-.

Truth is much too complicated to allow anything but lohn von Neuma.n (1901-19s7).

ap\oximation!

CHAPTER

THREE

Shallow Water Systems and lsentropic

Coordinates

describe a thin layff of constant density fluid in hydrostatic balance, rotating or not, bounded from below by a dgial surface and ftom above by a free suface, above which we suppos is another fluid of negligibl inertia. Such a configuration can b generalized to muttiple layers of tnmiscible
oNVENTToNALLY, 'TH E SHALLoW WATER EeUATIoNS

fluids lying one on top of anothr, forming a'stacked shallow water' system, and this class of svsrcms rs rhe main sublecr of rhi\ chapler.
Th single{ayer model is one of the simplest useful models in geophysical fluid al}rnamics, bcause it allows for a consideration of the effcts of rotation in a simple framework wittrout

the complicating effcts of stratification. By adding layers w can subsequently study the effects of stratification, and the model with just two layers is not orily a simple model of a stratified fluid, it is a suDrisingly good model of many phenomena in the ocean and aturcsphere. Indeed, the modls are more thanjust pedagogical tools we will find thar therc is a close physical and mathematical anatogy between the shatlow water equations and a description of the continuously statifid ocean or atmosphere $.ritten in isopycnal or ismtropic coordinates, lyirh a meaning beyond a cohcidntal similarity in the equations. We begin with the single'layI case.

1I

DYNAIVICS OF A SINCLE, SHALLOW LAYER

Shallow water dFamics apply, by definition, ro a fluid tayer of constant density in which the horizontal scale of rhe flow is much grearer rhan the layer depth. The fluial motion is fully detemined by the momentum and mass contiluity equations, and because of the assumed

t23

124

Chapter

3.

ShallowWater Systems and lsentropic Coordinates Fluid

l.
Topography

Fig. 3.1 A shallowwater system. h is the thickness of a water cotumn, I{ irs mean thickness. 4 the height of rhe fre surfac and 4, is the heisht ofthe tower, riqid, surface, above some arbitrary origin! typicalty (hosen such that the average of 4, is zero. A4 is the deviation free surface.height, so we have 4 = nb + h = E + An.

small aspct mtio the hydrostatic approximation is welt satisfied, anal we invoke this from the outset. Consider, then, fluid in a container above which is another fluid of negligibl dnsity (and therefore negligible inertia) retative to the fluid of interest, as illustrated in Fig. 3.1. As usual, our notation is that r = x.i + t/j + xrk is the thee dimensional vetocity and r, = xri + ,j is the horizontal vlocity. I!(r,Jr') is the thickness of the liquid cohjrnn, H is its mean height, and 4 is the height of the free surface_ In a flat-bottomed container 4 = l'!, whereas in general h = 4 4r, where t, is the height of the floor of the container. 3.1.1 Momentum quations
The vertical momentum equation is just the hydrostatic equation,

ap
and, because density is assumed constant, we may inte$at this to

(3.1)

p(x,y.zl = pg2 + po.

(3.2)

At the top of the fluid, z : r?, th pressure is detmined by the wighr of the overlying fluid and this is as$med to be negtigible. Thus, p = 0 at z : r| giving

p\x, y, z) = pgh\x,y ) - z).

(3.3)

The consequenc of this is that rhe horizontal gradient of plessule is independenr of height. That is
(3.4)

_ .a

.a

(3.s)

!
3.1 Dynamics of
a Single, Shallow Layer

I25

is the gradient opemtor at constant z. (In the rest of this chapter we will drop th subscript 2 ur ess that causes ambiguity. The three-dimensional gradient operator will be denoted by V3. we will also mostly use cartesian coordinates, but the shallow water equations indeed, 'Laplace's tidal equations' are may certainly be applied over a spherical planet essentially the shallow water equations on a sphre.) The horizontal momntum equations therefore become

DxI

w= pvP=-Svn

(3.6)

The nght-hand side of this equation is independent of the verocal coordinate z. Thus, if the flolv is initially independent of z, it must stay so. (This z independence is unrelated to that arising from the rapid rotation necessary for the Taylor-houdman effct.) The velocities u and u are functions of r., j'l and t on1y, and th horizontal momentum equation is therefore

Dx au au au or=ai+ t+r/aw:-.4v4.

(3.7)

That the horizontal velocity is indpendent of z is a consequence of tie hydrostatic equation, which ensures that the horizontal pressure gradient is hdependent of height. (Another starting point would be to take this independence of the honzontal motion with height as the defnihbn of shalow water flow. In real physical situations such independence does not hold exactly for example, ftiction at the bottom may induce a vertical dependence ol the flow in aboundary layer.) ln the presence of rotation, (3.7) easily genralizes to

Du, /u: 6l+r

wher

svn

(3.8)

Iatitude, so that on a spherical planet / : 3.1,2 Mass continuity quation

/k.

Just as with the primitive equations,

2r, sin

may be constant or may vary with and on the r-plane = /0 + p/.

Fron firct pfinciples

The mass contained in a fluid column of height h and cross sectional area A is given by Ia p h .lA (see Fig. 3.2). If thr is a net flux of fllltd across the colunm boundary &v advection) thn this must be balanced by a nt increase in the mass in A, ard therefore a net increase in the height of th water colurnn. The mass converyence into th column is given by

l-

= mass flux in

. dS, J lu

(3.9)

where S is the area of the vertical boundary of the column. The strfac area of the column is romposed of elements of area ,xn 51, where d, is a line elment ctcumscribing the colulm and n is a unit vector perpendicular to th boundary, pointing outwards. Thus (3.9) becomes

F^
F^ = -

I
|

Phu

dl'

(3.10)

Using the divergence theorem in two dimensions, (3.10) simplifies to

Iav

.(puhi M,

(3.11)

126

Chaptr

3.

Shallow Warer Systems and tsentropic Coordinates

Fis.3.Z The mass budget for a column of area A in a shallow wate. sys tem. The fluid leaving th column is

{phu ^dI \ hete r is the unit vec, tor normalto the boundary ofthe fluid column. There is a non-zero venical
velocity at the rop ofthe column ifthe mass convergenc into rhe column is

where the integral is over rhe cross-sctional area of the fluid column (tooking down ftom above). This is balanced by the local increase in hight of th water colunn, given by

o-=

ft1,a,: frl^,n* =J^,u**

(3.r2)

Because p is constant, the balaffe btlveen (3.11) and (3.t2) teads to

j^[ff.,
and because the area is

ou]ao=o,

(3.13)

a$itrary the integrand itselJ must vanish, whence,

a!t,.o6=o
or equivalently

(3.14)

Dh

|);+hv.

=0

(3.1s)

This dedvation holds whether or not the lower surface is flat. If it is, then h = 4, and if not h=4 4r. Equations (3.8) and (3.14) or (3.I 5) form a comptete set, surrunadzed in rhe
shaded box on the next page.

From the 3D mass conservation equation

Since the fluid is incompressible, the three-dimensionat mass continuity equation is V . = 0. Writing this our in component form

just

Au
az

\ax

ay

:-V

u.

(3.16)

-3.1 Dynamics of
a Single, Shallow Layer

127

The Shallow Water Equations

For a single'layff fluid, and ircluding the Coiolis terE, the inviscid shallow water

Dn6t+J/ =-gvn. Dh +lrv u-0 or mass continuit'4 Dt

momentum:

(sw1)

#+v.(hr)=0,

(sw.2)

where

is the ho zontal velocity, h is the total fluid thicloess, 4 is the height of the upper free surface and 4, is the height of the lower surface (the bottom topogaphy). Thus, h(t, y,t) = n\x,y,t) 4, (x, v ). The material derivative is

Daaaa J;=t-u o- at*uar*'av'

(sw.3)

with the rightmost explesion holding in Cartesian coordinat$.

Inte$ate this from the bottom of the flural (z = r,r) to the top (z = 4), noting that the
dght-hand side is indepndent of z, to grve

tuh)

-u(4b): -hv 'u.

(3.17)

Ar the top the vrtical velocity is the material derivative of the positron of a partiorlar fluid elment. But the position of the fluid at the top is just r], and thrcfore (see fig. 3 2)

wr =U. Dt'
At the bottom of the flurd we have similarly

(3.r8a)

u\nbt. =
where, apart from

Dnb

it

(3.18b)
'

ea hquakes and the like, Anrlaf = 0. Using (3.18a'b)'

(3.17) becomes

!{o
or. as in (3.15),

,o)

*no u=o

(3.19)

Dh Dt

+hV.n=0.

(3.20)

3,1.3 A rigid lid

lte

case where the uppr surface is held flat by the imposition of a rigid lid is sometimes of intIest. The ocean suggsts on such example, since the bath)T netry at the bottom of the

ocean prc!'rdes much Iarger vadations in fluid thclmess than do the small vaiiations in ttre

124

Chaptr

3.

Shallow Water Systems and lsentropic Coordinates

height of the ocean sufac. Suppose then that the upper surface is at a constant height
then, from (3.14) with

II

ahlaf = 0, th mass

conservadon equation becomes

(3.21)

vh.(lthbi =o,
where

h,

=H

nr. Note that (3.2r) alows us to define an incompressible mass'tlansport

Although the upper surface is flat, the pressur thre is no longer constant because a force must b provided by the rigid lid to keep the surface flat. The horizontal momentum equation is

Daa l= p"Pna' ot

where pxd is the pressure at the lid, and th complet quations of motion are then (3.2I) and (3.22).1 If the lower surface is flat, the trvo-dimensional flow itself is divrgence-free, and the equations reduce to the tlvo-djmensional incompressible Euler equations.

3.1a4 Stretching and the vertical velocity Berause thc hodzontal velocity is depth independent, the vertical velocity plays no role in advectioL Howevr, 1r/ is certaidy not zero for then the ftee surface would be unable to move up or don'n, but because of the vertical independence of th hodzontal flow 1, does have a simple vertical structurei to determine this we wrte the mass conseNation equatior

au
and integrat upwards ftom

. -

.v.
nb).

