Hughes OM2000

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Ocean Modelling 2 (2000) 73±83

www.elsevier.com/locate/omodol

A theoretical reason to expect inviscid western boundary


currents in realistic oceans
Chris W. Hughes
CCMS-Proudman Oceanographic Laboratory, Bidston Observatory, Bidston Hill, Prenton, Merseyside CH43 7RA, UK
Received 22 August 2000; received in revised form 20 September 2000; accepted 3 October 2000

Abstract
The conventional picture of an ocean gyre, based on an ocean with vertical sidewalls, assumes a balance
between an input of vorticity by wind stress curl, and a viscous ¯ux of vorticity through the boundary at the
same latitude, resulting from a viscous western boundary current which may be signi®cantly modi®ed by
nonlinear terms. Potential interactions with topography are also commonly acknowledged as a possible
complicating factor. In this idealized picture, the zonal momentum balance is taken to be geostrophic, as
numerous model analyses con®rm. A theoretical argument is given here which shows that, in an ocean with
sloping sidewalls, this geostrophic balance results in bottom pressure torques which balance the wind stress
curl at each latitude. This removes the requirement for a viscous western boundary current at each latitude
suggesting that the dynamics within a western boundary current may be essentially inviscid. While inviscid
western boundary currents have already been found in certain idealized systems, and in one set of diag-
nostics from an eddy permitting model, the generality of the argument presented here gives a strong reason
to believe that these are not special cases. Inviscid western boundary currents are in fact the rule, and the
vertical sidewall case is an unrealistic exception. Ó 2000 Elsevier Science Ltd. All rights reserved.

Keywords: Sverdrup balance; Bottom pressure torques

1. Introduction

The broad picture of how an ocean gyre circulation works was drawn some 50 years ago by
Stommel (1948) and Munk (1950) based on an assumption of linear dynamics and no interaction
with a sloping ocean ¯oor except at vertical sidewalls. Under those conditions, it can be shown
that an ocean interior ¯ow in Sverdrup balance develops, in which the net meridional transport is

E-mail address: cwh@pol.ac.uk (C.W. Hughes).

1463-5003/00/$ - see front matter Ó 2000 Elsevier Science Ltd. All rights reserved.
PII: S 1 4 6 3 - 5 0 0 3 ( 0 0 ) 0 0 0 1 1 - 1
74 C.W. Hughes / Ocean Modelling 2 (2000) 73±83

