the wind-
induced offshore Ekman transport inside and at front of the
CDW off the Changjiang is uniform in space. However,
owing to the presence of the frontal waves and eddies, the
interaction of wind- and buoyancy-driven currents in the
salinity front are spatially nonuniform. For example, in an
anticyclonic eddy, the southerly wind intensifies the off-
shore flow and weakens the onshore flow, increasing an
asymmetry of the swirl current and hence speeding up eddy
attachment. This is clearly shown seen in the distribution of
salinity and currents at day 30 after a southerly wind of 3 m/s
is initiated (Figure 13a).
[25] After the wind ceased, the current quickly adjusts to
the salinity field and surface elevation within a local inertial
timescale. The southward current is restored inside the
CDW plume to the prewind condition, while a significant
amount of CDW is transported offshore as a result of the
offshore detachment of anticyclonic eddies (Figure 13b).
[26] The spatial distribution of CDW varies with wind
direction and speed while the interaction process between
wind- and buoyancy-driven currents does not change. For
the same wind speed of 3 m/s, a southeasterly wind tends to
push more CDW northeastward, but the distribution of
CDW after the wind ceases is very similar to that found
in the case with the southerly wind (Figure 14, left). The
southwesterly wind enhances the southeastward transport of
CDW and resulting eddy patterns shift southward after the
wind relaxation (Figure 14, right). When the southerly
wind speed is increased to 5 m/s, the wind-induced Ekman
transport causes a larger offshore transport of CDW off the
Changjiang. After the wind ceases, less-saline core eddies
form east of the Changjiang as a result of the adjustment
between the current and salinity fields (Figure 15). Clearly,
the stronger southerly wind tends to enhance the offshore
transport of CDW as suggested in previous finite-difference
model results [Chang and Isobe, 2003]. The primary
difference here is that in ECS-FVCOM, the offshore
transport of CDW is a result of the nonlinear interaction
between the wind and unstable eddy fields at the front,
while in previous models, the offshore transport is caused
by the wind-driven Ekman transport applied to the core
body of CDW.
[27]Chen [2000] examined the formation of an isolated
low-salinity lens and its offshore detachment over the inner
shelf of the South Atlantic Bight (SAB). He found that the
detachment process happens through two steps. In the first
step the spatially nonuniform response of current to the
upwelling-favorable wind causes a wavelike frontal shape at
the outer edge of the frontal zone. Then the isolated low-
salinity lenses form at the crest, when water on the shore-
ward side of the crest is displaced by relatively high-salinity
water advected from the upstream trough south of the crest
and diffused upward from deeper off-shelf. The interaction
Figure 13. Near-surface CDW plume currents and salinity off the Changjiang on day 30 after a
southerly wind of 3 m/s blew (left) and on day 7 after the wind relaxed (right). The wind was added at the
end of the 30th day after the Changjiang discharge was added. The wind relaxation (wind speed set to
zero) was made at the end of the 30th day after the wind began to blow. Vectors were plotted in the same
spatial selection as that shown in Figure 10.
C02002 CHEN ET AL.: RIVER PLUME BAROCLINIC INSTABILITY
9of17
C02002
of wind- and buoyancy-driven currents described here is
then similar to the second step found in the SAB, except
that the eddy field forms at the CDW front due to baroclinic
instability.
[28] Our experiments suggest that when the CDW front
becomes unstable, CDW can be transported eastward along
the shallow ridge between the Changjiang and Cheju Island
as a result of the offshore detachment of anticyclonic eddies.
This offshore detachment process can be enhanced signif-
icantly by southerly winds, with transport increasing with
wind speed. To provide realistic simulation of this offshore
transport of CDW, a model needs to resolve the baroclinic
instability of the CDW front. This process is missing in
previous modeling experiments due to restrictions of hori-
zontal resolution and poor resolution of the ‘‘deep’’ water
passages inside and outside of the Changjiang.
3.4. Tracer and Lagrangian Particle Tracking
Experiments
[29] To examine the offshore transport of CDW, we
conducted both tracer concentration and Lagrangian particle
tracking experiments under different physical conditions.