(3.23)

tle bottom to give

= wb

- (v u)(z

13.24)

Thus, th vetical velocity is a linear function of height. Equation (3.24) can be written as

Dz Dnh

;;-#

._ ,o.u\tz

4b\.

(3.2s)

and at the upper surfac 1r/ = D4/Dt so that here we have

Dn _., Drl tv . u\tn_ DI = =:J:2 Dt

nb).

i3.26)

Eliminatirg the divergence telm ftom the last tlvo equations gives

Dtl,
which in turn gives

not=

2-n\D

n_;wt4-nrt,

(3.27)

Dl' nb\ D tz-nb\ ^ ,' /-* ot\u*1-Dt\

(3.28)

This means that the ratio of the height of a fluid parcl above the floor to the total depth of the column is fixd; that is, the fluid sttches udfo r y in a colunr\ and this is a kinematic prope(y of the shallow lrater system.

1.2

Reduced Cravity Equations

t29

','I

t_

Fig.3.3 The reduced gravity shallowwa

ter sYstem, An active layer lies over a

deep, more dnse, quiescent laver. ln a ommon variation the upper surfac k held flat bya risid lid, and 40 = 0.

3.1.5 Analogy with compressible flow

Th shallow water equations (3.8) and (3 14) are analogous to the compressible gas dt'namic quations in two dimensions, namelv

P!
and

= Dtp

!o"

(3.29)

a{+v.tupt=o,

{3.30)

(.p) The mass conseffation along with an equation of state wtlch we tak to be p = f : p quaiions (3.I4) and (3.30) are identical, with the replacement - h lf p Cpv' then {3 29)

D!

-L!!sp:

6ypt zsp.

(3.3r)

If y = 2 then the momentum equations (3 8) and (3.31i become equivalent' with p - h and (in Cy - g. In an ideal gas y = cpl.! and vatues tlpicallv are in fact less than 2 air y = 7/5); of y' for all values exact for h;wev;r, if the quations are linearized, then the analogv is I shallo$ and lhe linearized dp dp lhen (1.31) bFcomes at ar - po-r.aVp' wherF I ttgH'vh po H and .a so that - 9H Ihc wdrFr momentum cquar ion rs a lr /Al - -H
sound waves of a compressible fluial are then analogous to shallo$'water waves' which are considered in section 3.7

3.2 REDUCED CRAVITY EQUATIONS

Consider now a single shallow moMng laver of fluid on top of a deep, quiescent fluid layel (Fig. 3.3), and beneath a fluid of negligible inertia- This configuration is often used as a mo(lel of the upper ocean: the upper laver rcpreserts flow in perhaps the upper few hmdred mehes of the ocean, the lower layer bing the near_stagnant abyss lf we turn the model upside alown we have a perhaps slightly less iealistic modl of the atmosphere: the lower Iayer represents motion in the trcposphere above lvhich lies an inactive stratosphere The equations of motion are virtually the same in both cascs.

130

Chapter

3.

Shallow Water Systems and lsentropi( Coordinates

3.2.1 Pressure gradient in the active layer

We

will derive th equations for the oceanic case (active layer on top) in tlvo cases, whirh differ slightly in the assumption mad about th upper surfac.
I Free upper surface

The pressur in the uppr layer is given by inlegrating the hyahostatic equation down ftom I he upper suface. Thus, at a heighr z in rh upper ld)cr

pte)

Spt(no 2),

(3.32)

whre 40 is the height of the upper surfac. Hnc, evrywhere in the upper layer,

*',,
and the momentum equation is

-"'no,
gvno.

(3.33)

Dt+f/u=
p2e) =
But

Du

(3.34)

In the lower Iayer the pressue is also givn by the weight of the fluid above it. Thus, at som level z in the lower layer,

pts\no

nt) + pzs(h - z).

(3.3s)

if this layer is motionless the hodzontal pressuie gradient in it is zero and therelore

ptsno- ps
whete

4r

-constant.

r't.?6r

g'

g@2

pt) I pt rs r}]'e reduced graviq,.'l}le momentum equation becoms

?1 Dt"

r.. , - s'o,,,.

(3.37J

The equations are complted by the usual mass consewation equation,

ff+nv.z:0,
where ,1

(3.38)

no - 4r. Because 9 > g', (3.36) shows that surface displacements are much smdiier than th displacements at the interior interface. We see this in the real ocean wher the mean intedor isopycnal displacments may be seveml tens of metres but vadations in the mean height of ocean suface are of the order of cenrimetres.

II The rigid lid approximation The smallness of the upper surface displacement suggests that we will mak little error is we impose a rltid iid at the top of the fluid. Displacments ar no longer allowed, but the lid will in gnml impart a pressure force to the fluid. Suppose that this is P(r, /, f), thn the hodzontal pressure gradient in th upper layer is simply

vpr = VP.

(3.39)

3.3 Multi

Layer Shallow Water Equations

r3L

The pressue in rhe lower layer is again given by hydrostasy, and is

p2= -hSn1+ pzg\nr - zJ +P = pgh p28(h+z)+P,

so that

(3.40)

9pz =
Then

-Slpz

pr)Vh + VP.

(3.4r)

if Vp2 =

0 we have

Sl?z-Pl)vh=9P'
and the momenrum equation for ihe upperlayer rsjusl

(312)

D]+y"u=-sivn.
-

(3.43)

whcre g' = g\p2 pt) I p I . Thes quations differ from the usual shallow water equations only in the us of a rducd gravityr'in plac of itslf. lt is the densiay dr'frernce betrveen the two layers that is important. Similarly, if we tale a shallow water system, with the moving layer on the bottom, and we suppose that overlying it is a stationary fluid of finit density, then we would easily furd that the fluid equations for lhc moving layer are the same as if the fluid on iop had zero ineltia, except that would be replaced by an appmpriat reduced gmvity (problem 3.1).

3.3 MULTI.LAYER SHALLOW WATER EQUATIONS we now consider the d).namics of multiple layers of fluid stacked on top of each other. This is a crude representation of continuous statification, but it tums out to b a powerful
model of many geophysically interesting phenomena as well as being physically realizable in the laboratory, The prcssurc is continuous across the interface, but the density jumps discontiruously and this allows the horizontal velocity to have a corresponding discontinuity.
Thc set up is illustrated in Fig. 3.4. In cach layI pressure is given by the hyabostatic approximatioq and so an)ryhere in the interlorwe can ffnd the pressure by integrating dolvn from the top. Thus, at a height z in t}le first layer we have

pt=plg(no-zJ,
and in the second layer, p2 = whete

13.44)

pfiho -

4t) + p?gh)

- z)

= plsno +

hs\nr

pzgz,

(3.4s)

because orily the horizontal

pt) lpr, and so on. The term involving z is irrelevant for the dynamics, de vative cnters the equation of motion. Omitting this tetur, for the nth layer the dFamlcal prcssure is given by the sum from the top do$'n:

gi

= g@2

P^=Pr ZStth'
where 9;

(3.46)

=9(pi+r p)lh(bDrgo

= g).The interface displacements may b expressedin

terms of the layer thiclaresses by sumrling from the bottom up:

4^=4y+ \

ht

13.47)

132

Chapter

3.

ShallowWater Systems and lsntropic Coordinats

no

4r'
Fig. 3.4 The multi-layer shallow water system. The layers

n2

are numberd from the top down. The coordinats of th interfa.es are denoted by 4, and the layer thicknesses by h, so that hi = 4r - 4i r.

nF1,

4i

The momentum equation lor each layer may then be wdtten, in general,

Du" i+J\uh=

1vpn,

(3.48)

where the pressure is given by (3.46) and in terms of the layer depths usinC (3.a8). If we male the Boussinesq approximation then pn on the right-hand side of (3.48)is replaced by
Finally, the mass conservadon equation for each layer has the same form as the singlelayer case, and is

D!*n"v,u.=o.

(3.49)

The two- and threeJayer cases

The two.layer model (Fig. 3,5) is the simplest model to captwe the effects of stratification. Evaluating the presswes ustng (3.46) and (3.47) we ffnd:

Pt = Prgno = PP(111+ h2 + nb) p2 = p tts no + slntl = pt lg lh t + h2 + nb) +

The momentum equations for the Evo layers arc then

(3,s0a)

gihz + nb\1.

(3.50b)

Dr.'

fi'

zut =.

gV no

.gvthr -

h2 + nbt.

(3.51a)

and in the bottom layer

Dff r 1*u,

= fibvno + e\vn,)
:
- Pj[ov tnt
+

(3,51b)

n,

n) + siv

(h2

+ n,,l .

3.3 Mukl-Layer Shallow Water Equations

Fig.3.5 The two.layer shallowwater system. A fluid ofdensity p, lies over a denser pr. ln the.educed qravity case the lower layer may be arbitrarily thick and i5 ass!med stataonary and so has no horizontal pressLrre gradient. ln the 'rigid lid' approximation the top surface displa<enent is negleded, but there is then a non-zero pressure sradient induced by the lid.

fluid of density

In the Boussinesq approximation prlp, is replaced by unity. In a thrce-layer model the dynamical prcssures are found to be

where h = no added in a systematic fashion.

(3.52b) gi(hz + b + rtb)l (3.52c) pr = pt Ish + gilhz + h3 + nb) + sih3 + nb,l, = r,D + hr +hz+h a\d.gi= g\p3 pz)/pt.Morelayrscanobviouslybe
+

PPh p2 = pllgh
P1 =

(3.s2a)

3.3.1 Reduced'gravity multi-layerquation

As with a single active layer, we may envision multipl laycrs of fluid overlying a dceper stationary layer- This is a useful model of thc stratified upper ocean ove ying a nearly stationary and nearly unshatified abyss. hdeed se use such a modei to study the 'vcntilated thermocline' in chapter 16 and a detailed treatment may be found there. If we supposc there is a lid at the top, then thc rnodel is almost the same as that of the previous scction. However, now the horizontal pressure gradient in the lowest model layer is zero, and so we may obtain th pressures m all the active layers by integrating the hydrostatic equation upwards from this layer. Suppose we have N moving layers, lhen th reader may verify that th d),nami pressure in th rL th layer is given by

P^

Ptg,nt,

(3.s3)

134

Chapter

3.