determined by wind stress curl. Mass balance is achieved by a narrow return ¯ow in a western
boundary current in which viscous torques upset Sverdrup balance to permit a meridional return
¯ow.
Much attention has been given to the question of how nonlinear terms a€ect this balance,
mostly in barotropic models with ¯at ocean ¯oors and vertical sidewalls; a good summary is given
in Pedlosky (1996, Chapter 2). While nonlinear terms do signi®cantly alter the structure of the
western boundary current, they do not remove vorticity and, in fact, do not signi®cantly a€ect the
balance between wind stress curl and viscous torques at each latitude.
The question of topographic in¯uence, however, has received much less attention, although it is
commonly acknowledged as a possible complication. Holland (1973) noted that a continental
shelf slope could induce a recirculation in the Gulf Stream under certain circumstances, and
Wunsch and Roemmich (1985) suggested that topographic interactions are necessary to explain
observations in the North Atlantic. More recently, a number of authors have noted the existence
of idealized systems in which essentially inviscid western boundary currents occur, steered by
topography (Salmon, 1992; Straub et al., 1993; Salmon, 1994; Becker and Salmon, 1997; Griths
and Veronis, 1997, 1998; Becker, 1999), although in fact the simplest model (linear and homo-
geneous) may be easily deduced from a thermal analogy constructed by Welander (1968). Inviscid
western boundary currents have also been found in the Paci®c in diagnostics from an eddy per-
mitting model (Saunders et al., 1999). The question thus arises, are these examples of inviscid
western boundary currents the result of unrealistic idealizations, or special, regional dynamics,
and not particularly relevant to mainstream ocean theory, or is it the viscous boundary current of
the vertical sidewall case which is special and unrealistic?
The real ocean, of course, does have sloping sidewalls, but does not conform to the idealiza-
tions of the above studies. However, one approximation which is true of both idealized studies
and the real ocean is that the zonal momentum balance (outside the surface Ekman layer) is close
to geostrophic in a zonal integral. Since any mass ¯ux driven northwards in the Ekman layer must
be returned to the south (in geostrophic balance) at some depth, this means that the zonal integral
of zonal wind stress must be balanced by east±west pressure di€erences across the ocean basin, or
across topographic features, i.e., wind stress must be balanced by form stress at each latitude. This
is explicitly assumed in most analytical studies, in which nonlinear and viscous terms may be
important in the longstream momentum balance of boundary currents, but the cross-stream
balance and interior dynamics are geostrophic below the Ekman layer. It is also con®rmed from
modelling studies (Ponte and Rosen, 1994; Stevens and Ivchenko, 1997; Gille, 1997; Bryan, 1997)
and by implication (from the smallness of nonlinear terms) from observations (Oort, 1985; Bryden
and Heath, 1985; Morrow et al., 1994) for realistic ocean conditions. Most of these studies
concentrate on the Southern Ocean, where this geostrophic balance is a particular issue ®rst
mooted by Munk and Palmen (1951), but it is clear that the balance occurs at latitudes other than
those of Drake Passage. Indeed, the Southern Ocean is the focus of investigation because it is the
region where the balance seems most likely to be violated. The diagnostics from Gille (1997) are
reproduced here (Fig. 1), rescaled so as to show zonally integrated rather than averaged quan-
tities, for reasons which will be made clear later.
The purpose of this paper is to show that a geostrophic balance in the zonal integral of the
zonal momentum equation (below the wind-driven Ekman layer) must result in bottom pressure
torques which upset the Sverdrup balance, in an ocean with sloping sidewalls. Furthermore, those
C.W. Hughes / Ocean Modelling 2 (2000) 73±83 75

Fig. 1. Zonally integrated zonal stresses diagnosed from an eddy-permitting model (values taken from Gille (1997)),
showing (top panel) zonal wind stress (dotted) and form stress (solid), and (bottom panel) the residual on an exag-
gerated scale. Units are kg sÿ2 , equivalent to N mÿ1 . Divide by f to get the related (Ekman or geostrophic) meridional
mass transport.

bottom pressure torques balance the wind stress curl in a zonal integral (exactly if the geostrophic
balance is exact, but in practice the e€ect of nonlinear terms means that this balance only holds in
an average over a few degrees of latitude). This shows that the torque required to balance bV in
western boundary currents can be supplied by the bottom pressure torque, removing the need for
viscous torques except in the special case in which the topography is con®ned to a sidewall region
narrower than a Munk viscous boundary layer. In this special case, the boundary current cannot
¯ow above the region of bottom pressure torques, and viscosity is forced to play an important
r^
ole. In more realistic situations, with wide continental shelf regions and a continental slope which
is not vertical, ¯ow can occur above the topography, where the required bottom pressure torque
occurs, and we should expect essentially inviscid dynamics. Thus, inviscid western boundary
currents are the rule rather than the exception, it is the vertical sidewall case which is special and
unrealistic.

2. Sverdrup balance plus topography

The generalized form of Sverdrup balance for an ocean with bottom topography may be
deduced by taking the curl of the depth-integrated momentum equations. Starting with the
momentum equations
76 C.W. Hughes / Ocean Modelling 2 (2000) 73±83

