Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 108, NO. C7, 3226, doi:10.

1029/2002JC001473, 2003

A numerical study of dense water outflows and halocline anticyclones


in an arctic baroclinic slope current
Shenn-Yu Chao
Horn Point Laboratory, University of Maryland Center for Environmental Science, Cambridge, Maryland, USA

Ping-Tung Shaw
Department of Marine, Earth and Atmospheric Sciences, North Carolina State University, Raleigh, North Carolina, USA
Received 13 May 2002; revised 13 December 2002; accepted 18 April 2003; published 15 July 2003.

[1] We examine the relation between the dense water migration from the shelf to the deep
ocean and offshore generation of halocline anticyclones in a physical setting appropriate
for an Arctic baroclinic current system, using a nonhydrostatic numerical ocean model.
The model basin contains an outer shelf connected to the deep sea by a continental slope.
The ocean is cyclical in the alongshore direction. Sea surface is frictional to account for
the ice-exerted friction. A series of inflows and outflows is imposed on the shoreward side
of the shelf to facilitate the release of dense waters into the basin. The distribution of
inflow-outflow is sinusoidal in the alongshore direction. Inflow carries dense water
into the basin while the outflow withdraws ambient shelf water from the shelf. The
multiple sources produce an array of dense plumes entering the shelf. Seaward migration
of dense plumes is examined with and without an offshore baroclinic current bounded
to the left by the shelf. The baroclinic current, if present, is supported by slanted
isopycnals rising toward the shelf break. Without the baroclinic current, downslope
sinking plumes veer to the right and produce an undercurrent trapped by the continental
slope. With the baroclinic current, convection along slanted isopycnals can divert dense
plumes away from boundary trapping and produce intense subsurface anticyclones
offshore. Initial meanders embedded in the baroclinic current can trigger instabilities and
also produce subsurface anticyclones offshore. However, disturbances induced by sinking
plumes and instabilities of offshore origin generally do not interact constructively to
intensify subsurface anticyclones offshore. INDEX TERMS: 4255 Oceanography: General:
Numerical modeling; 4520 Oceanography: Physical: Eddies and mesoscale processes; KEYWORDS: dense
water outflow, eddy shedding
Citation: Chao, S.-Y., and P.-T. Shaw, A numerical study of dense water outflows and halocline anticyclones in an arctic baroclinic
slope current, J. Geophys. Res., 108(C7), 3226, doi:10.1029/2002JC001473, 2003.

1. Introduction topographic features such as submarine canyons or sharp


changes in coastline orientation may trigger active shelf-
[2] Over the Arctic shelf, the production and offshore
basin exchange. Off the Chukchi shelf, for example, the
migration of cold, saline waters (formed by brine rejected
dense outflow into the Arctic Ocean was observed to be
from growing ice) are essential to the maintenance of the
quite active through Barrow Canyon [Weingartner et al.,
halocline [Aagaard et al., 1981; Weingartner et al., 1998].
1998].
Brine ejection events in the latent heat polynyas over the
[3] The density of shelf waters regulates the maximum
shelf produce the dense water which in turn ventilates the
sinking depth. Weingartner et al. [1998] identified three
cold halocline layer of the Arctic Ocean [Aagaard et al.,
modes of waters in the outer Chukchi shelf available for
1981; Cavalieri and Martin, 1994; Roach et al., 1995]. The
offshore migration. The dominant mode had moderate
dense water migration from the shelf to the deep ocean can
salinity (32.5) and low temperature (<1.4C) character-
be widespread along the shelf break. This seems to be the
istic of Chukchi Sea winter water [Coachman et al., 1975].
prevalent mode of export from the Mackenzie shelf in the
The intermediate mode water is denser, having higher
northeastern Beaufort Sea, as the frontal structure seemed to
salinities (33  33.6). The hypersaline mode had salinities
vary little along the shelf break [Melling, 1993]. Over other
>34.0. The relative contribution from each mode varies
shelves, the offshore migration of dense waters is often
from year to year. If none of these modes mix as they
localized in a few preferred locations. Conceivably, distinct
migrate off the shelf, the dominant mode will ventilate the
upper halocline while the saltiest portion of the hypersaline
Copyright 2003 by the American Geophysical Union. mode water will sink to the bottom layer of Canada Basin.
0148-0227/03/2002JC001473 Offshore migration without mixing is of course an ideal-

19 - 1
21562202c, 2003, C7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2002JC001473 by Readcube-Labtiva, Wiley Online Library on [03/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19 - 2 CHAO AND SHAW: ARCTIC DENSE SHELF WATER OUTFLOW AND EDDY SHEDDING

ization. In reality, the dense shelf water does mix as it sinks also noted recently by Yoshikawa et al. [2001] in their
and in consequence, should ventilate the deep ocean over a investigations of a similar subject. Subsequent convection at
range of depths above the maximum allowable sinking depths follows slanted isopycnal surfaces and produces
depth. intense subsurface anticyclones on the offshore side of the
[4] Departing from the shelf break, sinking plumes are baroclinic current. The work established a possible link
subject to trapping by the continental slope. This is evident between slanted convection and offshore generation of
from results of previous numerical investigations [Chapman subsurface anticyclones. In reality, the lateral injection of
and Gawarkiewicz, 1995; Jiang and Garwood, 1996 and dense waters from the outer shelf rather than surface cooling
1998; Kikuchi et al., 1999; Gawarkiewicz, 2000] and triggers slanted convection. It remains to be demonstrated
rotating tank experiments [Etling et al., 2000; Lane-Serff that lateral injection of dense water can indeed lead to the
and Baines, 2000]. These studies excluded the presence of a offshore production of subsurface anticyclones.
baroclinic current offshore. In the otherwise quiescent [7] This paper continues to investigate the possible links
coastal ocean, sinking plumes tend to follow isobaths and among offshore migration of dense shelf water, seaward
intrude in the propagation direction of topographic Rossby dispersal of sinking plumes, and generation of offshore
waves. Eddies are occasionally shed by the sinking plumes, subsurface anticyclones. The ocean in this model contains
but these eddies appear to stay close to the continental slope a piece of the outer shelf connected to the deep basin by a
rather than move farther out into the deep basin. In other continental slope. The mesoscale basin is capped by a
words, the boundary trapping appears to be a strong frictional surface to account for friction exerted by sea
dynamical constraint against seaward dispersal. If this ice. Surface friction dampens the surface current which
trapping mechanism is unbreakable, the Arctic halocline would otherwise become unrealistically strong with contin-
cannot be ventilated efficiently. The apparent gap between uous forcing by offshore migration of dense shelf water.
models and observationalists’ perception of halocline ven- Localized topographic features such as submarine canyons
tilation motivates us to revisit the outflow problem by and sharp bends of the coastline are excluded for the
including more realistic flow conditions offshore. At issue purpose of identifying mechanisms in the simplest possible
is the anticyclonic Beaufort Sea gyre, which skirts around physical setting. A series of barotropic inflows and out-
the shelf break. Isopycnals associated with this baroclinic flows, sinusoidal in profile, is imposed on the shoreward
current are slanted, rising toward the shelf break. side of the outer shelf to initiate sinking plumes. The inflow
[5] Dynamically, there are reasons to suspect that the carries dense water into the outer shelf while the outflow
presence of a baroclinic current offshore could provide withdraws ambient shelf water out of the basin. Subsequent
mechanisms leading to the separation of sinking plumes dispersal of sinking plumes is examined with and without a
from the continental slope. First, the baroclinic current baroclinic current beyond the shelf break. A nonhydrostatic
flows against the subsurface density current induced by ocean model is used to better cope with downslope accel-
sinking plumes over the continental slope. This kinematic eration of sinking plumes. The baroclinic current offshore
mechanism cannot dominate because the baroclinic current can be initially stable if it is uniform alongshore, or mildly
such as the Beaufort Sea gyre is expected to be weak below unstable if it contains meanders. The instability of offshore
200 m depth, where isopycnal slopes begin to level [Wein- origin further complicates the seaward dispersal of sinking
gartner et al., 1998]. A second mechanism is due to the plumes.
slanted isopycnal surfaces associated with the baroclinic
current. The consequent convection more or less following 2. Model Formulation
slanted isopycnal surfaces may divert sinking plumes away
from boundary trapping. Finally, sinking plumes may add [8] The oceanographic setting of this model is similar to
potential energy to the baroclinic current and trigger bar- that of Chao and Shaw [2002] except for the addition of the
oclinic instability, which in turn may provoke seaward shelf and slope topography. Further, surface cooling is
dispersal. replaced by dense outflow from the shelf to trigger slanted
[6] Another motivating question is whether the dense convection. The formulation of the nonhydrostatic model
water outflow causes the ubiquity of subsurface anticy- essentially follows procedures as outlined by Williams
clones in the Beaufort Sea. These eddies have been known [1969] except for several major improvements suggested
to populate the halocline for some time [Newton et al., by Marshall et al. [1997]. Under the Boussinesq approxi-
1974; Manley and Hunkins, 1985; D’Asaro, 1988a; Muench mation, the three-dimensional momentum, continuity, and
et al., 2000] and their possible origin from the Chukchi density equations are
shelf outflow had been speculated [D’Asaro, 1988b]. Never-  
Dv 1 @ @v
theless, numerical modeling efforts have not been able to þ 2 k0 v ¼  rp  rgk  A4 r4H v þ n ;
establish such a link to date. In our most recent attempt Dt r0 @z @z
[Chao and Shaw, 2002], we have simulated the effect of
 