Shallowwater Systems and lsentropic Coordinates

Fiuid velocity, into page

Fis. 3.6 Geostrophic flow in a shallow water system, with a positive value of the Coriolis parametr /, as in the Northern Hemisphere. The prssure force is directed down the gradient ofthe height field, and this can be balanced by the Coiolis force ifthe fluid velocity is at right angles to it. lf/ were nesative, the geostrophic flow would be rcversed.

whereasbeforeg;

inte

=g(p,+r pj)/pr.Ifwehavealidatthetop,andtaketo=0,thenthe
by
(3.s4)

ace displacements are related to th Iayer thicknesses

\"

= >ht.

Irom thse er?ressions the momentum equation in each layer is easily constructed,

3.4 CEOSTROPHIC BALANCE AND THERMALWIND

Geostrophic balance occus in the shallow water equations, iust as in the continuously stratified equations, whn the Rossby number U/fl is small and th Coriolis term domhates the advective tIms in the momentum equation, In the singte layer shallow water equations the geostrophic flow is:

fxus=-gVn.
be geostrophic.)

(3.ss)

Thus, the geostrophic velocity is proportional to the alope of the surface, as skerched in Iig. 3.6. (For the rest of this sction, we will drop the subsc ptr, and rake.all velocities to

h both the singlelayu and multi-layI cases, th slope of an intedacial suface is dtecrly related to the aliffernce in pressure gradient on eithei side and so, by geostrophic balance, to the shear of the flow. This is the shalow water analoglle of the thermal wtud relation. To obtain an exprssion for this, consider the inteface, ,, betwen two laye$ Iabelled I arld 2, The prcssure in two layers is given by the hydrostatic relation and so,

h=A(x,y)-p\gz p2: A(x,y) p$4 + pzsln - zJ

=

(at some z in layer

l)

(3.56a)

Alx,y)

prg'rn

p2gz

(at som 2 in layI 2)

(3.s6b)

3.5 Form

Drag

Fig.3.7 Margules' relation: using hy'

drostasy, the difference in the horlzon talpressure gradient btween the upper and the lower layer ie given by g'prr, where r = Iand = AzlA! is the in terface slope and g' = \p, pt)lpt. ceostrophic balance then sives /(u, ,r) = 9'J, which is a special case of (3.60).

i\p :

pzg

Lz

where A(x,

y)

is a function ofintesration. Thus we find

*'o,
Using (3.57) thcn gives

,,t = n'''n'

(3.s7)
each layer, the velocity obeys

If the flow is gcostrophically balanced and Boussinesq then, ln

ft, = lxxvp,.
/(r.rr
or in general
r.rr r

(3.s8)

= k\siv4,

(3.s9) (3.60)

f
rclation' (Fig.
3.7).'?

(u" u"-) = kxs;v4.

Tbis is the thermal wind equation for the shallow water system. It applies at any interface, and it implies the slear rs proportional to the interface slope, a result known as the 'Margules
Suppose that we reprsent lhe atmosphere by two layers of fluid; a me dionally decreasitrg temperature may ther bc represented by an intedace that slopcs upwards toward the pole- Then, in cither hmisphere, !,"e have

' ,,
].5
FORM DRAG

s.'

Y o,

r.br,

and th tmperature gradient js associated with a positive shear (se problem 3.2).

\lhen the hterface between two layers varies with position that is, when it is $,avy the layers exert a prcssurc forcc on each other. Similarly, if the bottom of thc fluid is not flat then the topography and the bottom layer will in general exert forces on each other. This kind of force is known as /onn drdt and it is an important mcans whcreby momentum can be added to or cxtracted from a flow3 Consider a layer confind between two intedaces, 4r (x,l) and 4r(,.,j',). Then over some zonal interval I the average zonal pressure forc on that fluid layer is giten by

ir: r,

an

u, o'.

(3.62)

r36

Chaptr

3.

Shallow Water Systems and lsentropic Coordinates

Integrating by parts ffst in z and then in x, and noting that by hydrostasy ap lAz does nor depnd on horizontal position within the layer, we obtain

'r=

if,l*4'":.

dx

4fl. + an. +p an. = (ila; n,i = j;

p,

An'

r;,

(3.63)

where pr is the pressue at 4r, and similarly for pr, and to obtain the second line ive suppos that the integral is around a closed path, such as a cfcle of latitude, and the average is denoted with an overbar. These tems represent the tmnsfer of momentum from one layer to the next, and at a particular intedace, ;, we may define the folm drag, Tr, by

T,=lt,-=-n,-

an.

ap,

(3.64)

The form dmg is a strss and as the layer depth shrinks to zro its vertical dfvative, aTlaz, is the force (per unit volume) on the fluid, FoIm dlag is a particularly important means for

the vertical transfer of momntum and its ultimate removal in an ddying fluid, and is one of the main mechanisms whereby the wind sffess at the top of the ocean is cornmunicated to th ocean bottom. At the fluid bottom the form drag is wher 4, is the bottom topography, and this is propoflional to the momentum exchange with the solid Ea h. This is a significant mechanism for th ultimat removal of momentum in th ocean, especially in the Antarctic Circumpolar Culrent where it is likely to b much larger than bonom (or Ekman) drag arising from small-scale turbulence and friction. In the two-layer, flat-bottomed

Ftt

case the only folm drag occurrjng is that at the intface, and the momentum transfer between rhe larers is just p.ArnAx or -nta pjdi: then. the force on each Ia)r due ro lhe other is equal and opposite, as we would e).pect from momntum consenation, (Form drag is discussed more in an oceanographic context in sections 14.6.3 and 16.6.2.)

For flows in geostrophic balance, the folm drag is related to the meddional heat flux. The pressure gradient and velocity are related by p/r' : ap'lax and the interfacial displacement is proportional to the temperatue perturbation, ,' lin fact one may show that n . -b I'ablAzl.Thus -anp'rtax * u,. a.orresponden.e t}lar will rc ocrur$henhe

consido the -Eliaisen PaIn fux in chapter

7.

3.6 CONSERVATION PROPERTIES OF SHALLOW WATER SYSTEMS

There are l!\o cornmon M)es of conservalion properly m tluds: {i, material in\arjanrs and iii) integal invariants. Matedal invariance occurs whn a property (4 say) is conserved on 0. An integral invariant is on that is each fluid dmmt, and so obys th equation D+lDf conseNed followhg an integration over some, usually closed, volume; energy is an example.

3-6.1 Potential vo(icity: a material invariant

The vorticity of a fluid (considered at greater length in chapler 4), denoted ur, is dfind to be the curl of the velocity fidd. Let us also define the shallow water vorticity, ur* , as the curl

of th hodzontal velocity. We therefore havel

sr=Vxl,,

(tr*=Vx!r.

(3.6s)

r'(o);i =;F(a)
since F is arbitrary there are an different choices of F.

=0.

13.74)

infilte

number of materlal invariants corrspondrng to

,t"J

3.6 Conservatron Propertre5 of Shallow Water System5

Recatse

117

not

aulaz =

Ax

lAz = 0, only the vrtical component of @* is non-zero and

63)

-.-kr4 \dx
Using the vector identity

9'\=k(. dy )
. (Vxu\.

(3.66)

:is
we 64)

{x.V)l,l- ;Vrr'lr)w
te the momentum equation, (3.8), as

(3.67)

a#*-.".= v(on,l",).
Vx(@+xu) = (r. V)&r+ (uJ+. V) = (u. V)oJ+ + oJ*V .u,
+

(3.68)

az,

for
red

To obtain an evolution equation for the vorticity we take the curl of (3.68), and ma} use of ihe vector identity

@*V.x-xV

. @* (3.69)

his

ib
(or
Led

using the fact that V . @* is the divergence of a curl and thrcfore zro, and lu* . V)lt = 0 because @* is perpmdicular to the surface ir which & varies. Talqng the curl of (3.68) gives

ler
he ag
where <

dt -

a( r {r.. V)< - -(V (Dh

(3.70)

k . ur*. Now,

the mass conser,'ation equation may be written as

id
lat
and using this (3.70) becomes

Lv u=;w,

(3.71)

D(
Dt

<Dh - nDt'

13.72)

which simpMies to
0d

9/5\=o Dt \h/

(3.73)

is

ile

The important quantity </h, often denoted by Q, is kno$'n as the potential vorticity, aJ.d (3.73) is the potential vorticity equation. W r-derive this conservation law in a more gneral nlay in section 4.6 Because Q is conserved on parcels, then so is any function of Q; that is, F(Q) is a material invariant, where F is any function. To see this algebraically, multiply (3.73) by f'(Qr, rhe derivari\e ol I wirh respecl ro Q, Fving

l!

r38
Effects of rotation

Chapter

3.

ShallowWater Systems and lsentropic Coordinates

In

rotatjng frame of rcference, the shallow water momentum quation is

Dur't+Itu=
where (as before)

gv4,

'

(3.7s)

/k

This may be

tten in vector invaliant form as

a ;;
"ai

+((lJ-

rJ'.u-

I ,\ _/ v\sn! 2u ). tI+ttv / u.