qu†t ‡ qf k  u ˆ ÿrp ‡ sz ‡ a ‡ b: 1†
Here, bold font denotes two-dimensional vectors in the horizontal plane, except for k which is the
unit vector in the local vertical (upwards). Subscripts represent derivatives (t for time and z the
vertical coordinate), q the density, u the velocity, s the viscous stress transmitted vertically, a
represents terms nonlinear in u, b the force due to lateral viscous stress, and p is the pressure minus
a spatially and temporally constant atmospheric pressure (varying atmospheric pressure can be
accounted for but makes little di€erence while complicating the equations slightly). The Coriolis
parameter is f ˆ 2X sin /, where X is the angular rate of rotation of the earth, and / is the
latitude.
Integrating from ocean bottom (z ˆ ÿH ) to surface (z ˆ g), and taking the gradient operator
outside the integral in the pressure term, gives
Ut ‡ f k  U ˆ ÿrP ‡ pb rH ‡ s0 ‡ A ‡ B; 2†
Rg
where U; P ; A; B† ˆ ÿH qu; p; a; b† dz, s0 is the surface stress minus the bottom stress, usually
dominated by the wind stress, and pb is the bottom pressure (minus the constant atmospheric
pressure).
Taking the curl of (2) gives the barotropic vorticity equation
r  Ut ‡ kr  f U† ˆ rpb  rH ‡ r  s0 ‡ r  A ‡ B†: 3†
Assuming a steady state, and no mass sources or sinks (r  U ˆ 0), this reduces to
bV ˆ k  ‰rpb  rH ‡ r  s0 ‡ r  A ‡ B†Š; 4†
Rg
where b is the meridional gradient of f, and V ˆ ÿH qv dz. This represents Sverdrup balance
(bV ˆ k  r  s0 ), with possible interference from bottom pressure torques, nonlinear terms, and
lateral friction (note that s0 is the di€erence between wind stress and bottom friction, so the e€ect
of bottom friction is included here).
Some comment is perhaps warranted here on the use of the term ``torque'' to describe terms on
the right-hand side of (4). The term ``bottom pressure torque'' is commonly used in the literature
to describe r  pb rH† ˆ rpb  rH , although it has units of N mÿ3 , rather than N m as ap-
propriate to a torque. The reason for this may be an intuitive notion that the curl of a force is a
measure of the twisting due to that force, which is something like a torque. Actually, there is some
justi®cation for the term if the angular momentum balance of a vertical cylinder of ¯uid is
considered, over which r  pb rH † is constant. The torque exerted on the cylinder is given by
pr4 r  pb rH†=4, where r is the radius of the cylinder, so each of the terms in (4) can be in-
terpreted as 4=p†  torque about the local vertical axis, per unit (radius4 ). It is also worth noting
that the bottom pressure torque is a quite di€erent thing from the joint e€ect of baroclinicity and
relief (JEBAR) term which arises upon taking the curl of the depth averaged rather than depth
integrated momentum equation. The distinction is made most clearly by Mertz and Wright (1992).
For a ¯at bottomed ocean (H constant), or an ocean in which no currents penetrate to the
bottom (pb is a function of H), the bottom pressure torque is zero everywhere in the ocean in-
terior, although there can be a delta function on the vertical sidewalls. In that case, if a zonal
integral is performed around either a closed latitude circle, or from coast to coast in a basin closed
to the north or south, bV must integrate to zero by conservation of mass to the north or south of
the latitude line, leaving a balance between the zonal integrals of wind stress curl, lateral viscous
C.W. Hughes / Ocean Modelling 2 (2000) 73±83 77