surface cooling on a meandering jet along a flat-bottomed Dr @ @r
coastal ocean. Surface friction exerted by sea ice was also ¼ K4 r4H r þ k ;
Dt @z @z
included. Under that physical setting, convection becomes
intense in surface areas occupied by cyclonic vorticity,
funneling a large volume of dense water downward. The r  v ¼ 0;
intense convection in turn enhances surface cyclonic vor-
ticity. Interestingly, the positive feedback mechanism where v is the three-dimensional velocity vector (u,v,w), k0
between convection and surface anticyclonic vorticity was is a unit vector pointing upward from the North Pole, k is
21562202c, 2003, C7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2002JC001473 by Readcube-Labtiva, Wiley Online Library on [03/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
CHAO AND SHAW: ARCTIC DENSE SHELF WATER OUTFLOW AND EDDY SHEDDING 19 - 3

boundary, allowing the ambient shelf water to be withdrawn


from the shelf. The seaward boundary is open; the normal
gradient of all variables is zero. The model domain is
cyclical in the alongshore direction. Since all horizontal
mixing terms are biharmonic, each lateral boundary condi-
tion on the coastal and offshore boundaries must be paired
with a higher order boundary condition to make the problem
well posed. Let Ba = 0 be a boundary condition in which B
is the boundary operator and a a dependent variable. The
standard choice is taken with rH2Ba = 0 as the additional
constraint. Vertically, the ocean bottom is no-slip and
impermeable. The sea surface is capped by a rigid lid on
Figure 1. Model setting, dimensions, and a subsurface which the boundary constraint is a mixture of free-slip and
eddy generation scenario. no-slip conditions,

@ ðu; vÞ e
ð1  eÞ þ ðu; vÞ ¼ 0;
the local upward unit vector, and r is the perturbation @z z
density about a reference seawater density (r0 = 1025 kg m3).
The subscript H changes the gradient operator from where z is the vertical grid spacing and e is chosen to be
three-dimensional to horizontal. Horizontal mixing coeffi- 0.5, accounting for the partial friction exerted by sea ice.
cients are biharmonic with A4 = 5.6  108 m4 s1 and K4 = The choice for the value of e is meant to suppress
5.6  107 m4 s1. Biharmonic mixing is used instead of unrealistically strong surface currents induced by the
Laplacian mixing to suppress small-scale mixing and in the continuous offshore migration of dense shelf waters, which
process enhances eddy growth [Holland, 1978]. The latitude enhances surface currents markedly if e = 0 but marginally
is fixed at 70N, so that the Earth’s b-effect is neglected. The if e = 1.
vertical mixing coefficients depend on the Richardson [10] A baroclinic current can be prescribed offshore. The
number according to the formulation of Pacanowski and initial current can be uniform alongshore or contain mean-
Philander [1981]. The purpose is to enhance vertical mixing ders. Major experiments are initialized with a baroclinic
in convectively unstable regions. The model is not sensitive current without meanders. Initial meanders are occasionally
to this particular vertical mixing scheme, in the sense that used mainly to trigger the instability of offshore origin. The
sizeable constant mixing coefficients in the vertical would instability originated offshore turns out to be irrelevant if
produce qualitatively similar solutions. one wishes to link the dense outflow to the offshore
[9] Figure 1 illustrates the model ocean, with x axis production of halocline anticyclones. With or without
pointing alongshore, y pointing offshore, and z pointing meanders, the geostrophic jet can be initialized as follows.
vertically upward. The horizontal dimension is 140 km in Let d = 150 m be a depth scale of the jet, Ly = 40 km be the
the alongshore direction, 120 km in the cross-shelf direc- e-folding half-width of the jet, yc = 60 km be the mean
tion, and 500 m in the vertical. The outer shelf is chosen to distance between the jet axis and shoreward boundary, and
be 100 m deep. One could choose a shallower outer shelf if r = 2 kg m3 be the density anomaly supporting the
the major interest is the Chukchi Sea or east Siberian Sea. geostrophic jet. The reduced gravity can be defined as g0 = g
This type of fine adjustment is quantitative and will not r/r0 to characterize the jet. The alongshore current is
change the conclusions we intend to establish herein. "  #
 
Deeper reaches of the ocean are also truncated at 500 m g0 d 2px 2 2
because major developments are in the halocline. An u ¼ pffiffiffi exp  y  yc  a sin =Ly
fLy p l
alongshore series of sinusoidal barotropic inflows and out-   
flows is imposed on the shoreward boundary of the shelf. H
 tanhð z=d Þ þ tanh ;
The alongshore profile of inflow-outflow is sinusoidal. d
Dense water enters the shelf with the inflow, and an equal
amount of lighter ambient shelf water is withdrawn by the where a and l are meander amplitude and wavelength,
outflow. While the volume flux is conserved, a net density respectively. The shore-normal velocity for the meaner jet is
flux enters the outer shelf. The density of inflow (rin) is
constant, chosen to have a density equivalent to or greater pffiffiffi "    #  
than the halocline density. To a large extent, the inflow 2a0 gd p 2px 2 2 2px
v¼ exp  y  yc  a sin =Ly cos
density dictates the magnitude of coastal ocean response. f lLy l l
The inflows and outflows (U0) play a secondary role. If the  z  
H
inflow does not carry dense water onto the shelf, the  tanh þ tanh :
d d
circulation is mainly from sources to nearby sinks and will
not expand beyond the shelf break. With the density input, a
larger speed hastens the release of dense plumes onto the If there is no meander (a = 0), v becomes zero. The initial
outer shelf and therefore hastens the coastal ocean response. perturbation density field consists of two components,
On the other hand, a slower speed would retard the response
over and beyond the shelf break. The normal gradient of r ¼ ra ð zÞ þ rc ð x; y; zÞ:
density is zero on the outflow portion of the shoreward
21562202c, 2003, C7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2002JC001473 by Readcube-Labtiva, Wiley Online Library on [03/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19 - 4 CHAO AND SHAW: ARCTIC DENSE SHELF WATER OUTFLOW AND EDDY SHEDDING