Model dye with a concentration of 1.0 was injected
throughout the water column on the cross-river mouth
transect shown in Figure 16 at the end of 30 model days
after initiation of the Changjiang discharge. The dye was
then traced for an additional 30 days under conditions with
(1) only Changjiang discharge and (2) Changjiang discharge
plus a 5 m/s southerly wind, respectively. In the first case,
most of the dye appears in the unstable eddy zone along the
CDW front and in Hangzhou Bay, while some has been
carried eastward of the CDW front along the Changjiang
plateau (Figure 16a). In the second case, most of the dye is
advected and diffused further offshore to the east and
northeast over the Changjiang plateau (Figure 16b). After
wind relaxation after 30 d of wind forcing, much of the dye
stays over the plateau within the 50-m isobath.
[30] In the particle-tracking experiments, neutrally buoy-
ant particles were released at surface, local middepth, and
near-bottom on the same transect as the dye release oc-
curred. On this transect, middepth ranges from 5 to 8 m, and
near-bottom depths range from 10 to 16 m. At each level, 50
particles were released uniformly along the transect on the
end of 30 d of Changjiang discharge and then tracked for an
additional 30 d in the case with only Changjiang discharge
and for an additional 60 d in the case with both Changjiang
discharge and a 5 m/s southerly wind. In the first case,
particles released at the three levels show significantly
different trajectories, an indicator of the inhomogeneous
flow field in the vertical. The surface particle distribution is
similar to the distribution of dye at the surface. The surface
particles follow two main paths, one moving offshore
toward the CDW front and the other moving southward
along the coast and into Hangzhou Bay (Figure 17a). The
particles entering the frontal zone move either northward
or southward following the complex swirl eddy field.
Particles released at middepth also followed two primary
paths, one flowing first northward along the coast and then
turning clockwise back toward the south in frontal zone
and the other directly eastward into the CDW frontal zone
(Figure 17b). Some of these particles are ejected from the
Figure 14. Near-surface CDW plume currents and salinity off the Changjiang on day 7 after the wind
relaxed for the cases with southeasterly (left) and southwesterly (right) winds of 3 m/s. The wind ceased
at the end of the 30th day after the wind was added. Vectors were plotted in the same spatial selection as
that shown in Figure 10.
C02002 CHEN ET AL.: RIVER PLUME BAROCLINIC INSTABILITY
10 of 17
C02002
Figure 15. Near-surface CDW plume currents and salinity off the Changjiang on day 30 after a
southerly wind of 5 m/s blew (left) and on day 7 after the wind relaxed (right). The wind was added at the
end of the 30th day after the Changjiang discharge was added. The wind ceased at the end of the 30th day
after the wind blew. Vectors were plotted in the same spatial selection as that shown in Figure 10.
Figure 16. Distributions of model dye concentration at the end of day 60 after the Changjiang discharge
was added (left) and at the end of day 7 after the wind relaxed (right). In the wind case, a southerly wind
of 5 m/s was added into the model run at the end of day 30 after the Changjiang discharge was started.
The wind blew first for 30 days and then set to zero and the model run for an additional 7 d.
C02002 CHEN ET AL.: RIVER PLUME BAROCLINIC INSTABILITY
11 of 17
C02002
frontal eddies and move further east and south. Particles
released near-bottom follow similar paths as those released
at middepth (Figure 17c). Very few particles from these
two levels enter Hangzhou Bay.
[31] The particle trajectories differ significantly in the
case with a southerly wind of 5 m/s (Figure 18). At the
surface, the wind initially causes all surface particles to
move eastward offshore in the well-mixed region of the
CDW. After these particles enter the CDW frontal zone,
some are advected northward along the front while others
cross the front on anticyclonic eddy loops and then move
eastward toward Cheju Island (Figure 18a). The particles
released at middepth show some similarity with the surface
particles, with more particles drifting toward Cheju Island
(Figure 18b). Even more of the particles released near-
bottom escape the frontal zone and move toward Cheju
Island (Figure 18c). In this case with a 5 m/s southerly
wind, particles take about 40–45 d to travel from the mouth
of the Changjiang to Cheju Island. Once entering Cheju
Strait, they quickly accelerate and move into the Japan/East
Sea through the Tsushima/Korea Strait.