stress curl, and curl of the nonlinear term. Pedlosky (1996, Section 2.11) shows why the integral of
the nonlinear term should not be expected to be large in this situation, leaving a balance between
(wind-bottom) stress curl and lateral friction in a zonal integral. This is the reason why so much
attention has been given to the r^ ole of friction in western boundary currents. As has been well
rehearsed in the literature, and summarized in Pedlosky (1996, Chapter 2), friction can only be
1=3
signi®cant in a boundary layer of width given by the Munk scale dM ˆ AH =b† , where AH is the
lateral viscous di€usivity, meaning that a lateral di€usivity of about 2500 m2 sÿ1 is required for
viscous control of a 50 km wide current. Such a large di€usivity can only be justi®ed as an eddy
parameterization, and even then, as eddy activity is suppressed close to a sidewall, it is hard to
justify such large values at the ocean boundary. Nonetheless, since the nonlinear term cannot
contribute signi®cantly to the zonal integral of (4), the perception has persisted (true in the ¯at
bottom case) that the viscous torque r  B must be important at each latitude. The justi®cation
for this is derived not from a scaling of the term, but from the fact that it is the only term left to
balance r  s0 in a zonal integral.
When the ocean does not have vertical sidewalls, the bottom pressure torque need not be
concentrated in a delta function at the boundary. Very steep walls, admitting topographic slopes
only over a region narrower than the viscous Munk layer would not be expected to make much
di€erence to the solution, but the more realistic situation of a shelf slope broader than a Munk
boundary layer has the potential to make a signi®cant di€erence. An along-slope pressure gradient
equivalent to 1 cm of water for every 100 km along slope, on a slope of 1 in 100 typical of the
continental slope, produces a bottom pressure torque large enough to drive a northward volume
¯ux of 25  106 m3 sÿ1 for every 50 km of current width, so quite plausible bottom pressure
variations can provide the torque required by a typical western boundary current. The width of
the current then would be set by the scale of the slope region along which pressure is varying,
making a 50 km width quite plausible. However, we do not have to rely on ``plausible'' values for
the scaling of the along-slope pressure gradient, we can do better by taking values from the zonal
momentum balance at each latitude. Since bottom pressure di€erences balance the zonal wind
stress in this balance, it is easy to show, as will be demonstrated below, that the zonal integral of
the pressure torque balances the zonal integral of the wind stress curl in realistic conditions,
meaning that the zonally integrated bottom pressure torque is exactly that which is required to
balance bV in the western boundary current. With vertical sidewalls, this pressure torque is
concentrated in a delta function at the boundary, a singular situation requiring a Munk boundary
layer, but in more realistic circumstances the need for a viscous boundary layer is no longer clear,
and we should expect the torque and resulting current to be distributed along the continental
slope, where the western boundary current is actually found (although there are circumstances in
which deeper topography could be involved).
The connection between the boundary current dynamics and zonally integrated dynamics can
be made more explicitly by considering the case of an ocean with an interior in Sverdrup balance,
and a western boundary current. In this case, r  s0 ˆ bV in the interior, but mass conservation
means that ÿbV integrated across the western boundary current must equal bV integrated zonally
across the ocean interior, so the integral of ÿbV across the boundary current is equal to the zonal
integral of r  s0 . In order to have a dynamical balance in the western boundary current, bV here
must be balanced by one of the other terms in the barotropic vorticity equation (4), such as
bottom pressure torque or r  B, which must be negligible in the ocean interior if Sverdrup
78 C.W. Hughes / Ocean Modelling 2 (2000) 73±83

balance is not to be upset. Thus, whichever term balances bV in the western boundary current,
that term must equal ÿr  s0 in a zonal integral.
The converse is also true in this situation. If Sverdrup balance holds in the interior then only
r  s0 and bV can be signi®cant here. This means that, whichever term balances r  s0 in a zonal
integral, it must be acting in the western boundary current region. So, if bottom pressure torque
balances wind stress curl in a zonal integral, it must operate in the western boundary current and
balance bV here. Again, the vertical sidewall case is special since this makes the bottom pressure
torque act in a delta function at the boundary. It still balances bV when integrated across the
boundary current, but viscosity is needed to e€ectively spread out the delta function to a ®nite
width.
More generally, bottom pressure torques could act at other longitudes too, upsetting Sverdrup
balance in the ocean interior, but viscosity and nonlinear terms need not be important unless the
resulting ¯ows have strong enough shear or relative vorticity advection. As we shall see later, the
nonlinear term is locally important, but its e€ect is greatly reduced if an average is taken over a
few degrees of latitude. I will now show how the balance between zonal wind stress and form
stress in the zonal integral at each latitude, implies a balance between the zonal integrals of r  s0
and bottom pressure torque.
Consider the zonal integral of the depth-integrated momentum equation (2). The integral is
taken either around a closed latitude line, or from coast to coast. Since the terms in (2) involve
depth integrals of ®nite quantities, and the depth goes to zero (g ˆ ÿH ) at the coast, all terms are
zero at the coast. The zonal integral of Px is therefore zero and by conservation of mass, as for the
barotropic vorticity equation, the integral of V is zero, leaving for the steady-state balance
Z E
0ˆ pb Hx ‡ sx0 ‡ Ax ‡ Bx † dx; 5†
W