hydrostatic pressure (nonhydrostatic pressure) is solved by


forming a three-dimensional Poisson equation to enforce
zero velocity divergence as required by the continuity
equation. The three-dimensional Poisson equation is solved
by the preconditioned conjugate gradient (PCG) method
[e.g., Pozrikidis, 1997]. Integration in time is done by an
Adams-Bashforth scheme; a Euler time step is applied
occasionally to smooth the integration. The Arakawa-C
grid system is used, and spatial derivatives are center-
differenced to second-order accuracy. Vertical discretization
follows a z-coordinate formulation.

3. Numerical Results
[13] Table 1 lists a few selected experiments to be
discussed below. These experiments are arranged and
discussed in pairs for the purpose of isolating mechanisms.
Experiments 1 and 2 examine the ocean response to
sinusoidal inflow-outflow with and without a baroclinic
Figure 2. Vertical sections of initial geostrophic current current offshore, respectively. The baroclinic current is
(thick lines in cm s1) and perturbation density (thin lines in initially uniform in the alongshore direction. The amplitude
kg m3) in the absence of meanders. The flow is directed of inflow-outflow (U0) is chosen to be 20 cm s1, a
into the paper. In experiments without the baroclinic moderate speed for the observed dense outflow from
current, the density profile far offshore is used to initialize Chukchi shelf [Weingartner et al., 1998]. The width of
stratification everywhere. each source or sink (W) is 35 km on the shoreward side of
the outer shelf. The perturbation density for the inflow (rin)
is 1 kg m3. Referring to Figure 2, the eventual sinking
The first component (ra) provides background stratification depth (Dmax) is about 250 m or the lower halocline if the
below the depth of zm = 100 m characterized by a density dense inflow does not mix during the offshore migration.
difference of r0 = 1.5 kg m3 and a depth scale of Z0 = Thus the inflow water in experiments 1 and 2 corresponds
250 m. Mathematically, ra (z) is given by r0 + r0 to the intermediate mode water over the Chukchi shelf
tanh[(z  zm)/Z0] below 100 m depth and by r0 above available for halocline ventilation [Weingartner et al.,
it. The second component (rc) is in geostrophic balance with 1998]. Experiments 3 and 4 use narrower (23.3 km) inflow
alongshore current, given by and outflow to examine the model’s sensitivity to inflow
     z density (rin) with a baroclinic current offshore. The bar-
2px oclinic current is again uniform alongshore. The inflow
rc ¼ 0:5rerfc  y  yc  a sin =Ly = cosh2 ;
l d perturbation density is 1 kg m3 in experiment 3 and
where erfc is the error function. When initialized as such, 0 kg m3 in experiment 4. Referring to Figure 2, the
the meanders disappear if a = 0. Further, the baroclinic maximum sinking depth (Dmax) for rin = 0 kg m3 in
current can also be removed if yc ! 1. In this limit the experiment 4 is 500 m or well below halocline, cor-
initial coastal ocean is motionless but contains vertical responding to the hypersaline mode water over the Chuk-
stratification identical to that far offshore. chi shelf available for offshore migration [Weingartner et
[11] Figure 2 shows a vertical section of flow and density al., 1998].
fields for the initial baroclinic current if the meander [14] It was often suggested that the instability of a
amplitude is zero. The maximum current speed exceeds baroclinic current might lead to generation of subsurface
25 cm s1 at the sea surface and the jet axis is 40 km off the anticyclones offshore [Manley and Hunkins, 1985]. Exper-
shelf break. The overall density distribution resembles the
observed transect off the Chukchi shelf [see Weingartner et
al., 1998, Figure 10]. Seaward of the front, the overall Table 1. List of Numerical Experiments, Amplitude of Inflow-
vertical stratification is about 3.5 kg m3, comparable to Outflow (U0), Width of Inflow or Outflow (W), Inflow Perturba-
that of the Beaufort Sea halocline. In the subsurface depths tion Density (rin), Presence or Absence of Offshore Current, Initial
(below 200 m) of interest to subsurface anticyclone gener- Meander Wavelength (l), and Maximum Sinking Depth for the
ation, the initial jet speed is much below 5 cm s1. As a Inflow (Dmax)
reference, the first baroclinic Rossby radius is about 21.5 U0, W, rin, kg Initial Offshore l, Dmax,
km seaward of the baroclinic current. Experiment cm s1 km m3 Current km m
[12] The horizontal and vertical resolutions are 2 km and 1 20 35 1 yes none 250
25 m, respectively. The time step of integration is 60 s. The 2 20 35 1 no none 250
formulation of the nonhydrostatic model essentially follows 3 20 23.3 1 yes none 250
4 20 23.3 0 yes none 500
the procedure outlined by Marshall et al. [1997]. The 5 0 none none yes 70 none
depth-averaged velocity and hydrostatic pressure in the 6 20 35 1 yes 70 250
three-dimensional momentum equation are calculated as 7 0 none none yes 46.6 none
in traditional hydrostatic models, while the deviation from 8 20 23.3 0 yes 46.6 500
21562202c, 2003, C7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2002JC001473 by Readcube-Labtiva, Wiley Online Library on [03/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
CHAO AND SHAW: ARCTIC DENSE SHELF WATER OUTFLOW AND EDDY SHEDDING 19 - 5

Figure 3. Flow and negative vorticity fields at (a, c) 212.5 m depth and (b, d) 62.5 m depth driven by
wavenumber-2 sinusoidal inflow-outflow from the shelf ( y = 0 km) for 15 days. Left-hand panels
(Figures 3a and 3b) are from experiment 2 in which the coastal ocean is initially motionless. Right-hand
panels (Figures 3c and 3d) are from experiment 1, which initially contains a baroclinic current offshore.
The baroclinic current is initially uniform in the alongshore direction. The inflow-outflow profile is given
at the bottom of Figures 3b and 3d. Contours of negative vorticity begin from 0.05 f and decrease at
intervals of 0.05 f.