Figure 17. Trajectories of fluid particles released at (a) the
surface, (b) local middepth, and (c) near-bottom on the
transect off the Changjiang for the case with only
Changjiang discharge. The particles were released at the
end of day 30 after the Changjiang discharge was added and
then traced for additional 30 d.
Figure 18. Trajectories of fluid particles released at (a) the
surface, (b) local middepth, and (c) near-bottom on the
transect off the Changjiang for the case with the Changjiang
discharge plus a 5 m/s southerly wind. The particles were
released at the end of day 30 after the Changjiang discharge
was added and then traced for an additional 60 d.
C02002 CHEN ET AL.: RIVER PLUME BAROCLINIC INSTABILITY
12 of 17
C02002
[32] The model particle trajectories with southerly wind
forcing agree the trajectories of drifters observed in Sep-
tember 1986 [Beardsley et al., 1992]. The anticyclonic loop
shown in the observed drifter track indicates an anticyclonic
eddy at the CDW front, which is well resolved in both our
dye and particle experiments using ECS-FVCOM. In par-
ticular, the particle-tracking experiments (Figures 18b
and 18c) show that a model drifter drogued at 10 m and
released off the Changjiang can be carried northeastward
into the Cheju Strait and JES as observed in fall 1986.
[33] Note that with a southerly wind, the dye and par-
ticles, which were released on the same transect, follow
completely different pathways. Dye does not spread to
Cheju Strait while many particles arrive there in about
40–45 d. Chen et al. [2007] examined physical mechanisms
for cross-frontal transport of a neutral tracer on Georges
Bank. They found that the movement of the center of the
dye concentration (or ‘‘patch’’) is driven by the ensemble
horizontal velocity and the vertical velocity shear-related
concentration flux of the dye patch. In an inhomogeneous
flow field the movement of the dye patch can differ
significantly from trajectories of individual particles. In
our case here, it is inappropriate to study the cross-frontal
water exchange based on trajectories of a limited number of
particles. During summer, circulation in the ECS is quite
inhomogeneous in the vertical. In the CDW frontal zone, the
current can be considered as a two-layer system: southward
flow of CDW and northward flow of TWC water. In the
CDW frontal zone, the formation of an anticyclonic eddy in
the upper layer is compensated with a cyclonic eddy near the
bottom. Fluid particles release in the upper layer mainly
represent the advective movement of a water parcel in the
Ekman layer with little interaction with the surrounding
water, while the dye movement includes both horizontal and
vertical diffusion. Caution is needed when interpreting
particle trajectories for water exchange in the ECS.
4. Discussion
[34] The high-resolution ECS-FVCOM simulations re-
veal that the CDW plume is baroclinically unstable and
that CDW can be advected offshore as a result of the
detachment of anticyclonic eddies that form in the CDW
plume frontal zone. The interaction between wind- and
buoyancy-driven flows at the front tends to enhance the
offshore CDW transport. A momentum balance analysis
shows that the CDW currents off the Changjiang are quasi-
geostrophic. In both xand ydirections the Coriolis term is
balanced to first-order by the difference of barotropic
(surface elevation) and baroclinic (density) gradient forces
(Figures 19 and 20). This quasi-geostrophic condition is still
valid when a uniform southerly wind is added and then
turned off. Inside the Changjiang, the presence of wind
forcing does not change significantly the along-river distri-
bution of surface elevation: highest at the upstream end of
the river and downward toward the mouth (Figure 21). Thus
the river flow is controlled by the surface elevation gradient
force with little influence from wind or the Coriolis force.
This surface elevation gradient plays a key role in stabiliz-
ing the CDW flow inside the river and near the coast. Off
the Changjiang, the Coriolis force becomes important, and
the CDW moves with a balance between the surface
elevation gradient and Coriolis forces in the vertically
well-mixed region and with a balance between the surface
elevation gradient, density gradient, and Coriolis forces in
the stratified region. Under such a quasi-geostrophic con-
dition, the evolution of unstable waves and formation of
eddies in the CWD front satisfy instability theory [Pedlosky,
1979; Qiu et al., 1988]. This theory shows that the insta-
bility develops through linear and nonlinear stages. The
eddy kinetic energy grows exponentially as a result of
transfer from potential energy and mean kinetic energy in
the linear stage and decreases gradually as the instability
saturates in the nonlinear stage. These two processes are
clearly evident in an idealized experiment with multiple
river discharges made by Qiu et al. [1988], which should be
also valid for the CDW plume.