(3.76)

and taldng the curl of this gives the vorticity equation

atu.vtte -It=

tJ-77')

This is the same as the shallow water vorticity equation in a non_lotating fram, save that 6 is replaced by 6 + /, the reason lor this being that is the vorticitv that the fluid has by vhtue of the background rotation. Thus, (3:77) is simply tle equation of motion for the total or absolute vorticity, &r" = a* + .f = le + f)k. The potential vo icity equation in the rotating case follows, much as in th non-rotating case, by combining (3.77) with the mass conservation equatron, grving

-q(#)=.
That is, Q = (< + invariant.
Vo ni

(3.78)

/)/h,

the potential vorticity in a rotating shallow system' is a material

city

a n

d ci rcul ation

Although vorticity itslf is not a material invariant, its integal over a horizontal matedal area is invariant, To demonstate this in the non-rotating case, consider the integral

,=!^<ao=J^onae,
over a surface A, the cross_sectional area of a cohmn of height h (as in matedal derivative of this glves

(3.79)

lig

3:2) Taking the

ff:J ffrar*J o-loaer.

99=9 [ rao = o. Df DtJrtkorem, it
may b \TitteD as

(3.80)

The fust teIm is zero, by (3.72); the second telm is just the derivative of the volune of a column of fluid and it too is zero, by mass consewation. Thus,
{3.81)

Thus, the integral of the voticity over a some crcss_sectional area of the fluid is unchanging' although both the vorticity and area of the fluid mav individualy change. Using Stokes'

DC

J,, DI - DIJ

D-

. ar.

{3.82)

3.6 Conservation Properties of Shallow Watr Systems

r39

where th line integral is around the boundary of A. This is an example of Kelvin's circulation theorem, which we shall met again in a more general form in chapter 4, where we also

considu the rotating

case.
J

A slight generalization of (3.8r) is possibl. Consider the integral J = again F is any ditrerentiable function of its argumen(- It is clear thal

F(Q)h

d?4

where

-qJ.tor,'oo=0.

(3.83)

IJ the area of integration in (3.68) or (3.83) is the whole domain (enclosed by frictlonless walls, for example) then it is clear that the intcgral of ,f(Q) is a constant, including as a specjal casc rhc intcgral ol <.

3.6.2 Energy conservation: an lnteqral invarlant

Sincc we have made various simplifications in deriving the shallow watr system, ir is not self-evident that energy should be conse cd, or indeed what folm the energy rakcs. The kinelic nergy density iKE), that is the kinetic ncrgy per unit area, is porrt?/2. The potntial

energy dcnsity of th fluid is

ez=

['

ooo,a, = loosn'.

(3.84)

The factor p0 appears in both kinetic and potential energis and, because it is a constant, we will omit it. For algebraic simplicity we also assume the bottom is flat, at z = 0. Using the mass conservation equation {3.r5) we obtain an equation for the evolution of Potential energy dcnsity, namely

ft{*on'v.u=o

(3.85a)

*'4., ("+)t+, "=,

a-z
a
Addins (3.8sb) and (3.86b) we obtajn

(3.8sb)

From the momentum and mass continuity equations we obtain an equation for the evolution of kinedc energy density, namely

hi

u2h

ht

(3.86a)

hr2 I h,\ su v2 h, dr z-tv \u 2 )

\)

(3.86b)

Sl

@"'

en') * v

ll"@n'*n'*sn')l :0,

(3.87)

ff*v.r=0,

(3.88)

140

Chapter

3.

Shallow Water Systems and lsentropic Coordinates

wherct=KE+?E= (htt2

+ ghz)l2lsrhe density of the total eneigy ?nd F = ttlhuz + gh2 + gh2) 12is r}le energy flrlx. If the fluid is confined to a domain boundd by rigid wals, on which the nolmal component of velocity vanishes, then on integating (3.87) over that aiea and using Gauss's theorem, the total energy is sen to be conserved; that is

ai ra l thuz+ah2\da:o. dt 2dt Ja
found in the

(3.89)

Such an energy principle also holds in the case with bottom topogaphy. Note that, as w case for a compressible fluid in chapter 2, the enrgv llu,.( in (3.88) is not just

the enrgy density multiplied by the velocity; it contains an additional tetlr:. guh2 12, and this represents the enryy transfer occurdng when the fllrid dos work against the pressue force (see problem 3.3).

3.7 SHALLOW WATER WAVES

Let us now Iook at the gravity waves that occur in shallow water. To isolate the essenc of the phenomena, we \4'ill consider waves in a single fluid layer, with a flat bottom and a ftee uppI surface, in which gravity provides the sole restorurg force.

3.7.1 Non{otating shallow water waves

civen a flat bottom the fluid thickness is equal to the free surface displacement (lig. 3.1), and we let

h(x, Y,t) : H + h' lx,Y,t) = H + utx, y,l t - u tx.y.It.

The mass

4'

\x'Y't)'

(3.90a) (3.90b)

conse ation equation, (3.15), then becomes 4.1' at

-," , n''o. -u .v4 =0,

d,t. ,

(3.91)

and nglecting squares of small quantities this ).rlds the linar equation

at'"",,

r.

o.

. /
= 0, yields

(3.S2)

Similarly, linearizing the momentum equation, (3.8) with

i
gence of (3.93) Ieads to

Au'

= -sv tt

(3.93)

Eliminating velocity by differentiating (3.92) with respect to time and taking the diver'

a#
4/ =

oav'n'

= o,

(3.94)

which may be recogEized as a wave equation. w can find the dispersion relatronship for this by substituting the trial solution
Re

ier(ft

'-dt,

(3.9s)

3.7 Shallow Water

Waves

t4t

wher is a complex constant, = ik +jl is the horizontal wavenumber and Re indicates that the real pa of the solution should be tal(en. Il for simplicity, we restrict attennon to the one dimensional problem, with no va ation in the jv-direction, then substituting into (3.94) leads to the dispersion relationship

(3.96)

where c = .]3F; that is, th wave sped is proportional to the square root of the mean fluid depth and is independent of the wavenumber the waves arc dispersior ess. The gcneral solution is a superposition of all such waves, wlth the amplitudes of each wave (or lou er componen0 being derermined by the Fouier decomposition of the lnitial conditions. Because the waves are dispersionless, th general solution canbe written as

n'\x,t) = ;EG cl)

where

F(x + ct)l

(3.97)

is the height field at f = 0. lrom this, it is easy to se that th shape of an initial ahsturbance is preserved as it propagates both to the dght and to the left at speed c, (se
also problem 3.7).

F(r)

3.7.2 Rotating shallow watr (Poincar6) wavs

We now consider the effects of

rotation on shallow water waves, Linearizing the rotating, flat-bottomed /-plane shallow water equations li.e., (Sw.1) and (Sw.2) on page 1271 about a stale of rst we obtain

u:,:

- ,"n -

nu*

u{

no

-nuuo".

+ , ,(#
=

dy

I(3.98a,b,c)
(3.99)

We non-dimensionalize thes equations by

writing

(x,y)

L(i,i), tu',u')=u6,i),
^, dn

Hi,

and (3.98)becomes

dlt

ar --

ai

dx

.ah laa at\ = 0. (3.100a,b,c) ^=-c'-, ^-ai " -:+l-+-l dy at at \ax oJl -+ln

where a = n?E/ U is the non-dimensional speed of non-rotating shallow watcr waves. (It is also the inverse ot rhe Froude number U /\,tF., To obldn a dispersion relationshjp \^e lel

(i,i,i)
wfrere 0 =

= (a,t,D'1[

6i),

(3.r0r)

[i

ana A ls the non.alimensional frequency, and substilute into (3.100), giving

/ ia -r. r,z[\ rfir lrn -ii i.'rllal=0.

\ ik | -i6) \nt

(3.102)

142

Chaprer

3.

ShallowWater Systems and lsentropic Cootdinates

Fig.3.a Dispersion relation for Poincar6 waves (solid) and non-rotating shallo\{ water waves (dashed). Frquency is scaled by the Coriolis fre quencv f, and wavenumber
by the inve6e deformation ra-

^3

dius JsHlf. For smallwav numbe6 the frequency is ap' proximatly /ifor hiqh wavenumbers it asymptotes to that of non-rotating waves.

-t

wavenumber (k x

ld)

This homogeneous equation has non-tdvial solutions orily if the dterminant of the matrix vanishs. This condition givs

6@2

j3

t'R'

= o.

(3.103)

wtrereR,=l'?+i'zrtrereare$4'oclassesofsolutionto(3.103).Thetustissimply6:0,ie. time inalepndnt flow coresponding to geostrophic balance in (3 98) (Because geostrophic
balance gives a divergence-free velocity field for a constant Coriolis pammeter the equations

are satisfied by a time independent solution.) The second set of solulions satisfies the disDrsion relation rr.r o4i I il -

A/ ji

i,'k

t,

whi.h in dimensional folm is

@2=f&+gH(k?+Lz)

(3.10s)

Th corresponding waves are knolvn as Poinca,'d waves,a and the dispersion relationship is illustrated in lig. 3.8. Note that the frequncy is always greater than the Co olis ftequency /0. There are two interesting limits.

(i)

The short wave

linit.If

*'-#'
:
0 and

(3.106)

where = k2 + r'?, then the dispersion relationship reduces to that of the non_rotating (3.96). This conairtion is equivalent to requirjng that the wavelength be much case

f,

shofier than the deform ation radius, Ld = 17h.lf. Specifica]ly, if I

is the wavelength, the condition is
A2 <<

l = 2nlt
(3.107)

Lien)z

3.7 ShallowWater

Waves

143

The nume cal factor of (2rr)'? is more than an order of magnitude, so care must be takn when deciding if the condition is satisfied in particular cases. Futhermore, the havelength musr still be longer thar the depth of the fluid, othenvise the shallow watcr condition is not mct.

(ii)

The lonq wave

linit.

r,

K/
that js

.,
'9

.
I/,

rr.ro8r
then th (3.r09)

if the wavelength is much longer than the delomation radius dispersion rlationship is
'I

hcsl dr( I nuM dr inpl' ial o\.illation\. Ihc cquations ol morion gi\ ing risF Io rhem are

!-t*':0, ff * fou' :0,

which are quivalnt to material equations for free particles in straincd by prcssurc forres, namely
a

(3.1i0)
rotating fram, uncon'

d'7x
See also

ar, t'u =ir, ?;i+J'u=o.

d2v

(l.r r r)

problem 3.9.