iment 5 illustrates the outcome of this offshore generation or sink (35 km) is somewhat larger than the first baroclinic
scenario by switching off the inflow-outflow over the Rossby radius (21.5 km) far offshore. Flow fields and
shelf. The instability is triggered by initial meanders within negative vorticity contours are shown at 212.5 m depth
the offshore current having a wavelength (l) of 70 km. (Figures 3a and 3c) and 62.5 m depth (Figures 3b and 3d).
Experiment 6 explores the relation between responses in- The inflow-outflow profile is indicated at the bottom of
duced by sinking plumes and baroclinic instability of bottom panels as a reference. The ability of an offshore
offshore origin. Dense water is produced over the shelf by baroclinic current to remove sinking plumes trapped over
the inflow-outflow identical to that in experiment 1, while the continental slope is striking. Sinking plumes from the
instability offshore is produced by initial meanders within shelf manifest a subsurface undercurrent trapped on the
the offshore current as in experiment 5. In terms of the continental slope in the case without a boundary current
offshore production of subsurface anticyclones, the effects (Figure 3a). The trapping by the continental slope is
of sinking plumes and instability originated offshore gener- expected in light of previous investigations of similar
ally do not interact constructively. Experiments 7 and subjects using hydrostatic models [Jiang and Garwood,
8 essentially repeat experiments 5 and 6, respectively, 1996, 1998; Gawarkiewicz, 2000]. In the presence of a
except with a shorter wavelength of offshore meanders, a boundary current, the undercurrent leaks offshore and forms
narrower width of the inflow-outflow and denser inflow. two subsurface anticyclones at 212.5 m depth (Figure 3c).
The lack of constructive interaction between responses The drift to the left (negative x-direction) during offshore
induced by sinking plumes and by instability of offshore migration is a result of advection by the baroclinic current.
origin persists in these two experiments. The intensity of subsurface anticyclones in Figure 3c is
about 0.42 f at day 15. Anticyclonic vorticity at 62.5 m
3.1. Effects of Offshore Currents on Sinking Plumes (Figure 3d) is much weaker. This subsurface maximum is
[15] Figure 3, from experiments 1 and 2, contrasts model caused partly by surface friction exerted by sea ice and
responses after 15 days of dense outflow in cases with a partly by the descent of sinking plumes. Although the
baroclinic current offshore (Figure 3 (right)) and without a coastal ocean is initially at rest in experiment 2, sinking
baroclinic current (Figure 3 (left)). The width of each source plumes subsequently induce a meandering surface current
21562202c, 2003, C7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2002JC001473 by Readcube-Labtiva, Wiley Online Library on [03/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19 - 6 CHAO AND SHAW: ARCTIC DENSE SHELF WATER OUTFLOW AND EDDY SHEDDING

vorticity of subsurface anticyclones increases from 0.26 f


at day 10 to 0.42 f at day 15 and levels off to 0.43 f at
day 20. The anticyclones gain strength during the detach-
ment. Conceivably, the slanted convection involves poten-
tial energy release which works against dissipation. Further
seaward migrations thereafter depend on the ambient flow
condition and will not be illustrated.
[17] In reality, the boundary current over the continental
slope varies seasonally. The efficiency of subsurface anticy-
clone generation depends on the strength of the boundary
current. We have conducted a few numerical experiments to
explore the model sensitivity to the initial current strength.
Results are quite predictable and therefore will only be
commented on briefly without illustrations. If the initial
boundary current is much weaker than the chosen initial
current as illustrated in Figure 2, the associated initial iso-
pycnals will not rise as sharply toward the shelf break. In
consequence, a much weaker boundary current will not be
able to divert subsurface anticyclones away from boundary
trapping. Conversely, if the initial boundary current becomes
unrealistically strong, the breakaway subsurface anticyclones
are often torn apart as thin filaments due to the strong
advection by the surface current. Leaving extremely strong
and extremely weak boundary currents aside, a modest
increase in the boundary current strength generally enhances
shedding of halocline anticyclones slightly.
[18] Figure 5, from experiment 1, shows flow fields and
isopycnals at day 10 (Figure 5 (left)) and day 20 (Figure 5
(right)). The horizontal slice is at 212.5 m depth in upper
panels (Figures 5a and 5c) and 62.5 m depth in lower panels
(Figures 5b and 5d). Sinking plumes induce large meanders
embedded in the baroclinic current which is initially uni-
form in the alongshore direction. At 62.5 m depth, the
isopycnal with largest meanders roughly indicates the axis
of the meandering jet. The perturbation density for this axial
isopycnal is 2 kg m3 at day 10 (Figure 5b) and 2.5 kg m3
at day 20 (Figure 5d). When averaged alongshore, the axis
of the surface jet is at y = 60 km initially (Figure 2),
meanders about y = 40 km at day 10 (Figure 5b) and moves
seaward to y = 75 km at day 20 (Figure 5d). Sinking plumes
apparently induce the shoreward movement of surface jet
axis and surface isopycnals before day 10. The shoreward
Figure 4. Flow and negative vorticity fields at 212.5 m intrusion of surface buoyant waters is accompanied by
depth from experiment 1 at (a) day 10, (b) day 15, and (c) subsurface seaward intrusion of dense water into the hal-
day 20. The inflow-outflow profile is given at the bottom. ocline (Figure 5a). Sinking plumes also induce an under-
Contours of negative vorticity start from 0.05 f and current adjacent to the continental slope, meandering at day
decrease at intervals of 0.05 f. Maximum vorticity for 10 (Figure 5a) and shedding halocline anticyclones at day
anticyclones are given in each panel in units of f. 20 (Figure 5c).
[19] From day 10 to day 20, the surface meander jet
moves seaward but surface isopycnals do not (Figures 5b
pointing essentially to the negative x-direction (Figure 3b). and 5d). The driving force is apparently in the halocline.
The surface current is supported by rising isopycnals near Halocline anticyclones move seaward and carry the surface
the shelf break induced by convection; this point will be jet aloft with them. Conceivably, subsurface motions can
further illustrated later. influence currents aloft through vertical mixing and by
[16] Focusing on experiment 1 with a baroclinic current changing surface pressure distributions. Meanders of sur-
offshore, Figure 4 shows how subsurface anticyclones at face isopycnals generally lag somewhat behind those of
212.5 m depth break away from boundary trapping. The subsurface isopycnals if one follows the main (negative x-)
flow over the continental slope begins to extend seaward at direction of the mean jet. The phase lag is a telling feature
two isolated locations before day 10 (Figure 4a), forming of baroclinic instability at work [Phillips, 1954].
two anticyclonic meanders. These meanders continue to [20] Figure 6, also from experiment 1, shows two cross-
grow at day 15 (Figure 4b), pinching off into isolated shelf sections of density and shore-normal flow fields
subsurface anticyclones at day 20 (Figure 4c). The negative approximately half a meander wavelength apart. The
21562202c, 2003, C7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2002JC001473 by Readcube-Labtiva, Wiley Online Library on [03/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
CHAO AND SHAW: ARCTIC DENSE SHELF WATER OUTFLOW AND EDDY SHEDDING 19 - 7

Figure 5. Flow fields and isopycnals from experiment 1 at (a, b,) day 10 and (c, d) day 20. Top panels
(Figures 5a and 5c) are at 212.5 m depth, while bottom panels (Figures 5b and 5d) are at 62.5 m depth.
The inflow-outflow profile is given at the bottom of Figures 5b and 5d. Contour values refer to
perturbation density in units of kg m3. Contour intervals for isopycnals are 0.1 kg m3 at 212.5 m depth
and 0.5 kg m3 at 62.5 m depth.