[35] Both previous theoretical and current numerical
experiments show that the baroclinic instability of the
CDW plume occurs for small horizontal Ekman number
that is directly proportional to horizontal diffusion and
inversely proportional to river discharge rate. This means
that the plume is more unstable during the summer flood
season than during the winter dry season. Horizontal diffu-
sion in FVCOM is calculated by the Smagorinsky param-
eterization method defined as
Ah¼0:5CHWz
Prffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
@u
@x
2
þ0:5@v
@xþ@u
@y
2
þ@v
@y
2
sð2Þ
where C
H
is a constant parameter and W
z
is the area of the
individual tracer control element and P
r
is the Prandtl
number. The value of A
h
is thus proportional to the area of
the individual tracer control element and the gradient of
horizontal velocities: decreasing as the grid size or
horizontal velocity gradient becomes smaller. In general,
A
h
is a function of time and space related to the horizontal
current shear and bathymetry. For the high-resolution ECS-
FVCOM grid, A
h
20 m
2
/s in the CDW plume. This value
is much smaller than values used in previous ECS model
experiments [Choi, 1980; Yanagi and Takahashi, 1993;
Isobe et al., 2002]. In early summer of 1999, a dye
experiment was conducted in the frontal zone on the
southern flank of Georges Bank to examine cross-frontal
water exchange processes. By tracing the dye patch over
time, Houghton [2002] used a Fickian model and the dye
patch spreading to estimate the cross-frontal horizontal
diffusion coefficient and obtained a time-averaged value of
A
h
in a range of 18 ± 4.5 m
2
/s. Chen et al. [2007] simulated
the dye experiment by running FVCOM with different
horizontal resolution on Georges Bank and found that the
model converged to a value of A
h
of 20 m
2
/s when
horizontal resolution was 500 m. Although there has been
no in situ measurements made to estimate A
h
in the CDW to
our knowledge, we believe that its value should be in the
same order of magnitude as that inferred in the frontal zone
on the southern flank of Georges Bank because of similar
frontal dynamics. Similarity of the eddy fields found in the
high-resolution MODIS SST imagery and the ECS-
FVCOM simulations implies that the order of the horizontal
diffusion coefficient estimated by the Smagorinsky method
is robust. We encourage direct in situ measurements in the
ECS to improve knowledge of horizontal diffusivity.
C02002 CHEN ET AL.: RIVER PLUME BAROCLINIC INSTABILITY
13 of 17
C02002
Figure 19. Distributions of the major terms in the vertically averaged x-momentum equation on day 60
for the case with only Changjiang discharge (left column), on day 30 after a southerly wind of 3 m/s was
added (center column), and on day 7 after the wind relaxed (right column). The top shows the Coriois
term, center shows the surface elevation gradient force, and the bottom shows the baroclinic pressure
gradient force term. Units are m
2
/s
2
.
C02002 CHEN ET AL.: RIVER PLUME BAROCLINIC INSTABILITY
14 of 17
C02002
Figure 20. Distributions of the major terms in the vertically averaged y-momentum equation (left) on
day 60 for the case with only Changjiang discharge, (middle) on day 30 after a southerly wind of 3 m/s
was added and (right) on day 7 after the wind relaxed. Also shown is (top) the Coriois term, center
panels the surface elevation gradient force and (bottom) the baroclinic pressure gradient force term.
Units are m
2
/s
2
.