where dx is shorthand for R cos / dk, with R the radius of the earth and k is the longitude, and a
superscript x represents the zonal component of a vector. In this balance we know, as illustrated
in Fig. 1, that the main terms are wind stress sx0 , and bottom form stress pb Hx , with model di-
agnostics showing the nonlinear term to be much smaller, and the lateral friction smaller again
(Gille, 1997). However, each term in (5) is related to the zonal integral of the corresponding term
in (4) by Stokes' theorem, so the terms which are important in the zonal integral of (4) must be the
same terms which dominate in (5). To see this, consider integrating a term in (4) over the area
illustrated in Fig. 2: a zonal strip bounded by two latitude lines and two sections of coastline.
Since each term in (4) represents the curl of a corresponding term in the depth-integrated mo-
mentum equation (2), we can use Stokes' theorem to write the area integral of that curl as a line
integral over the bounding contour, for example for (wind-bottom) stress,
Z Z
r  s0 †  dS ˆ s0  ds: 6†
C dC

However, dC consists of two zonal sections, and two sections along which the depth is zero
(meaning that each of the terms in (2) is zero), so we can write
Z Z Z
x
r  s0 †  dS ˆ s0 dx ÿ sx0 dx; 7†
C /1 /2
C.W. Hughes / Ocean Modelling 2 (2000) 73±83 79

Fig. 2. Schematic depicting a zonal strip of ocean area, C, bounded by a closed contour dC which consists of two zonal
sections at latitudes /1 and /2 , and two sections of coastline.

and similarly for the bottom pressure torque


Z Z Z
rpb  rH †  dS ˆ pb Hx dx ÿ pb Hx dx 8†
C /1 /2

with similar relations for the other terms, where /1 is the southern latitude bounding area C and
/2 is the northern latitude. We immediately recognize the right-hand sides of Eqs. (7) and (8) as
terms from (5), the zonal integral of the zonal momentum equation, evaluated at the two
bounding latitudes /1 and /2 , and di€erenced.
If the balance in (5) is purely between wind stress and bottom form stress, at each latitude (in
other words, if the Ekman transport across each latitude is balanced by a purely geostrophic
return ¯ow at depth), then
Z Z Z Z
x x
s0 dx ÿ s0 dx ˆ pb Hx dx ÿ pb Hx dx;
/1 /2 /1 /2

so, from (7) and (8),


Z Z
r  s0 †  dS ˆ rpb  rH †  dS;
C C

and the area integral of (4) has no contribution from lateral friction or nonlinear terms. The zonal
line integral of (4) is just the limit of an area integral as the width of the strip tends to zero. In this
limit, the zonal integral of each term in (4) is simply d=dy of the corresponding term in (5). A
balance between wind stress and form stress at each latitude implies, by a mathematical identity, a
balance between wind stress curl and bottom pressure torque at each latitude.
In practice, though, it makes sense to consider a zonal strip of ®nite thickness. The reason is that,
although the nonlinear term in the zonally integrated zonal momentum balance is small compared
to wind stress and bottom form stress, it varies with latitude much more rapidly than wind stress
(see Fig. 1). For a very narrow strip, the area integral of (4) is predominantly a balance between
bottom pressure torque and nonlinear terms (this is shown explicitly from model diagnostics by
Wells and de Cuevas (1995), it is only when the strip becomes wide enough (a few degrees of
latitude) that the wind stress curl integral becomes larger than the nonlinear torque, and the
balance becomes dominated by wind stress curl and bottom pressure torque on this and all larger
80 C.W. Hughes / Ocean Modelling 2 (2000) 73±83