3.7.3 Kelvin waves

The Kelvin wav is a particular F/pe of gravity wave that exists in the presence of both rotation and a laleral boundary. Suppose thcrc is a solidboundary at 0; clearly harmonic solutions in the j/-directlon ar not allowable. as these would not satisfy the condition of no normal flow at the boundary. Do any wave like solutions exist? The afffmative answer to this question was provided by Kelvin and thc associated waves arc now eponlnously knor\.n as (elvin wdves.s We begin with the linearized shallow water quations, namly

/ :

au

ol

Jou

s
0 at

dn
y

ax' at
:0

da to' "

dn

ar'

ao H ../"u

at

\a,
r' :

au\
ay

The fact that

r' :

)(3.112a,b,c)
(3.1r3a,b,c)

o'

suggcsts that r{c look for a solution with

0 cvcrtrvhcre,

r\'hence thse equations becom

E=-sa,'

Au

An

r"" = qa!. 4 ol 0L *oY ax

A2u' (, A2u'

=o.

Equations (3.r13a) and (3.113c)lead to the standard wave cquation

at

ax/

(3.rr4)

$here c

/gH,

th usual wave speed of shallow water waves. Thc solution of (3.114) is

u'

Ft\x

+ cL,

yl + Fz\x -

ct,

yl,

(3.115)

144
r4aith

Chapter

3.

Shallow Watr Systems and lsntropic Coordinates

conespondlng surface alisplacement

n' -,18 I s

[-Flx

+ ct, y) + Fzlx - ct, yr].

(3.116)

The solution represents the superposltlon of two waves, one (Fl) travelling in the negative r-clhection, and the other in the positive r-direction. To obtaln the dependerce of these functions we use (3.113b) which givs

1FL .k _ ay .Jgt7
with solutions

iFz ay
F2

fo \1911

(3.117)

h
where

F(x + ct\ertLd

G(x

cue-rtLn,

(3.118)

ld

vE;I/JFo is the radius oI deformation, The solution Fr grows exponentially away

from the wall, and so fails to satisfy the condition of boundedness at intrity. lt thus must be rliminated, leaving tbe geneml solution

u'=etlLdG@-ct)t
n' -- ,,lEDe vll'G(x

x,' =

O,

ct)

(3.1r9)

Th$e are Kelirln waves, and they decay e8)onendally away from the boundary. If f0 is
positive, as in the Northern Hemisphere, th boundary is to the right of an observet movturg wlth the wave. Given a constant Coriolis palameter, we could equaly well have obtained a

solution on a meddional wall, itr which case we would find that the wave again moves such that the wall ls to the right of the wave dfection. (This is obltous once it is reatized that /-plane d),namics are isotropic in x and Jr'.) Thus, in the Northem Hernisphere the wave moves anticlockwise rcund a basin, and conve$ely in the Southem Henisphere, ald in both
hemispheres the dtection is cyclonic.

3.8 GEOSTROPHTC ADJUSTMENT

in chapter 2 that the large-scale, extratropical circulation of the a@osphere is in near-geostrophic balance. Why is this? Why should the Rossby number be small? tuguably, the magnitude of the velocity in the aunosphere and ocean is rtrlhately given by
We noted the sEengrh of the forcinS, and so ultimately by the diferentlal heating between pole and equator (although even thls argument b not satisfactory, since the forcing mairily detemines the energy throughput, not drectly the energy ltse]J, and the forcing ts ltselJ dpendent on the atmospherc's response). But even supposing that the velocity magnltudes are glven, there is no a pdori guaranrce thal the forcing or the dynamics will produce length scales that arc such that the Rossby number is small. However, there is in fact a powerful and ubiquitous process whereby a fluid in an initially unbalanced state ftturally evolves toward a state of geostrophic balance, namely ,eostlophic a4irstmen! This process occurs quite genenlly in rotating fluids, whether stmtiffed or not. To pose the problem in a simple folm we will consider the free evolution of a single shallow layer of fluid whose initial state is madfesdy unbalanced, and we will suppose that suiface displacemelts ale small so that ttre

3.8 Ceostrophic Adjustment

145

cvolution ofthc system is described by the linearized shallo$' equations ofmotion. These

i+J/u=

au

g\/tl,

a]+av u:0,

(3.r20a,b)

whre t is the free surface displacement and ,q is the mean fluid dpth, and we omit the primes on the linarized variables. 3.8.1 Non rotating flow
We considerfust the non-rotating problem set,

Mthlitleloss of genemtity, in onc dimension.

we suppose that inilially the fluid is at rest but with a simple discontinuity in the height

and u().,t = 0) = 0 everywher. We can realize these initial conditions physically by separating two fluid masses of differeni depths by a thin dividing wall, and then quickly removing the wall. What is the subsequent cvolution of the fluid? The general solution Io the linear problem is givenby (3.97) where the functional form ls determind by the initial conditions so that her (3.122) F(x) = 4(x,t = 0) = 40sgn(x). Equation (3.S7) stats that this initial pattern is propagated to thc right and to the left. lhdr i\.lho di!(onrinuiries in fluid heighl move Io lhc righl and lell al a speed. \&F. Specificaly, the solution is

4(x,t=0)=1 L 4o )'>u

l+n, x<o

(3.r2r)

4(x,t)

= ;4olscn(r+ct)+sgn(r

ct)1.

(3.123)

waves are dispersioilss, the solution is

Thc initial conditions may be much more complex than a simpl ftont, but, because the still simply a sum of the translation of those initial conditions to th right and to the lft at speed.. Th velocity field in this class ofproblm

which sivs, usins (3.97),

au. An i = s a"'
u

(3.r24)

+lF\x + ct) - F(x -

ct)1.

(3.r25)

Conslder the case with initial conditions given by (3.121). At a given location, away from the initial disturbance, the fluid remains at rest and undisturbed unll the ftont arrives. After the front has passed, the fluid surface is again undisturbed and the velocity is uniform and non zro. Spcifically:

- f-lo"gn,*, ,,_
lu

lo ':1''o''"

t"t lxl

(3.r26)

The solution wilh 'top-haf initial conditions in the hight lield, and zero initial velocity, is a superposition two discontinuities similar to (3.126) and is illustmted in Fig. 3.9. Two fronts propagate in either direction from each discontinuity and, in this case, the final velocity, as well as the fluid displacement, is zero aftr all the ftonts have passed. That is, the disturbance is radiated compltely away,

146

Chapter Pefturbation height

3.

ShallowWater Systems and lsentropic Coordinates

t=0

\Tt

t=2
0 0

Fig. 3,9 The time development of an initial 'top hat' height disturbance, with zro initial velocity, in non rotating flow. Fronts propagate in both directions, and the velocity is non-zero between fronts, but ultimatelythe velocityand height distubance are radiated away to infnity.

3.8.2 Rotating flow Rotation makes a profound difference to the adjustment problem of the shallow water system, because a steady, adjustcd, solution can exist with non-zero gladients in the height field the associated pressure gradients being balanced by the Coriolis force - and potential voticity conservation provides a powerful consEaint on the fluid volution,o Lr a rotating shalow fluid that cons ation is represnted by

af * u.oA:
where Q

o,

\3.127)

= (< + /)/h. ln the linear

case

with constant Coriolis parameter, (3.127) becomes

(3.128)

a#=", ,=(< n*).

This equation may be obtained either from the linearized velocity and mass conservation equations, (3.120), or ftom (3.127) directly. h the latter case, we write

a=T* "jre*r.r (r-fi)

having used
becomes

"

i(^.,

k+) = r;. +

(3.129)

nimnortant, as is the

/o

>

l<l

and

l,.V4, and this is second order in perturbd quantities and so is neglectd. Thus, making these approximations, (3.127) reduces to (3.128). The potential vorticity field is therefore fixed in space! Of cours, this was also tm in the non rotating case where the fluid is initially at rest- Thn a : < = 0 and th fluid rcmains inotational throughout the subsequent evolution of the flow. However, this is rather a weak constraint on the subsequent evolution of the fluidt it does nothing, for example, to prevent the conversion

H > lt . The term /0/H is a constant and so dtnamically Il t factor multiplying 4. Further, the advective telm r . VQ

3.8 Ceosrophic

Adjustment

147

of all the potcntial energy to kinetic energy. In the rotating case the potntial vorticity is non-zero, and potential vorticity conservation and geostrophic balance are all w need to infer the final steady state, assuming it exisls, without solving for the details of the flow
volution,
as we

now sce.

With an initial condition for the hight fild given by (3.121), the initial potential vorticity is givn by

-J

atx

. lt=l | - fonntu x <o fonulH x>0,

(3.r30)

and this remains unchangcd throughout th adjustment process. The final steady state is then the solution of the equations

kft - a,*,tt, t^" - -n31, h,

- sa*.

(3.131a,b,c)

wherc<=A1)lax-A lay.

Because the Coriolis parameter is constant, the velocity

ficld

is horizontally non-divergent and we may define a streamfunclion {, = (3.131) then reduce to

t4l/0.

Equations

(''-fi)*=',",,',
where rter

(3.r32)

ght
rnd

l,j = \'?E//0 is known as the Ross}/ rddru s of defonnation or often just the 'defor mation radius' or the 'Rossby radius'. It is a naturally occuning length scale in problms involving both rotation and gxavity, and arises in a slightly diffrcnt form in stratiffed lluids. The initial conditions (3.130) admit of a nice analytic solution, for the flow will rcmain uniform inJl/, and (3.132) reduces to
(3.133)

-.;)

scpantely for x > 0 and x < 0 and then match the solutions and their first derivatives at = 0, also imposing the condition that the streamfunction dccays to zerc as ,a t@. The solution is
We solvc this

28)

- ['Bqoliottl-e't"t
l+rsao71or,:

x>o
y <0.