C02002 CHEN ET AL.: RIVER PLUME BAROCLINIC INSTABILITY
15 of 17
C02002
[36] Our experiments clearly show the advantages of an
unstructured grid model in resolving the complex geometry
and frontal instability in the ECS. This result is consistent
with the other recent application of FVCOM in the ECS by
Isobe and Beardsley [2006]. By resolving complex bottom
topography over the slope, they successfully simulated
nonlinear frontal waves and eddies produced by the inter-
action of Kuroshio with the spatially varying bottom topog-
raphy. Since the ECS circulation is strongly influenced by
the Changjiang discharge and the offshore detachment of
CDW is relevant to the baroclinic instability of the CDW
front and eddy formation, an numerical mass-conservative
unstructured-grid model with geometry flexibility is highly
recommended in that region.
[37] ECS-FVCOM is also configured with a higher hor-
izontal resolution Changjiang Estuary subdomain. This
subdomain (with a resolution of 0.10.5 km) resolves all
major islands off Hanghzhou Bay and the narrow water
passages in the Changjiang. This subdomain model can be
driven through the grid-nesting approach from the regional
ECS-FVCOM. We ran the subdomain model to check
whether the horizontal resolution used in the regional model
is sufficient to resolve the spatial structure of the Chang-
jiang plume in the inner shelf and also examine the
contribution of tidal forcing to the plume instability. We
found little difference in the plume structure and behavior,
so in this paper we present only the regional ECS-FVCOM
model results with a horizontal resolution of 0.5 km in the
Changjiang. The fine-grid Changjiang subdomain model
results have been compared with recent field measurement
data and the results will be presented in a separate paper.
5. Summary
[38] Physical processes that cause offshore detachment of
CDW into the ECS have been examined using the high-
resolution, unstructured-grid, finite-volume coastal ocean
model (FVCOM). The numerical experiments show that
the summertime ECS regional circulation is strongly influ-
enced by the Changjiang discharge. As a result of the large-
scale adjustment, the TWC, YCCC, YCWC, and Tsushima
Warm Current are intensified. Low-salinity CDW can be
carried eastward over the submerged plateau bounded by
the 50-m isobath by either enhanced clockwise vorticity
motion of TWC or intensified southeastward transport of
YCCC.
[39] Given the horizontal resolution of less than 1.0 km,
the CDW front is baroclinically unstable, and CDW can be
transported eastward toward Cheju Island by the eddies that
detach from the front. This detachment process is enhanced
by southerly winds due to the spatially nonuniform inter-
action of wind-induced Ekman flow and frontal eddy
formation and detachment. Characteristics of the model-
predicted eddy field are consistent with previous theoretical
studies of the baroclinic instability of buoyancy-driven
coastal currents and high-resolution MODIS SST imagery
for the CDW area taken in the summer of 2005. The
baroclinic instability of the CDW front is controlled by
the horizontal Ekman number. For the Changjiang, this
number is much smaller than the criterion suggested by
theoretical analysis.
[40] The model particles show similar trajectories as
drifters released off the Changjiang during the summer of
1986. The comparison between these particle trajectories
and dye released at the same location suggests that with
southerly winds the near-surface particles can be advected
to Cheju Island and then enter the JES over a time scale of
40–45 d while the dye spreads more slowly eastward and is
still confined to the submerged plateau. Both horizontal and
vertical diffusive processes need to be taken into account in
order to accurately simulate offshore CDW transport.
[41] We note here that our experiments have focused on
an idealized process-oriented study to explore the physical
mechanisms for the offshore transport of CDW in the ECS.
Figure 21. Distribution of the surface elevation (left) at the end of day 60 for the case with only
Changjiang discharge, (middle) at the end of day 30 day after a southerly wind of 3 m/s was added,
(right) and on day 7 after the wind relaxed.
C02002 CHEN ET AL.: RIVER PLUME BAROCLINIC INSTABILITY
16 of 17
C02002
More realistic simulations should be carried out to validate
the model through comparisons with direct in situ and
satellite observations. Since the instability of the CDW
front is closely related to local bathymetry and the multiple
discharge sources of CDW in the Changjiang mouth, we
strongly recommend using an unstructured-grid model with
local mass, momentum, and tracer conservation for future
studies of coastal processes in the ECS.