scales. If the question to be answered is ``what balances wind stress curl in a zonal integral'', the
answer is that this is only a meaningful question on a relatively large meridional length scale, and
on that length scale it is the bottom pressure torque which balances it. On shorter length scales,
wind stress curl is a small term in the balance compared with a dominant balance between bottom
pressure torque and the nonlinear term. The reason for plotting Fig. 1 in terms of zonal integral
rather than zonal average terms should now be clear. In this form, the accuracy of the balance
between wind stress curl and bottom pressure torque can be read o€ directly from the ®gure. Over a
zonal strip bounded by two latitude lines, the wind stress curl integral is given by the di€erence
between the wind stress integral in Fig. 1 at the two bounding latitudes, and similarly for the
bottom pressure torque integral. Clearly these di€erences do not balance for very narrow strips, for
which the nonlinear term is important, but do balance for strips a few degrees of latitude wide, or
wider. The zonal line integral of terms in (4) is simply the slope of the corresponding curve in Fig. 1.
To summarize, for a zonal strip as in Fig. 2, the area integral of each term in the barotropic
vorticity equation (4) is identically equal to the di€erence in the zonal integrals of the corre-
sponding term in the depth- and zonally integrated zonal momentum equation (5), as evaluated at
the two bounding latitudes of the zonal strip. Since the two terms which balance in (5) are the
wind stress and bottom form stress terms, the terms which must balance in the area integral of (4)
are wind stress curl and bottom pressure torque, at least for strips wide enough that the di€erence
in zonal wind stress from north to south is signi®cant. Broadly speaking, wind stress curl is
balanced by bottom pressure torque, as long as the zonal wind stress is balanced by bottom form
stress. This is equivalent to the requirement that the net Ekman transport across a latitude line,
driven by wind stress, be returned in a geostrophic ¯ow at some depth. This requirement is met by
all the analytical and numerical ocean models of which I am aware.

3. Discussion

Although the analysis above shows that wind stress curl is balanced by bottom pressure torque
at each latitude, given some averaging in the meridional direction, it says nothing about the
longitude at which the bottom pressure torque occurs. Given that torques are balanced by bV , it
seems sensible to assume that the bulk of the bottom pressure torque occurs in the western
boundary current, where bV is largest, but there is no a priori reason to limit the pressure torque
to occur on the west of the ocean. In fact, in a barotropic ocean which gets deeper towards the
pole (so that the gradient of f =H in the ocean interior is opposite to the gradient of f), the cir-
culation produces a strong eastern boundary current (Welander, 1968) (although that solution
also assumes vertical sidewalls, and has bottom friction but no lateral friction).
When topography is permitted, the arguments against a signi®cant role for nonlinear terms in
the zonal integral balance also fail. In this case it is dicult to predict the size of nonlinear terms
from ®rst principles, but model diagnostics can help. We see the local importance of nonlinear
terms from Fig. 1, and the fact that the slopes of the two curves do not match at the shortest scales.
However, we also see that the slopes do match when averaged over a few degrees of latitude,
meaning that the integrated e€ect of nonlinear terms becomes negligible over these length scales.
Thus we should expect to ®nd that nonlinearity is locally very important in the barotropic vorticity
equation (4), but with an e€ect which integrates to zero over a typical length scale of a few degrees.
C.W. Hughes / Ocean Modelling 2 (2000) 73±83 81