cross-section at x = 80 km (Figure 6a at day 10 and Figure 6b mistakable in Figure 6d. Around y = 85 km, isopycnals
at day 20) is between two halocline anticyclones at day 20. are lifted above about 150 m depth and deepened below it.
Corresponding sections (Figures 6c and 6d) at x = 114 km The volume of reduced stratification harbors the halocline
slices through the center of a halocline anticyclone at day anticyclone.
20. If one looks for dominant circulation patterns over and [21] In experiment 1, the choice of inflow density (rin =
beyond the shelf break, snapshots of cross-sections contain 1 kg m3) is crucial to maximize offshore production of
transient alongshore variability because of moving mean- halocline anticyclones. The maximum sinking depth is in
ders and eddies. Nevertheless, several persistent features are the lower halocline. A much higher inflow density would
noteworthy. The first is the frontal circulation over the shelf have the adverse effect of diminishing halocline anticy-
and slope. Sinking plumes induce a shelf break front sloping clones offshore. We will use experiments with narrower
in the same sense as that supporting the baroclinic current inflow and outflow to illustrate the model’s sensitivity to the
offshore. Consequently, the offshore baroclinic current inflow density below.
gains additional strength from sinking plumes. Downslope
currents along the continental slope are often accompanied 3.2. Narrower Inflow and Outflow
by shelfward upwelling from halocline depths rather than [22] Experiments 3 and 4 use narrower inflow and out-
landward intrusion of surface waters. Meanders and eddies flow, each having a width of 23.3 km, which is comparable
occasionally disrupt the persistent frontal circulation. For to the first baroclinic Rossby radius far offshore (21.5 km).
example, the would-be shelfward upwelling from halocline The inflow perturbation density (rin) is 1 kg m3 in
depths in Figure 6a is offset by subsurface meanders (see experiment 3 and 0 kg m3 in experiment 4, corresponding
Figure 5a). Landward intrusion of surface waters in Figure respectively to the intermediate mode and hypersaline mode
6c is part of surface meanders at day 10 (see Figure 5b). waters available for offshore migration over the Chukchi
Note also the depressed isopycnals over the continental shelf. The maximum sinking depth of dense plumes is in the
slope below 200 m depth. The depression usually occurs lower halocline (250 m) in experiment 3 and well below
below the penetration depth of sinking plumes and, by halocline (500 m) in experiment 4. Both experiments
virtue of geostrophy, below the core of the undercurrent. include an offshore baroclinic current which is initially
The signature of the halocline anticyclone is also un- uniform alongshore. We will not illustrate cases without a
21562202c, 2003, C7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2002JC001473 by Readcube-Labtiva, Wiley Online Library on [03/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19 - 8 CHAO AND SHAW: ARCTIC DENSE SHELF WATER OUTFLOW AND EDDY SHEDDING

Figure 6. Two cross-shelf sections of isopycnals and cross-shelf flow from experiment 1. Left-hand
panels at (a) day 10 and (b) day 20 are at x = 80 km, which slices through between two subsurface
anticyclones at day 20. Right-hand panels at (c) day 10 and (d) day 20 are at x = 114 km, which cuts
through the center of a halocline anticyclone at day 20. Contour values for perturbation density are in
units of kg m3.

baroclinic current herein because they do not generate controlled by two competing factors. The first factor is
halocline anticyclones offshore. numerical damping brought about by biharmonic mixing
[23] Figure 7 shows flow fields and negative vorticity used in the model. The consequent damping rate of subsur-
contours after 20 days of forcing at 212.5 m depth (Figures face anticyclones is inversely proportional to the fourth
7a and 7c) and 62.5 m depth (Figures 7b and 7d). To lowest power of eddy size, weakening smaller anticyclones induced
order, the offshore baroclinic current still reduces boundary by narrower inflow-outflow width at a faster rate. On the
trapping as before. The undercurrent leaks offshore to other hand, smaller anticyclones in experiment 3 must
manifest anticyclones with the aid of the baroclinic current expand to a scale comparable to the Rossby radius and in
(Figures 7a and 7b). Halocline anticyclones from experi- the process gain strength through potential energy release.
ment 3 (Figure 7a) are much stronger because the dense Similar mechanism is absent in experiment 1 because the
outflow is destined for the lower halocline. In contrast, they inflow-outflow width is considerably larger than the Rossby
become shallower and more than 3 times weaker in exper- radius. When weighing two competing factors together, the
iment 4 (Figure 7c) because the dense outflow sinks too strength of halocline anticyclones could either increase or
deeply. With rin = 0 kg m3, the maximum sinking depth decrease as a function of the inflow-outflow width in the
(500 m) would be well below the halocline. In this extreme present model. Not surprisingly, the resulting strength of
the sinking plumes penetrate through the halocline. The subsurface anticyclones is not too sensitive to the inflow-
slanted convection along isopycnal surfaces of the offshore outflow width because two competing factors tend to offset
baroclinic current still occurs as a way to divert dense each other.
plumes away from boundary trapping, but becomes much [25] Figure 8, from experiment 4, shows flow and density
weaker. In consequence, the halocline anticyclones offshore fields at day 15 (Figure 8 (left)) and day 20 (Figure 8 (right)).
become much weaker. The process leading to the inefficient At 212.5 m depth (Figure 8 (top panels)), the ultra-dense
production of halocline anticyclones by ultra-dense plumes plumes induce meanders in isopycnals as in the case of
will be examined in Figure 8 and the associated discussion. wavenumber-2 forcing (Figure 5). The meander wavelength
[24] A comparison between Figure 7a (from experiment 3) is understandably shorter with the wavenumber-3 forcing.
and Figure 4c (from experiment 1) illustrates the model’s Seaward intruding meander crests carry dense waters off-
dependence on the inflow-outflow width alone because both shore and trap anticyclones in between (Figure 8a). Because
experiments employ the same inflow density destined for the the density of sinking plumes is too high, seaward intruding
lower halocline. By doing so, the dependence of the strength meander crests at depths become expansive and widen in
of halocline anticyclones on inflow-outflow width is mainly time, narrowing the troughs further (Figure 8c). Anticyclonic
21562202c, 2003, C7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2002JC001473 by Readcube-Labtiva, Wiley Online Library on [03/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
CHAO AND SHAW: ARCTIC DENSE SHELF WATER OUTFLOW AND EDDY SHEDDING 19 - 9

Figure 7. Flow and negative vorticity fields at (a, c) 212.5 m depth and (b, d) 62.5 m depth driven by
wavenumber-3 sinusoidal inflow-outflow from the shelf for 20 days. Left-hand panels (Figures 7a and
7b) are from experiment 3 with modest inflow density (rin = 1 kg m3). Right-hand panels (Figures 7c
and 7d) are from experiment 4 with ultra-dense sinking plumes (rin = 0 kg m3). The baroclinic current is
initially uniform in the alongshore direction. The inflow-outflow profile is given at the bottom of Figures
7b and 7d. Contours of negative vorticity begin from 0.05 f and decrease at intervals of 0.05 f. The
maximum vorticity for subsurface anticyclones is given in Figures 7a and 7c.