[42]Acknowledgments. The development of FVCOM is supported
by the Massachusetts Fisheries Institute through NOAA grants DOC/
NOAA/NA04NMF4720332 and DOC/NOAA/NA05NMF4721131 and al-
so the U.S. GLOBEC Northwest Atlantic/Georges Bank program through
NSF grants OCE-0234545 and OCE-0227679, NOAA grant NA160P2323
and ONR subcontract grant from Woods Hole Oceanographic Institution.
C. Chen serves as Zi Jiang Scholar at the State Key Laboratory for
Estuarine and Coastal Research, East China Normal University. P. Ding
is supported by the Chinese National Key Basic Research Project grant
2002CB412403. X. Mao is supported by the National Natural Science
Foundation of China (NSFC) grant 40576079. This paper is 07-0201 in the
SMAST Contribution Series, School of Marine Science and Technology,
University of Massachusetts-Dartmouth.
References
Beardsley, R., R. Limeburner, H. Yu, and G. A. Cannon (1985), Discharge
of the Changjiang (Yangtze River) into the East China Sea, Cont. Shelf
Res.,4, 57– 76.
Beardsley, R., R. Limeburner, K. Kim, and J. Candela (1992), Lagrangian
flow observations in the East China, Yellow and Japan Sea, La Mer,30,
297 – 314.
Chang, P. H., and A. Isobe (2003), A numerical study on the Changjiang
diluted water in the Yellow and East China Sea, J. Geophys. Res.,
108(C9), 3299, doi:10.1029/2002JC001749.
Chen, C. (2000), A modeling study of episodic cross-frontal water trans-
ports over the inner shelf of the South Atlantic Bight, J. Phys. Oceanogr.,
30, 1722– 1742.
Chen, C., R. Beardsley, and R. Limeburner (1992), The structure of the
Kuroshio southwest of Kyushu: velocity, transport and potential vorticity
fields, Deep Sea Res.,39(2), 245– 268.
Chen, C., R. Beardsley, and R. Limeburner (1994), Comparison of winter
and summer hydrographic observations in the Yellow and East China
Seas and adjacent Kuroshio during 1986, Cont. Shelf Res.,14, 909– 929.
Chen, C., L. Zheng, and J. Blanton (1999), Physical processes controlling
the formation, evolution, and perturbation of the low-salinity front in the
inner shelf off the southeastern U.S.: A modeling study, J. Geophys. Res.,
104, 1259– 1288.
Chen, C., H. Liu, and R. Beardsley (2003), An unstructured grid, finite-
volume, three-dimensional, primitive equations ocean model: Application
to coastal ocean and estuaries, J. Atmos. Oceanic Technol.,20(1), 159 –
186.
Chen, C., G. Cowles, and R. C. Beardsley (2004), An unstructured grid,
finite-volume coastal ocean model: FVCOM user manual, 1st ed., Tech.
Rep. 04-0601, 183 pp., School of Mar. Sci. and Technol., Univ. of Mass.,
Dartmouth.
Chen, C., R. C. Beardsley, and G. Cowles (2006a), An unstructured grid,
finite-volume coastal ocean model (FVCOM) system, Oceanography,
19(1), 78– 89.
Chen, C., G. Cowles, and R. C. Beardsley (2006b), An unstructured grid,
finite-volume coastal ocean model: FVCOM user manual, 2nd ed., Tech.
Rep. 06-0602, 315 pp., School of Mar. Sci. and Technol., Univ. of Mass.,
Dartmouth.
Chen, C., H. Huang, R. C. Beardsley, H. Liu, Q. Xu, and G. Cowles
(2006c), A finite-volume numerical approach for coastal ocean circula-
tion studies: Comparisons with finite-difference models, J. Geophys.
Res.,112 , C03018, doi:10.1029/2006JC003485.
Chen, C., Q. Xu, R. Houghton, and R. C. Beardsley (2007), A model-dye
comparison experiment in the tidal mixing front zone on the southern
flank of Georges Bank, J. Geophys. Res., doi:10.1029/2007JC004106, in
press.
Choi, B. H. (1980), A tidal model of the Yellow Sea and the Eastern China
Sea, Rep. 80-02, 72 pp., Korea Ocean Res. and Dev. Inst. (KORDI),
Ansan, Korea.