This is precisely the balance which was diagnosed by Saunders et al. (1999) for western boundary
currents in the Paci®c Ocean, from the global quarter degree ocean circulation model OCCAM. The
balance integrated over the area of each boundary current was shown to be dominated by bV and
bottom pressure torque, with the nonlinear term also contributing signi®cantly in some cases (the
biggest contribution is 37%, in the Kuroshio, for an average over 3° of latitude by 2° of longitude).
The lateral viscous torque is only signi®cant in the deep western boundary current at the Samoan
Passage (a very small region, 1.5° of latitude by 0.75° of longitude centred on about 190°E, 9.5°S), in
which region it balances 37% of bV . In none of the other western boundary current regions does the
lateral viscous torque exceed 10% of bV , and in several cases it is of the wrong sign to balance bV .
The argument given above shows why this result should not be a surprise, and should not be in-
terpreted as a quirk in a particular region of a particular model. This result is in fact what should be
expected of any model in which the zonal momentum balance is dominated by wind stress and form
stress, and topographic slopes are wider than Munk boundary layers.
One ®nal complication which should be addressed is the e€ect of a longshore wind stress. The
quantity which has been used throughout this paper is s0 , which represents the di€erence between
wind stress and bottom stress. It was argued earlier that all the terms in (2) tend to zero at the
coast, since they are depth integrals of ®nite quantities and the depth tends to zero. As s0 is zero at
the coast, this means that the wind stress must be balanced by bottom stress as the depth tends to
zero (otherwise an in®nite acceleration would result). Clearly, then, it is not possible to neglect the
e€ect of bottom stress in a coastal boundary layer.
This is not a problem if the wind stress at the coast is directed exactly on or o€ shore, since it is
only the longshore component of s0 which is considered when connecting the zonal integral (5) to
the area integral of the barotropic vorticity equation (4) via Stokes' theorem. In many cases of
theoretical or idealized modelling studies, the wind stress is contrived to have no longshore
component, so the e€ect of this coastal boundary layer is negligible. In reality, of course, there will
be a longshore wind stress near the coast so we must adjust what we mean by ``wind stress curl''
slightly to include the e€ect of these boundary layers. This is signi®cant when considering a global
integral of wind stress curl, since a global integral of r  s0 must be zero, so any global integral of
wind stress curl must be balanced by a global integral of bottom stress curl which, however, can be
con®ned completely to the action of bottom stress in shallow boundary layers. Although this
implies a r^ole for viscosity in balancing wind stress curl, it is only the global integral (i.e., long-
shore wind stress at the coast) which need be balanced this way, and this clearly has nothing to do
with western boundary current dynamics as it implies a coastal boundary layer at all coasts, east,
west, north or south. Strictly, though, I should not say that wind stress curl is balanced by bottom
pressure torque, but that (wind-bottom) stress curl is balanced by bottom pressure torque, with
wind stress and bottom stress being equal at the coast, although the distinction only matters when
there is a longshore wind stress (or when the zonal integral of bottom stress is important as in the
atmosphere, or in unrealistic models of the Southern Ocean without topography).

4. Summary

A theoretical argument (essentially Stokes' theorem) shows that, if zonal wind stress is balanced
by bottom form stress in a zonal integral at each latitude (a generally accepted assumption
82 C.W. Hughes / Ocean Modelling 2 (2000) 73±83

supported by theoretical and modelling studies, and by inference from observations), then wind
stress curl must be balanced by bottom pressure torque rather than lateral viscous torques, when
integrated over a zonal strip. The width of the strip depends on the size and meridional length
scale of the residual imbalance between wind stress and form stress, but model diagnostics suggest
a few degrees (or any wider strip) should be sucient. For narrower strips, nonlinear terms are
expected to become important, together with the bottom pressure torque, and the lateral viscous
torque could also have a local e€ect. The lateral viscous torque is only expected to be the dom-
inant term in the ocean interior in the rather special case where the topography is con®ned to a
region narrower than a Munk boundary layer, in a near-vertical sidewall. Even then it is the
bottom pressure torque which dominates when the sloping region is included in the zonal integral.
This suggests that the barotropic vorticity balance in western boundary currents is essentially
inviscid in realistic ocean models, in accord with diagnostics from one such model. Thus the
apparently special, idealized studies discussed in the introduction, of systems with inviscid western
boundary currents, are in fact less special than the commonly used assumption of vertical side-
walls and viscous western boundary currents.
Rather than expecting viscous western boundary currents, and seeing inviscid examples ex-
amples as an occasional exception, we should expect inviscid western boundary currents and
question the realism of any system which displays a balance between bV and viscous torques in
the western boundary current.

Acknowledgements

Many people have read and commented on versions of this paper, thanks to all, and to the
referees for constructive comments. This work was supported by the UK Natural Environment
Research Council, and the UK Defence Research Agency.