vortices at 212.5 m depth weaken in time as a result of tively. These meanders are sufficiently small in amplitude
reduced trough areas. The halocline anticyclones become and therefore induce mildly unstable meanders and eddies
increasingly confined to upper depths in the process. It can in time. Surface and subsurface meanders are vertically
be concluded that ultra-dense plumes do not favor the aligned initially. The inflow-outflow profiles shown at the
generation of halocline anticyclones. The signature of bar- bottom of each panel are not used in this subsection, but
oclinic instability is also evident in Figure 8. Upper level will be used in the next subsection when we examine the
isopycnal meanders lag those at depths if one follows the interaction between sinking plumes and instability of off-
main direction of the mean jet. shore origin.
[26] The model becomes sensitive to the maximum depth [28] Figure 10 shows meanders and eddies induced by the
of the ocean if the shelf water becomes exceedingly dense. instability of offshore origin. The instability evolves more
If the maximum sinking depth of the dense shelf water rapidly with wavenumber-3 meander forcing (Figure 10
exceeds 500 m, then the maximum bottom depth should be (right)) than with wavenumber-2 forcing (Figure 10 (left)).
placed at or below the maximum sinking depth to allow for For this reason, flow and density fields at 212.5 m depth
deeper descent of sinking plumes. However, this issue is (Figure 10 (top panels)) and 62.5 m depth (Figure 10
quantitative and will not change our major conclusion that (bottom panels)) are displayed at day 15 for the wave-
deep sinking plumes generally weaken the production of number-2 case and at day 10 for the wavenumber-3 case.
halocline anticyclones. Regardless of the meander wavelength, consequences are
similar. The instability leads to a series of subsurface anti-
3.3. Instability of Offshore Origin cyclones on the offshore side of the jet and subsurface
[27] Excluding dense plumes from the shelf, meanders cyclones on the shoreward side. Offshore anticyclones are
initially embedded in the baroclinic current can also trigger enclosed by the isopycnal with a perturbation density of
instabilities. Figure 9, from experiments 5 and 7, shows 1.3 kg m3 at 212.5 m depth (Figures 10a and 10c). The
initial surface flow and density fields for (Figure 9a) wave- diameter of each subsurface eddy is about one half of the
number-2 and (Figure 9b) wavenumber-3 meanders, respec- initial meander wavelength. The subsurface anticyclones
21562202c, 2003, C7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2002JC001473 by Readcube-Labtiva, Wiley Online Library on [03/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19 - 10 CHAO AND SHAW: ARCTIC DENSE SHELF WATER OUTFLOW AND EDDY SHEDDING

Figure 8. Flow fields and isopycnals from experiment 4 at (a, b) day 15 and (c, d) day 20. Top panels
(Figures 8a and 8c) are at 212.5 m depth, while bottom panels (Figures 8b and 8d) are at 62.5 m depth.
The inflow-outflow profile is given at the bottom of Figures 8b and 8d. Contour values refer to
perturbation density in units of kg m3.

offshore are generally weak and shallow. With wavenum- 212.5 m depth and 0.33 f at 62.5 m depth. These anti-
ber-2 meander forcing, the maximum vorticity of offshore cyclones migrate offshore in time. They are more energetic
anticyclones (Figure 10a) is about 0.14 f at 212.5 m depth with wavenumber-3 forcing and consequently reach the
and 0.23 f at 62.5 m depth. Corresponding vorticity with offshore boundary before day 20. With wavenumber-2
wavenumber-3 forcing (Figure 10c) is about 0.14 f at meanders, these anticyclones reach the offshore boundary

Figure 9. Initial surface currents and surface perturbation density (in kg m3) with (a) wavenumber-2
meanders and (b) wavenumber-3 meanders. Inflow-outflow profiles given at the bottom of each panel
may or may not be used. See text.
21562202c, 2003, C7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2002JC001473 by Readcube-Labtiva, Wiley Online Library on [03/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
CHAO AND SHAW: ARCTIC DENSE SHELF WATER OUTFLOW AND EDDY SHEDDING 19 - 11

Figure 10. Flow fields and perturbation density contours (in kg m3) forced by (a, b) wavenumber-2
meanders at day 15 from experiment 5 and (c, d) wavenumber-3 meanders at day 10 from experiment 7.
Both experiments exclude density forcing from the shelf. Top panels (Figures 10a and 10c) are at 212.5 m
depth and bottom panels (Figures 10b and 10d) are at 62.5 m.

at about day 20. The signature of baroclinic instability is interactions between responses induced by sinking plumes
also unmistakable in Figure 10, as surface meanders gener- and by offshore meander forcing are examined below.
ally lag behind those at depths if one follows the main
direction of the mean jet. 3.4. Sinking Plumes and Offshore Meander Forcing
[29] This type of meander-induced instability, arising [31] Many types of interactions occur when both shelf
from differential advection, is well known in atmospheric input and offshore meander forcing occur simultaneously.
literature. Since advection of meanders is faster near the jet Our objective is not to consider all possibilities but rather to
axis, meanders on both sides of the jet tilt backward in time focus on the destructive and constructive processes leading
relative to the axial meander. The backward tilt of meanders to the production of halocline anticyclones. The inflow and
and eddies induces a transfer of eddy momentum fluxes up outflow are stationary, but the offshore meanders move with
the mean gradient or toward the jet axis, enhancing the jet the jet in time. The phase relationship between the two
speed. The so-called negative viscosity phenomenon [Starr, dictates much of the interaction. To search for the most
1968] need not be related to baroclinic instability as defined productive interaction, we have shifted the alongshore
in the classical sense of the term, but often becomes more profile of inflow-outflow a quarter of the wavelength at a
pronounced in the baroclinic regime because it involves time and monitored the strength of offshore anticyclones as
potential energy release. a function of the shift. The most constructive interaction
[30] The development of an undercurrent off the shelf occurs if the inflow-outflow profile is 180 out of phase
break is also noteworthy. Whether the meander forcing is with the initial meanders offshore, as we illustrated in
wavenumber-2 or wavenumber-3, a meandering under- Figure 9.
current against the initial baroclinic current develops on [32] Figure 11 shows subsurface (Figures 11a and 11c)
the shoreward side of the surface jet (Figures 10a and 10c). and near-surface features (Figures 11b and 11d) as a result
The undercurrent is part of the baroclinic response to of this interaction. Left-hand panels (Figures 11a and 11b)
shoreward intrusions of surface buoyant waters with the are wavenumber-2 solutions at day 15 while right-hand
meander (Figures 10b and 10d). A similar development was panels (Figures 11c and 11d) are wavenumber-3 solutions at
examined earlier in a flat-bottomed coastal ocean [Chao and day 10. Offshore anticyclones persist as in cases without
Shaw, 2002]; details will not be repeated here. We also note sinking plumes from the shelf (Figure 10). At 212.5 m
that sinking plumes from the shelf induce a similar under- depth, these anticyclones are delineated by isopycnals with
current off the shelf break (Figures 5a and 5c). Possible a perturbation density of 1.3 kg m3. Their diameter is
21562202c, 2003, C7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2002JC001473 by Readcube-Labtiva, Wiley Online Library on [03/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19 - 12 CHAO AND SHAW: ARCTIC DENSE SHELF WATER OUTFLOW AND EDDY SHEDDING

Figure 11. Flow fields and perturbation density contours (in kg m3) forced by (a, b) wavenumber-2
meanders at day 15 from experiment 6 and (c, d) wavenumber-3 meanders at day 10 from experiment 8.
Both experiments also include shelf inputs of dense plumes. The inflow-outflow profiles are shown at the
bottom of Figures 11b and 11d. Top panels (Figures 11a and 11c) are at 212.5 m depth, and bottom panels
(Figures 11b and 11d) are at 62.5 m.