-c'",

(3.r34)

The velocity field associated with ttns is obtained from (3.13lb,c), and is
29)

u=O,

"= ffi.-",,.
a

(3.rrs)
jt foms along the initial

1a
is

l'

Th vclocity is perycndicular to lh slope of the free suface, and discontinuity, as illustrated in Fig. 3-10.

he
he on

The important point of this problem is that th variations in the height and field are not mdiated away to infinity, as in the non-rotating problem. Rather, potential vorticiry consrvation constrains the inlluence of the adjustment to withln a defomratioD radius (we see now why this name is appropiatc) of thc initial disturbance. This property is a general
one in gostrophic adjustment

jump, as considered in problem 3.12.

it also arises if the inilial condition consists ofa velocity

148

Chapter

3.

Shallow Water Systems and lsentropic Coordinates

'6

-1

_!p

.!'
-E

-ou

.$r

5.r

!o

-2 x

lLa

Fig. 3.t0 Solutions of a ljnar geostrophic adjustment problem. Top panet: the initial height field, given by (3.12l) with 40 : 1 Second panel: equilibrium (finat) heishtfield, 4 given by(3.134) and q : fa| l g-Thnd panel: equilib um geostrophic velocity (normal to the gradint of height feld), given by (3.t3s). Bottom panet: potential vonicity, qiven by(3.130), and this does note evolve. The disrance, x is non dimensionalized by the defor.nation radius rd = a&F//o, and the velocitv by 40@ lfaLd). changes to the initial stat occur only within o(Id) of th initiat discontinuity; and as r - ao the initial state is unaltered.

3.8.3 * Energetics of adjustment How much of ttre initial potential energy of the flow is lost to infinity by gravity wave radiation, and how much is converted to kinetic enrgy? The linear equations (3.120) Iead to

|fttuu'
E=

* sa'1 *

sIrY

1ur1 =

x'

(3.136)

so that energy conseruation holds in the form

!J@u't sn\d,,

d-E

dt =0'

(3.r37)

Fovided Ore irtglal of the divergence telm vanishes, as it normally witl in a closed domain. The fluid has a non-zero potentiaf energy, ( I / 2 ) J: ,r42 dr, if there arc variations in fluid

1.8 Ceostrophic adjustment

height, and witll the initial conditions (3 121) the imtial potential energy is

t49

Ph
This is nominatly inffnite if th fluid

[l

(3.138)

'rao'

has no boundaries, and the imtial potential nergv

density is ,11612 everlHhere. In the non rotating case, anal with initial conditions (3 121)' after the front has passed' the potntial energy density is zero and the kinetic nergy density is I11r2 / 2 = g nt lZ' usi"g (3.126) and d'? = 9H. Thus, all the potendal energy is locally converted to kinetic energv as th front passes, and ventually the kinetic energy is distributed uniforrr[y along the line. In the case illustrated in lig. 3.9, the potential energy and kinetic energy are both ndiated away from the initiat disturbance. (Note that althowh we can superpose the solutions ftom diffrent initial conditions, we carmot superpose their potential and kintic energies.) The gmeral point is that the evolution of the disturbance is not conlined to its initial location. In contrast, in the rotating cas th conversion ftom potential to kinetic energy a larget confined to within a deformation radius of the initial disturbance, and allocations far from the inilial disturbanc th inilial state is ssentially unaltered. The conservation of potential rcrticity has prevented the complete conversion of potential enrgv to kinetic energv' a result that is not sensitive to the precise form of the initial conditions (see also problem 3.r0). In fact, in the rotatiru case, some of the initial potential energy is converted to kinetic nergy, some ftmains as potential energy and some is lost to infinity; let us calculate thse amounts. The final potential energy, after adjustment, is, using (3 134)'
PEF =

]r'a

fij

(' " -"'1' a" * J'- 1r ci.,)'a'].

12u

(3.139)

This is nominally infinite, but the change in potential energv is finite and is given by

PEt-PEF=sqBI'
(3.135),

'rr"

.-2,r1"14.,s

s-g4gr..

(3.140)

The initial kinetic energy is zcro, because the fluid is at rcst, and its fiinal value is' using

rr"I u,ax - uls-'!^ 1 I ' lu 2 J \JLal )0

".'"

a,

- snBLo

(J.141)

Thus one-third of the difference btwen the initial and final potntial energies is converted to khetic nergy, and this is trapped within a distance of the order of a deformation radius of the disturbance; the remainder, an amount gld l]6 is radiated away and lost to infinity. In ant ffnite region surrounding the initial aliscontinurty the final energy is less than the initial nergy.

3J,4 ' General initial conditions

Brause of the lincarity of the (linear) adjustment problem a spectral viewpoint is useful, in riich the fielals are represented as the sum or integtal of non'interacting lo]flifl modes. Ior

150

Chapter

3.

Shatlow water Systems and lsentropic Coordinates

exampl, suppose that the height field of the initial distu$ance is a two_dimensional field
given by

n(o) =

d&dl lJ iu(o)ei(t'{')

(3.142)

where the Foufier coefncients flr,r (0) are given, and the initial velocity field is zero Then the initial (and final) potential voticity field is given by

u=

-f;

.l.lio,,<ot

t'*u"

ul"or.

(3.143)

To obtain an expression for the final height and velocity fields, we express the potential vorticiw field as

'

fs- Jd dkdt

r,.l44)

The potenual

vo

rcity fild does not evolve, and it is rlatd to the initial height field by

fu,1= ffin*do).

(3.14s)

In the final, geosuophically batanced state, the potential vo{rcity is related to the height field by

a-, fa q=::v-n_:n

ano

ar,=

(,ftx, - t;)ir.,,
-r;r,.,,'t

. (3.146a.b)

where K2 = k'z+ r'z. Using (3.145) and (3.146), the Fouder componnts of the qnal height

field satisfy

7*'

r;)

nr,,

(3.r47)

ffrr(0) ^ n,.,=E,tri. lx physical space the final height field is just the spectml integxal of this, namely
n

(3.r48)

- lJr.,,",'*,,' arai - lJ

D^

*!;1"",''-

arar.

(3.r49)

we see that at large scales (f'zIe << 1) 4u is almost unchanged from its iniual state; the velocity field, which is then detrmined by geostrophic balance, thus adusts to the preexisting height field. At large scales most of the energy in geostrophicallv balanced flow is potential eneyi thus, it is nergetically easier for the velocity to change to come into balanc with tle height field than vice versa. At small scales, however, the final height field
bas much le(s variability ttan il did initially. Conversely, at small scales the height field adjusts to the velocity fild. To see this, let us suppose that the initial conditions contain vorticity but have zero height displacement. Specifically, if the iniual vorticity is V'zVJ(0), where q/(0) is the initial strearifunction, then it is straightforward to show that th final streamfunction is given by

q = 0-,, ei(r"') lJ

dkd t =

J!

rl4pffff2

auat

(3.1s0)

3.8 Ceostrophic

Adjustment

l5l

l2)

.3t ral

The final height fild is then obtaind ftom this, via geostrophic balance, by 4 = Ofo /9 ) q.,. Evidently, for small scales K2t, > 1) the sueamfunction, and hence the vortical component of the velocity field, arc alnost unaltered from their initial values. On the other hand, at large scales the final streamfunction has much less variability thar it does initially, and so the height field will be largely govemed by whatevr variatlon it (and not the velocity field) had initially. In general, the final state is a superposition of the states grven by (3.149) and (3.150). The divergent component of the initial velocity field does not affect the final stale because it has no potential vorticity, and so all of the associated energy is eventually lost to

infinity.
Fhally, we remark that just as in the pioblem with a discontinuous initial height profil, the change in total energy during adjustment is negative this can be seen from the form ofthe integals above, although we leave the specifics as a problm to th reader. That is, some of the iritial potential and kinetic enrgy is lost to inffnity, but some is trapped by the

.t)

Polendal voriicity constraint.

5)

3.8.5 A vaJiational perspctlve

ht
b)

In the non-rctating problem, all of the initial potential energy is eventually radiated away to inJinity. ln the rotating problem, the final stare contains both potential and kinetic energy. why is the energy not all radiated away to infinity? It is because potcntial vorticity conservation on parcels prevents all of the energy bcing dispersed. This suggests that it may b informative to thinl of the geostrophic adjustment ploblem as a varidtional ptoblem: r^.'e seek to minimize the energy consistent with the conservation of potential vorticity. We stay in the linar approximation in which, because the advection of potendal vorticity is neglected, potential vorticity remains constant at cach point. The energy of the flow is given by thc sum of potential and kinetic energies, namely
eneryy =
+ In@' sa'')

an,

(3.1s1)

(nhre d-4

= dx dy)

and the potential vorticity field is

a=t-kff=o.-.s
e

f"fi,

(3.rs2)

nhere the subscripts x,]y denote derivatives. Thc problem is then to extrmize the energy $bject to potntial vordcity conservatiorl This is a constrained problem rn the calculus of Eriations, sometimes calld an isopnmetric problem because of its origins in ma,yimizing ihe area of a sudace for a given perimetei,T The mathematical problem is to extremize the iDteglal

r-!ln@'*,'),on'

+ Alx,y)[11)"

- uy) fonlHll M,

(3.rs3)

siere ,\ (x, J', ) is a tagrange multiplier, undetern ned at this stage. It is a function of space: il it were a constant, the integral would mercly exEemize enryy subject to a given intgral of potential vorticity, and rearrangements of potential vorticity {which hcre we wish to
disallow) would leave the integral unaltered.

a. rh.ra ?ro ih.pp indehanrlanr w"ri,hlp. rhara,ra rhrea Fnl.r-I .hrhdp .nnrfinna rh*

152

Chapter

3.