Galperin, B., L. H. Kantha, S. Hassid, and A. Rosati (1988), A quasi-
equilibrium turbulent energy model for geophysical flows, J. Atmos.
Sci.,45, 55– 62.
He, K. L., P. X. Ding, and Y. Z. Kong (2006), Salinity variation in the
southern branch and salt water intrusion in the northern branch of the
Changjiang, Adv. Nature Sci.,16(5), 584 – 589.
Houghton, R. W. (2002), Diapycnal flow through a tidal front: a dye tracer
study on Georges Bank, J. Mar. Syst.,37, 31 – 46.
Ichikawa, H., and R. C. Beardsley (2002), The current system in the Yellow
and East China Seas, J. Oceanogr.,58, 77 – 92.
Isobe, A., and R. C. Beardsley (2006), An estimate of the cross-frontal
transport at the shelf break of the East China Sea with the finite-volume
coastal ocean model, J. Geophys. Res.,111, C03012, doi: 10.1029/
2005JC003290.
Isobe, A., M. Ando, T. Watanabe, T. Senjyu, S. Sugihara, and A. Manda
(2002), Freshwater and temperature transports through the Tsushima-
Korea Straits, J. Geophys. Res.,107(C7), 3065, doi:10.1029/
2000JC000702.
Kong, Y. Z., S. L. He, P. X. Ding, and K. L. Hu (2004), Characteristics of
temporal and spatial variation of salinity and their indicating significance
in the Changjiang estuary, Acta Oceanol. Sinica,26(4), 9 – 18.
Lie, H. J., C. H. Cho, J. H. Lee, P. Niiler, and J. H. Hu (1998), Separation of
the Kuroshio water and its penetration onto the continental shelf west of
Kyushu, J. Geophys. Res.,103, 2963– 2976.
Lie, H. J., C. H. Cho, J. H. Lee, and S. Lee (2003), Structure and eastward
extension of the Changjiang River plume in the East China Sea, J. Geo-
phys. Res.,108(3), 3077, doi: 10.1029/2001JC001194.
Mellor, G. L., and T. Yamada (1982), Development of a turbulence closure
model for geophysical fluid problem, Rev. Geophys.,20, 851 – 875.
Pedlosky, J. (1979), Geophysical Fluid Dynamics, 624 pp. Springer, New
York.
Qiu, B., and N. Imasato (1990), A numerical study on the formation of the
Kuroshio counter current and the Kuroshio branch current in the East
China Sea, Cont. Shelf Res.,10, 165– 184.
Qiu, B., N. Imasato, and T. Waji (1988), Instability of buoyancy-driven
coastal density currents, J. Geophys. Res.,93, 5037 – 5050.
Smagorinsky, J. (1963), General circulation experiments with the primitive
equations, I. The basic experiment, Mon. Weather Rev.,91, 99– 164.
Yanagi, T., and S. Takahashi (1993), Seasonal variation of circulations in
the East China Sea and the Yellow Sea, J. Oceanogr.,49, 503– 520.
R. C. Beardsley, Department of Physical Oceanography, Woods Hole
Oceanographic Institution, Woods Hole, MA 02543, USA.
C. Chen, G. Cowles, G. Gao, J. Qi, Q. Xu, and P. Xue, Department of
Fisheries Oceanography, Intercampus Graduate School for Marine Science
and Technology, University of Massachusetts, New Bedford, MA 02744,
USA. (c1chen@umassd.edu)
P. Ding, State Key Laboratory for Estuarine and Coastal Research, East
China Normal University, Shanghai, 200062, China.
C. Li, Department of Oceanography and Coastal Sciences, School of the
Coast and Environment, Louisiana State University, Baton Rouge, LA
70803, USA.
H. Lin, Marine Extension Service, University of Georgia, Athens, GA
30602, USA.
X. Mao, State Key Laboratory of Satellite Ocean Environment Dynamics,
Second Institution of Oceanography, State Oceanic Administration,
Hangzhou, 310012, China.
M. Shi, College of Physical and Environmental Oceanography, Ocean
University of China, Qingdao, 266100, China.
C02002 CHEN ET AL.: RIVER PLUME BAROCLINIC INSTABILITY
17 of 17
C02002