about half of the meander wavelength as before. They are also of halocline anticyclones in an Arctic Ocean setting. These
shallow and weak as in cases without density forcing from the subsurface eddies appear to be important to halocline
shelf (Figure 10). To quantify further, the maximum vorticity ventilation, for they deliver shelf water parcels into the
for the offshore anticyclones is about 0.17 f at 212.5 m basin interior. The conceptual difficulty in establishing
depth (Figure 11a) and 0.22 f at 62.5 m (Figure 11b) with the linkage lies in the trapping mechanism introduced by
wavenumber-2 forcing. In the wavenumber-3 setting, the the continental slope. Sinking plumes from the shelf tend to
corresponding vorticity is about 0.13 f at 212.5 m depth be trapped by the continental slope, dispersing alongshore
(Figure 11c) and 0.34 f at 62.5 m (Figure 11d). Compared to following the propagation direction of topographic Rossby
cases without the shelf input (Figure 10), sinking plumes waves. The trapping mechanism is a strong dynamical
contribute little and sometimes negatively to the growth of constraint against seaward dispersal and must be overcome
offshore anticyclones. to open up the possibility for offshore migration of hal-
[33] Figure 11 is actually the best case scenario in terms of ocline anticyclones. Fixing the inquiry at this level, we
positive interactions between the two mechanisms. In other have conducted a series of simple numerical experiments to
cases the shifted inflow-outflow profile along the shelf leads point out a promising mechanism which could lead to
to even more negative interaction. Offshore anticyclones are seaward leakage of dense outflow and generation of
often annihilated by the negative interaction, especially if the halocline anticyclones offshore. A series of inflow and
inflow-outflow profile and offshore meanders are initially in outflow is imposed on the shoreward side of the shelf to
phase. The conclusion is not surprising because instabilities facilitate release of dense plumes into the basin. The dense
of shelf and offshore origins do not produce exactly com- plumes produce a subsurface density current trapped by the
patible disturbances that can interact constructively. continental slope as expected. A key catalyst to activate
seaward leakage is an offshore baroclinic current bounded
to the left by the shelf break. Isopycnals rise toward the
4. Discussion and Conclusions shelf break to support the baroclinic current. In conse-
[34] This paper seeks to establish a possible link between quence, convection along slanted isopycnals diverts sinking
dense water outflow from the shelf and offshore production plumes away from boundary trapping and, in our simple
21562202c, 2003, C7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2002JC001473 by Readcube-Labtiva, Wiley Online Library on [03/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
CHAO AND SHAW: ARCTIC DENSE SHELF WATER OUTFLOW AND EDDY SHEDDING 19 - 13

Figure 12. Cartoons showing different three-dimensional responses to forcing by sinusoidal inflow-
outflow (a) without and (b) with an initial baroclinic current offshore. In Figure 12a, tilted isopycnals
(dotted lines) induced by sinking plumes (short arrows) support a surface current (open arrows) bounded
to the left at the edge of the shelf and an undercurrent (solid curved arrows). In Figure 12b, solid curves
and solid arrows indicate near-surface isopycnals and currents, respectively. Corresponding features
below 200 m are dashed. Sinking plumes destabilize the baroclinic current offshore, producing large-
amplitude meanders intruding seaward at subsurface depths and halocline anticyclones in between. An
undercurrent is also generated adjacent to the continental slope.

model setting, produces intense halocline anticyclones presents a departure from the physical setting of the
offshore. Chukchi shelf. We argue for now that our model responses
[35] Figure 12 illustrates the three-dimensional relation- are primarily driven by the inflow rather than the outflow
ship between density and flow fields induced by sinking because only the former provides the potential energy input
plumes. The left panel (Figure 12a) illustrates the ocean needed to drive the system. While this argument is intui-
response induced by the sinusoidal inflow-outflow without a tively acceptable, it is still desirable to demonstrate gener-
baroclinic current offshore. Sinking plumes raise isopycnals ation of halocline anticyclones by a source without a
over the shelf break but depress them below the sinking compensatory return flow. We will demonstrate that this is
depth. The perturbed density field supports a surface current indeed possible in a paper which incorporates the effect of a
bounded to the left by the outer shelf, and an undercurrent submarine canyon [Shaw and Chao, 2003].
over the slope above the sinking depth. In this case, trapping [37] The lateral injection of dense plumes provokes
of sinking plumes by the continental slope is a strong baroclinic instability, and the instability in turn provides
dynamical constraint against seaward dispersal. The right pathways for the seaward dispersal of dense plumes and
panel (Figure 12b) illustrates the corresponding response offshore generation of anticyclones. For this mechanism to
with a baroclinic current offshore. Solid curves and solid be effective, sinking plumes must be primarily responsible
arrows indicate near-surface isopycnals and currents, respec- for baroclinic waves offshore. In the presence of active
tively. Corresponding features below 200 m are dashed. In baroclinic waves originated offshore, sinking plumes still
this case, sinking plumes destabilize the baroclinic current leak offshore but become ineffective in generating or
offshore, producing large-amplitude meanders intruding reinforcing subsurface anticyclones offshore.
seaward at subsurface depths. Trapped between subsurface [38] Other experiments using sinusoidal inflow-outflow
meander crests are subsurface anticyclones. Near-surface have been performed. The response over and beyond the
meanders lag somewhat behind subsurface meanders if we shelf break is found to be negligible if the inflow water has
follow the main direction of the mean current. Since waters the density of the ambient shelf water. Consequently, the
are heavier shoreward of a meander and lighter seaward of it, ocean response is driven by lateral injection of dense waters
the phase lag in upper depths ensures that the upper and rather than the mechanical energy input from the inflow-
lower halves of a halocline anticyclone are characterized by outflow. If the density of the inflow water is increased
positive and negative density anomalies, respectively. The slightly, sinking plumes reach greater depths, enhancing
result shows that sinking plumes add potential energy to the subsurface meander growth and anticyclone generation.
baroclinic current and trigger baroclinic instability, as sug- However, ultra-dense plumes, such as the hypersaline mode
gested in our earlier study [Chao and Shaw, 2002]. This water over the Chukchi shelf [Weingartner et al., 1998], will
process enhances seaward dense water dispersal and the penetrate through the halocline rather than being diverted
generation of subsurface anticyclones, suggesting a promis- seaward by the slanted isopycnal surfaces and therefore are
ing mechanism to ventilate the halocline. unable to generate intense halocline anticyclones offshore.
[36] Although the use of inflow and outflow serves the These experiments would suggest that the intermediate
purpose of releasing dense water onto the outer shelf, the mode water over the Chukchi shelf is the most effective
landward return flow imposed on the shoreward boundary in producing halocline anticyclones offshore. This tentative
21562202c, 2003, C7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2002JC001473 by Readcube-Labtiva, Wiley Online Library on [03/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
19 - 14 CHAO AND SHAW: ARCTIC DENSE SHELF WATER OUTFLOW AND EDDY SHEDDING