Shallow Water Systems and lsentropic Coordinates

must be solved in order to minimize J. These arc

aL aaL
an
ax
AL

awher

aL_aaL _aaL_aaL_^ ax Au, ay du" "' Au ax aux

a

an,

aaL
ay

an"

-'
Ay Av,

aaL :0,

(3.1s4)

is the integrand on the dght hand side of (3.153). Substituting the expression for

into (3.154) gives, after

litde algebra,

tgn-AI^=4, -alal zu-)v

and then eliminating

0, zu ay-O,

(3.1ss)

gives the simple relationships

s an fooy'

(3.1s6)

which are the quations of geostrophic balance. Thus, in the linear apprcximati oI', geostrophic balance is the minimum energy state for a given field of potential vofticity.

3.9 ISENTROPICCOORDINATE5 return to the continuously stratified primitive equations, and considr the use of potential density as a vertical coordilrate, In pmctice this means using potential temperature in the atmospher and (for simple equations of state) buoyancy ir rh ocean; such coordinare systems are generically ca\ed isentropic coordinates, and sometimes iso/ycnal coorilinates if density iB usd. This may seem an odd thing to do but for adiabatic flov,' the res! ting equations of motion hav an attractive form that aids the interpretation of large,scale flow Th themod''namic equaoon becomes a statement for the conservation of the mass of fluid v{ith a given value of potential density and, because the flow of both the atmosphere and the ocean is largely along isentropic surfaces, the momentum and vorticity equations have a quasi-two-dimensional f orm. The particular choice of vertical coordinate is detemined by the form of the ihemo, d)'namic equation in the equation-set at hard; thus, if the thermod).namic equation is DdlDf = O, we traisform the equations from (r,JL7,z) coodinates to (x,Jr', d) coordinars. The material derivativ in this coordinate system is
We now

D A /a\ /a\ a- at-u\a*lu '\a"[

3.9.1 A hyd.ostatic Boussinesq fluid

DeA r' ae- at

A

Vr, 6$,

(3.1s7)

where the last term on the dght'hand side is zero for adiabaric flow.

In the simple Boussinesq equations (see the table on page 72) the buoyancy is the relevant thmodynamic variable. With hydrostatic balance the horizontal and vertical momentum quations are, in height coordinates, D1t

D'

+fxu=-v4,

O=

u4,

(3.1s8)

3.9 lsentropic Coordinates

t53

where b is the buoyancy, the variable analogous to the potential temperatue 0 of an ideat gas. The thermod),namic equation is

Db. Dt -'

(3.1s9)

and because = -gdp/po, isenftopic coordinates are the same as isopycnal coordinates. using {2,142) the horizontal pressur gradient may b transformed to isentropic coordi-

/aC\ _ tdo\ taz\ aC. -

\ax/,- \axlb

\ax/baz

faO\ "\ax/b6(ar\ (aM\

\ax/b
zb.

\ax,/b'

(3.160)

M=O

(3.t6r)

Thus, the horizontal momentum equation becomes

#+Jxr=-vbM.
where the material derivative is given by (3.157), with b replacing hyfu ostatic equation becomes
AM

(3.r62)

0.

UsinC

(3.f6r) the
(3.163)

ah- ''
A2

The mass continuity equation may be dedved by noting that for a Boussinesq fluid th mass element may be written as

6m = po-fr8b 5x 6y.
The mass continuity equation,

(3.164)

DamlDt = 0, becomes

D az az *i * avot ''

= o'

(3,16s)

whereV3r,=Vb

t, + a r / a, is the thee'dimensional derivative of the velocity ln isentropic coordinates. Equation (3.165) may thus be wfitten

Do + OVh .u =
where

ab (t-t

(3.r66)

o = azlab is a measure of the thickness beMeen two isentropic sudaces and the material derivative is given by (3.157) with d replaced by r. &uations {3.162), (3.163) and (3.166) compdse a closed set, $.ith dependent vadables r., M and z in the space of iDdependent variables x, / and D.
3.9.2 A hydrostalic ideal gas Dedving th equations of motion for this system requires a littl more work than in the Boussinsq case but the idea is the same. For an ideal gas in hydrostatic balance we hav, usinc (r.Ir2),

605T6D6T5alrT=7+

e=i

-=-oMt

(3.r67)

154

Chapter

3.

shallow Water Systems and lsentropic Coordinates

where M = cp T + O is the 'Montgomery potentiat', equal to the dry static eneryy. (We use some of the same slmbols as in the Boussinesq case to facilitate comparison, but tlreir meanings are slighdy diffrent.) hom this

DM

ae

n -"'

(3.r68)

wherc n = cpT le = cp(p/pi)R/" is the 'E .1ff function. Equation (3.168) represents the hydostatic relation in isentropic coordinates. Note also that M = dII + O. To obtain an appropdate form for the horizontat pressue gradient force f[sr note rhat, in the usual height coordinates, it is given by

v,p pwhere 17 =

ov

-n,

(3.r69)

crTld.

Using (2.142) gives

9!a 0v.n = ev6n ' gaz "6.

l :Vzp-(pVe7 - VeO-VeM. p
Thus, the horizontal momentum equation is

(3.170)

Then, using the definition of 11 and the hydrostatic approximation to help evaluate the vertical dedvativ, w obtain
(3.171)

D]+fxu=
Much as

vsM.

'll.l72)

ir

the Boussinesq case, the mass continuify equation may be deived by noting

that th mass elemnt may be witten as

t.a,\ae a^ - - gat * dt.

The mass continuity equation,

(3.173)

Ddn /Dt =

0, becomes

Ao Dt;a + ;6v1 v=t)

DaD

13.t74)

Itn aA E+ove.u= oi,

(3.162) and (3.166).

(3.175)

where norv d = apla6 is a measure of the (plessule) thickness between two isenropic surfaces. Equations (3.168), (3.172) and (3.175) foDn a ctosed set, analogous to (3.163),

3.9.3 * Analogy to shallow water equations

The quations of motion in isentropic coordinates hav an obvious anatogy wirh th shallow watr equations, and we may think of the shallow water quations as a finite-alifference

3.10

Available Potential Energy

I55

reprsentation of the primitive equations {'ritten in isentropic coordinates, or tbinl of the latter as the continuous Limit of the shatlow water equations as the number of layerc incrases. For example, consider a two-isentlopic_level representation of (3.168), (3 f72) and (3.175), in which the lower boundary is an isenhope A natural finite differencing gives

IIoAqa Mr-Mz=I Aet'

Mt =
$here th Ads are constants, and the momentum equations for each layer become

(3.176a)

(3176b)

Dff+7xu:
Dff
+

ddovrro

\3.I77a)

f xuz: -Mavrro

aelvnr.

(3.r77b)

Together with the mass continuity quation for each level thse are just like the two_laver shallow water equations (3.51). This means that rcsults that one might easilv derive for the shallow water quations will oftn have a continuous analog

3.IO AVAILABLE POTENTIAL ENERGY

We now revisit tlle issue of the internal and potential energv

stratifred flow, motivald by the folowirg remarks. In adiabatic, inviscid flow the total amormt of energy is conseNed' and ther are conversions betlveen intemal nrgv, potential energy and kinetic energy ln an ialal gas the potntiat energy anal the internal energv of a column extending thoughout thir sum is called the total potential the atmosphere are in a constant ratio to ach other energy. In a simple Boussinesq fluid, energetic conversions involve only th potential and kinetic energy, and not the intemal energy. Yet, plair v, in neither a Boussinesq fluid nor an ideal gas can ali the total potential eneryy in a fluid be conve ed to kinetic energy, fbr then all of th fluid would be aaljatnt to the ground and rhe fluid would have no thickness' which intuitively seems impossible. Given a state of th atmosphere or otan, how much of its total potential energy is available for conversion to kinetic energv? In particular, because total energy is conserveal only in adiabatic flow, w mav usefu y ask how much potential energy is available for conversion to kinetic energv undr an adiabatic rearrangement of fluid parcels? (or Suppose that at any given time the flowis slably stratified' but that the isentropes 11 The as in lig. 3 potntial are sloping, densi9 more generally the surfaces of constant potential energy of the system woulal be reduced iJ the isentropes wre flattened, for then heavrer fluid would be moved to lower altitudes, with lighter fluid replacing il at highr altitudcs. In an adiabatic rearrangemnt thc amount of fluid betwen the isentrcpes would

ill

remain constant, anal a state with flat isentropes (meaning parallel to the geopotntial surfaces) evidently constitutes a state of minimum lotal potential encrgy. The difference between the total potential energy of the fluid and the total potential enrgv after an adiabatic rearrangement to a state in which th isentropic surfaces are flat is called Ihc available
potential energy,

ot

AJE.3

t56 lnitial state

Chapter

3.

5hallow lvarer Systems and Irentropi( Coordinates

e3

Minimum potential energy stale

er
0z
Ar

warm, liSht
0z
01

cold, dense
Flg. 3.1 I lf a stably stratifed lnltlal state wlth s lopinq isntopes (left) ls ad iabatically rearranged then the state of rninimum potential energy has flat isennopes, as on

the right, but the amounr of fluid.onrained beMeen.ach isentroplc sufa(e

is

unchanged. Th differnce betwen the potential energles ofthe two gtates is the ava i la ble potentla I ene rgy.

3.lO.l A Boussinesq nuid

The potential energy of a column of a Boussinesq fluid of unit area is given by

r = - [' u,a" = - !'

b-a',.