References

Becker, J.M., 1999. E€ect of a western continental slope on the wind-driven circulation. J. Phys. Oceanogr. 29, 512±518.
Becker, J.M., Salmon, R., 1997. Eddy formation on a continental slope. J. Mar. Res. 55, 181±200.
Bryan, F.O., 1997. The axial angular momentum balance of a global ocean general circulation model. Dyn. Atmos.
Oceans. 25, 191±216.
Bryden, H.L., Heath, R.A., 1985. Energetic eddies at the northern edge of the antarctic circumpolar current in the
southwest paci®c. Prog. Oceanogr. 14, 65±87.
Gille, S.T., 1997. The southern ocean momentum balance: evidence for topographic e€ects from numerical model
output and altimeter data. J. Phys. Oceanogr. 27, 2219±2232.
Griths, R.W., Veronis, G., 1997. A laboratory study of the e€ects of a sloping side boundary on wind-driven
circulation in a homogeneous ocean model. J. Mar. Res. 55, 1103±1126.
Griths, R.W., Veronis, G., 1998. Linear theory of the e€ect of a sloping boundary on circulation in a homogeneous
laboratory model. J. Mar. Res. 56, 75±86.
Holland, W.R., 1973. Baroclinic and topographic in¯uences on the transport in western boundary currents. Geophys.
Fluid Dyn. 4, 187±210.
Mertz, G., Wright, D.G., 1992. Interpretations of the JEBAR term. J. Phys. Oceanogr. 22, 301±305.
C.W. Hughes / Ocean Modelling 2 (2000) 73±83 83

Morrow, R., Coleman, R., Church, J., Chelton, D., 1994. Surface eddy momentum ¯ux and velocity variances in the
southern ocean from geosat altimetry. J. Phys. Oceanogr. 24, 2050±2071.
Munk, W.H., 1950. On the wind-driven ocean circulation. J. Meteorol. 7, 79±93.
Munk, W.H., Palmen, E., 1951. Note on the dynamics of the Antarctic Circumpolar Current. Tellus 3, 53±55.
Oort, A.H., 1985. Balance conditions in the earth's climate system. Adv. Geophys. A 28, 75±98.
Ponte, R.M., Rosen, R.D., 1994. Oceanic angular momentum and torques in a general circulation model. J. Phys.
Oceanogr. 24, 1966±1977.
Pedlosky, 1996. Ocean Circulation Theory. Springer, Berlin, 453 pp.
Salmon, R., 1992. A two-layer Gulf Stream over a continental slope. J. Mar. Res. 50, 341±365.
Salmon, R., 1994. Generalized two-layer models of ocean circulation. J. Mar. Res. 50, 341±365.
Saunders, P.M., Coward, A.C., de Cuevas, B.A., 1999. Circulation of the paci®c ocean seen in a global ocean model:
ocean circulation and climate advanced modelling project (OCCAM). J. Geophys. Res. 104, 18281±18299.
Stevens, D.P., Ivchenko, V.O., 1997. The zonal momentum balance in an eddy-resolving general-circulation model of
the Southern Ocean. Quart. J. R. Meteor. Soc. 123, 929±951.
Stommel, H., 1948. The westward intensi®cation of wind-driven ocean currents. Trans. Amer. Geophys. Union 29, 202±
206.
Straub, D.N., Killworth, P.D., Kawase, M., 1993. A simple model of mass-driven abyssal circulation over a general
bottom topography. J. Phys. Oceanogr. 23, 1454±1469.
Welander, P., 1968. Wind-driven circulation in one- and two-layer oceans of variable depth. Tellus 20, 1±15.
Wells, N.C., de Cuevas, B.A., 1995. Vorticity dynamics of the southern ocean from a general circulation model. J. Phys.
Oceanogr. 25, 2569±2582.
Wunsch, C., Roemmich, D., 1985. Is the north atlantic in Sverdrup balance?. J. Phys. Oceanogr. 15, 1876±1880.

You might also like