conclusion awaits future confirmation with more realistic Coachman, L. K., K. Aagaard, and R. B. Tripp, Bering Strait: The Regional
Physical Oceanography, 172 pp., Univ. of Wash. Press, Seattle, Washing-
sources of dense shelf water and topographic features. The ton, 1975.
compatibility between inflow width and the baroclinic D’Asaro, E. A., Observations of small eddies in the Beaufort Sea, J. Geo-
Rossby radius is also crucial to maximize offshore produc- phys. Res., 93, 6669 – 6684, 1988a.
tion of halocline anticyclones. This is conceivable because D’Asaro, E. A., Generation of submesoscale vortices: A new mechanism,
J. Geophys. Res., 93, 6685 – 6693, 1988b.
the Rossby radius is the scale of choice for patchiness. If Etling, D., F. Gelhardt, U. Schrader, F. Brennecke, G. Kuhn, G. C. d’Hieres,
sources are much wider than the first baroclinic Rossby and H. Didelle, Experiments with density currents on a sloping bottom in
radius, the ability to generate subsurface anticyclones is a rotating fluid, Dyn. Atmos. Oceans, 31, 139 – 164, 2000.
Gawarkiewicz, G., Effects of ambient stratification and shelf break topog-
reduced. Very narrow sources and sinks are not sufficiently raphy on offshore transport of dense water on continental shelves,
separated, and the array of dense plumes entering the shelf J. Geophys. Res., 105, 3307 – 3324, 2000.
tends to coalesce during the offshore migration, becoming Holland, W. R., The role of mesoscale eddies in the general circulation
model of the ocean: Numerical experiments using a wind-driven quasi-
more or less continuous in the alongshore direction. Since geostrophic model, J. Phys. Oceanogr., 8, 363 – 392, 1978.
patchiness is reduced, eddy shedding becomes less likely. In Jiang, L., and R. W. Garwood Jr., Three-dimensional simulations of over-
this extreme case, the oceanic response over and beyond flow on continental slopes, J. Phys. Oceanogr., 26, 1214 – 1233, 1996.
shelf break resembles those driven by a line source. Jiang, L., and R. W. Garwood Jr., Effects of topographic steering and
ambient stratification on overflows on continental slopes: A model study,
[39] The possible model dependence on the bottom slope J. Geophys. Res., 103, 5459 – 5476, 1998.
presents another complicating factor, since Arctic slopes Kikuchi, T., M. Wakatsuchi, and M. Ikeda, A numerical investigation of the
have large variations in the depth gradient. In general, a transport process of dense shelf water from a continental shelf to a slope,
J. Geophys. Res., 104, 1197 – 1210, 1999.
steeper slope enhances boundary trapping of density Lane-Serff, G. F., and P. G. Baines, Eddy formation by overflows in strat-
currents and therefore hinders seaward dispersal of dense ified water, J. Phys. Oceanogr., 30, 327 – 337, 2000.
water. In a rotating tank setting, for example, Lentz and Lentz, S. J., and K. R. Helfrich, Buoyant gravity currents along a sloping
bottom in a rotating fluid, J. Fluid Mech., 464, 251 – 278, 2002.
Helfrich [2002] have demonstrated faster propagation of Manley, T. O., and K. Hunkins, Mesoscale eddies of the Arctic Ocean,
density currents along the boundary if the bottom slope J. Geophys. Res., 90, 4911 – 4930, 1985.
increases, indicative of enhanced trapping by a steeper Marshall, J., A. Adcroft, C. Hill, L. Perelmam, and C. Heisey, A finite-
slope. In the present physical setting, however, the strong volume, incompressible Navier-Stokes model for studies of the ocean on
parallel computers, J. Geophys. Res., 102, 5753 – 5766, 1997.
stratification greatly reduces the effect of bottom slope to Melling, H., The formation of a haline shelf front in wintertime in an ice-
trap the dense outflow. In consequence, shedding of hal- covered arctic sea, Cont. Shelf Res., 13, 1123 – 1147, 1993.
ocline anticyclones is not visibly retarded if the bottom Muench, R. D., J. T. Gunn, T. E. Whitledge, P. Schlosser, and W. Smethie,
An Arctic Ocean cold eddy, J. Geophys. Res., 105, 23,997 – 24,006,
slope is steepened by a factor of 2 or so. 2000.
[40] While the present findings are encouraging, more Newton, J. L., K. Aagaard, and L. K. Coachman, Baroclinic eddies in arctic
efforts are needed along this line of investigation. Foremost ocean, Deep Sea Res., 21, 707 – 719, 1974.
Pacanowski, R. C., and S. G. H. Philander, Parameterization of vertical
of all, it is important to examine the relation between a single mixing in numerical models of tropical oceans, J. Phys. Oceanogr., 11,
source of dense shelf water and possible generation of 1443 – 1451, 1981.
halocline anticyclones offshore. Distinct topographic and Phillips, N. A., Energy transformations and meridional circulations asso-
coastline features must also be examined, because the pres- ciated with simple baroclinic waves in a two-level, quasi-geostrophic
model, Tellus, 6, 273 – 286, 1954.
ence of a baroclinic current may not be the only way to divert Pozrikidis, C., Introduction to Theoretical and Computational Fluid
dense plumes away from boundary trapping. Take the Chuk- Dynamics, 675 pp., Oxford Univ. Press, New York, 1997.
chi shelf as an example. The geometry of the active outflow Roach, A. T., K. Aagaard, C. H. Pease, S. A. Salo, T. Weingartner,
V. Parlov, and M. Kulakov, Direct measurements of transports and water
region is quite complex, including shoals, canyons, sharp properties through the Bering Strait, J. Geophys. Res., 100, 18,443 –
coastline bends, and seaward isobath protrusions. These local 18,457, 1995.
features are also potentially capable of diverting dense Shaw, P.-T., and S.-Y. Chao, Effects of a baroclinic current on a sinking
plumes away from boundary trapping. Our modeling efforts dense water plume from a submarine canyon and heton shedding, Deep
Sea Res., Part I, 50, 357 – 371, 2003.
to account for these additional mechanisms are still ongoing. Starr, V. P., The Physics of Negative Viscosity Phenomena, 256 pp.,
McGraw-Hill, New York, 1968.
[41] Acknowledgments. This study was partly supported by the Weingartner, T. J., D. J. Cavalieri, K. Aagaard, and Y. Sasaki, Circulation,
Arctic Sciences Section, Office of Polar Programs of the National Science dense water formation, and outflow on the northeast Chukchi shelf,
Foundation under grants OPP-9709952 and OPP-9614107. A discussion J. Geophys. Res., 103, 7647 – 7662, 1998.
with T. Weingartner helped initiate this study. Williams, G. P., Numerical integration of the three-dimensional Navier-
Stokes equations for incompressible flow, J. Fluid Mech., 37, 727 –
750, 1969.
References Yoshikawa, Y., K. Akitomo, and T. Awaji, Formation process of intermedi-
Aagaard, K., L. K. Coachman, and E. C. Carmack, On the halocline of the ate water in baroclinic current under cooling, J. Geophys. Res., 106,
Arctic Ocean, Deep Sea Res., 28, 529 – 545, 1981. 1033 – 1051, 2001.
Cavalieri, D. J., and S. Martin, The contribution of Alaskan, Siberian, and
Canadian coastal polynyas to the cold halocline layer of the Arctic Ocean,
J. Geophys. Res., 99, 18,343 – 18,362, 1994. 
Chao, S.-Y., and P.-T. Shaw, A numerical investigation of slanted convec- S.-Y. Chao, Horn Point Laboratory, University of Maryland Center for
tion and subsurface anticyclone generation in an Arctic baroclinic current Environmental Science, Cambridge, MD 21613-0775, USA. (chao@hpl.
system, J. Geophys. Res., 107, 3019, doi:10.1029/2001JC000786, 2002. umces.edu)
Chapman, D. C., and G. Gawarkiewicz, Offshore transport of dense shelf P.-T. Shaw, Department of Marine, Earth and Atmospheric Sciences,
water in the presence of a submarine-canyon, J. Geophys. Res., 100, North Carolina State University, Raleigh, NC 27695-8208, USA.
13,373 – 13,387, 1995. (pt_shaw@ncsu.edu)

You might also like