(3.178)

and the potential energy of the entire fluid is given by the hodzontal integral of rhis. The minimum potential energy of th fluid aises after an adiabatic ranangement in which the isopymals are flattened, and the resulting buoyaDcy is only a function of z. The available potential energy is then the difrerence between the energy of the initial state and of this minimum state, and to obtain an approximate exprcssion for this we first integrate (3,178) by parts 1o give

-il:

22db-

l+li=;t:

22db-

b^H2

(3.179)

where ,a is the maximum value of b in the domain, and we may fomdly take th upper boundary to have this value of, without affcting the final result. The minimum potenoal energy state adses when z is a function only of r, z = Z(b) say. Because mass is conserved in the rearrangement, Z is equal to the horizontally avemgd value of z on a given isopycnal surface, Z, and the sufaces Z and t thus dfine each other completely. The average available potential energy, pr unit area, is then given by
APE

;J",,' -zz) db = ;)o

7db,

(3.r80)

where

z = Z + z'i that is, z' is t}le height variatio[ of an isopycnal surface, and the last term on the right'hand side of (3.179) has carcelled with an identical term in the expression for the potential energy of the re-arranged state. The available potential energy is thus

3.1

Available Potential Energy

157

2rz+ a2l -l _b'

ablaz

lb-olz)l

abt lz - 2(6\l ? z l.=z

_b (ab\ -. \a, )'

Fig.3.12 An lsopy(nal surfac, b = r, and the constant heisht sudac, z = Z. z is the height of the isopycnal surface after a rearrangement to a minimum potentialenergy state, equalto the average hi9ht ofthe lsopycnal suface. The values of z on the isopycnal surface, and of , on the constant height surface. (an be obtained by the Taylor expanrions shown. For an ideal gas in pressure coordinatet, replae z by p

proportional to the integial of the variance of the altitude of such a surfac, and it is a positive-definit quantity. To obtain an expression ir z coordinates, we express the height variatlons on an isopycnal surface in terms of buoyancy vadations on a surface of constant

hetht

by Taylor expanding th height about its value on th isopycnal surface. Referring to

FIg.3.12 this gives

.<at =z

fi\,_nta - aet =z - klo+b

(3.r81)

wljlre b' = b(z) - 6 \s conesponding buoyancy perturbation on the z surface and b is the averag value of b on the z sudace. Furthermor, azlabl=, - azlAb - @6lAz)-t, aLfi
(3.181) thus becomes

t =ab-z--a'(fi)"-dh
wz

(3.182)

w}lerc z' = z(Lt) - z is lhe height perturbatlon of the isopycnal surface, from its average ralue. Using (3.182) iD {3.180)we obtain an expression for the APE per ullit area, to 1\,lt

ece

dz = ',l^ -:_ abtaz

(3.r83)

The total A?E of the fluid is the horizontal inlegal of the above, and so is proportional to the vaiance of the buoyancy on a height surface. we emphasize that A?E is not defined for a single column of fluid, for it dpends on the vanations of buoyancy over a ho zontal surface. Note too that the derivation neglcts the effects of topographyi this, ard the us of abasic-state stratification, effectively restrict the use of(3.183) to a siryle ocean basin, and 6'en for that the approximations used limit the accuracy of the expressions.

I58
3.10.2 An idealgas

Chapter

3.

Shallow Vvater Systems and tsentropic Coordinates

The e\T'ression for the ApE for an ideal gas is obtaineal, mrrads mrrandb in the same way as for a Boussinesq fluid and the trusting reaaler may skip directly to (3.191). The intmal energy of an ideat gas column of unit ara is given by

r=l
r = [- oe,a.

cvTpdz =

r;Trde.
eaz =

(3.r84)

where pJ is the surface pressure, and the conesponding potential enrgy is given by

t'' zdp = l'

f'4'o,'

{3.r8s)

In (3.184)weuse hydrostasy, alldin(3.185) the equalities make successive use of hidrostasy, an integration by pafiq hydrostasy and the ideal gas relation. Thus, the total potential neruy GPE) is given by

or=,,r_9
TPE

Jo.rdv.

(3.186)

Using the ideal gas equation of state we can $,rit thjs as

after an integation by parts. (We omit a telm propotional to pJd, that arises in the integation by parts, because it cancels in a similar fashion to the boundary term in the Boussllesq derivation; ff taLe dr = 0.) The total porential energy of the entfe fluid is equal to a hodzontal integral of (3.187). The miDjmum totat potential energy arises when the pressue itr (3.187) is a function or y of A, p = P(d), where by conservation of mass p is the average value of the odginal pressure on the isentropic surface, p = U. The average available potential energ_y per unit area is thetr givn by the difference betweetr the initial
state and this mfuimnm, namely

?re). t^r= ^:i!all Gl.' d.e,

(3.187)

^*=;,**'';;l(hl.' - (;)-.'] *,
which is

(3.r88)

a positive'dfinite quantity. A useful approximation to this is obtained by exFessing the right-hand side in tems of rhe variance of the potential tmperature on a pressrue sMace. W fust use the binomial expansion to e).Tand px+l = (p + p')k+r. Neglecting third and higher-order tems (3.188) becomes

*'=*t(i,)'.'H *
Th vadable

(3.r89)

p' = p(d) 7is a pressuie perturbation on an isentropic sruface, ard is related to the potential temperature prturbation on an isobaric suface by [cf. (3.182)]

. "'"

! d0^- aqlAp
a.t

(3.rs0)

-Notes and

Problems
7

I59
surface. using

wherc 0'
r]
al

d(zJ is the potcntial temperature perturbatlon on thc (3.r90) in (3.I89) we finalty obtain

: 0(p)

aer' o" *'=YIi'u' I sar) \

ap

(3.r9r)

Th A?E is thus proportional to th variancc of the potntial temperature on thc pressure surface or, from (3.189), proportional to the va ance of the pressure on an lsentropic

3.1O.3 use, interpretation, and the atmosphere and o.ean The potcntial energy of a fluid ls reduccd s'hen the d)'namics acts to flatten the isntropcs Consider, for example, the Earth's atmospherc, with isentropes sloping upwards toward the pole (Iig. 3.Il with thc pole on the dght). Flattcning these isentropes amounts to a sinling of dcnse air and a rising of light air, and this reduction of potcntial nrgy leads to a corresponding production ofkinetic energy. Thus, if the dynamics is such as to reduce the temperaturc gradient between equator and pole by flattening the iscntropes thenAPEis conve ed to KE by that proccss. A statistically steady state is achieved because thc heating from the Sun continually acts to restore the horizontal temperature gradint betlveen equator and pole, thus rcplenishing the pool of APE, and to this extent the large-scale atmospheric cfculation acts like a hat cngine. It is a uscful exercise to calcl ate thc total potential eneryy, the available potntial energy and the kinetic energy of atmosphere and th ocan, (problem 3.16) one finds TPE>> APE> KE
(3.192)

bith, vcry approximarly, TPE - lo0APtand APE - I0 Kt Thc first inequalitv should not surpns us (as it was this that led us to definc APE in the ffst instance), but th second inequality is not obvious {and in facl the ratio is largcr in the ocean) It is rclated to the
fact that th instabilities of the atmosphcrc and ocean occur at a scale smaller than the size of the domain, and are unable to release all the potential enrgy that might be availabl fnderstanding this more fully is the topic of chaptrs 6 and 9.

The alsorithm to solve these equations numerically differs from that of the free surface shallowwater equations because the mass conservation equation can no longer be stepped foMard in time. Rather, an elliptic equation for prd must be derived bv eliminating time deivatives between (3.22) usinq (3.21), and this is then solved at each timestep.
relate the enerqy offronts to their slope ln this same paper the notion of available potential energy arose. The expression form draq'ls also commonly used in aerodynamics, and the two usages are related. ln aerodynamics, form draq is the force due to the pressure difference between the front and he rear of an object, or any other 'form', moving through a fluid. Aerodynamit

2 After N4argules (1903). Marqules sought to 3

t60

Chaprer

3.

Shallow Water Systems and lsentropic Coordinates

form draq may include frictional effects between the wind and the surface itself, but these are omitted in most 9ophysical uses.
0ules) Henri Poincare (l854-l912) was a prodigious French mathematician, physicist and philosopher, reqarded as one ofthe qreatest mathematicians living at the turn ofthe twentieth centurv. He is remembered for his oriqinal work in alqebra and topology, and in dynamical svstems and celestial mechanics, obtaininq many results in what would be called nonlinear dynamics and chaos when these fields re emerged some 60 yea6 later - the notion of 'sensitive dependence on initial onditions', for example, is present in his work. He also obtained a number of the results ofspeciaLrelativity independently ofEinstein, and worked on

the theory ofrotating fluids hencethePoincarwavesofthisthapter.Hewroteextensively and successfullyforthe generalpublir on the meanins, importanc and philosophyofscience. Among otherthings he discussed whether scientific knowledqe was an arbitrary convention, a notion that remains discussed and controversial to this day. (His answer: '.onvenlion', in part, yest arbitrary" no.) He was a proponent ofth role of intuition in mathematicaland scientinc prosress, and did not believe that mathematics could ever be wholly reduced to

Thomson (1869). w. Thompson later became Lord Kelvin. See Gill(1982) or Philander (i990)
6 As considered by Rossby

(l938). of

7 An introduction to variational problems may be found in weinstock (1952) and a number

other textbooks. Applications to many traditional problems in mechanics are discussed by Lanczos (1970).
l4argules (]903) introduced the.oncept of potntialenergy that is avaiable for conversion to kinetic energy, Lorenz (1955) clarified its meanins and derived useful, approximate formula
APE is just the non kinetic part of the pseudoenergy, an interpretation that naturally leads to a number ofxtensions of the concept. There are a host ofother papers on ihe subject, including that of Huang (1998) who looked at some ofthe limitations ofthe approximate expressions in an oceanic context.

for its computation. Shepherd (l993) showed that the

Gill, A. E., 1 9A2. Atnosphere-Ocean Dynamics.

This remains a good reference for geostrophic adjustment and gravity waves. The time
dependent geostrophic adjustment problem is discussed in section 7.3.

3.1

Derive the appropriate shallow water equations for a single movinq layer offluid ofdensity pr above a risid floor, and where above the movins fluid is a stationary fluid ofdensity p0, where po < pr. Show that as (polpr) 0 the usual shallow water equations emerge.

3.2 @ Model the atmosphere as two immiscible, 'shallow water' fluids of different density
stacked one above the other. Using reasonabl values for any required physical parameters, estimate the displacement ofthe interfacial surfa(e associated with a pole equator temperature gradient of 60 K, Similarly estimate an interfacial displacement in the ocean associated with a temperature gradienr of20K over a distance of4000km. Ohis is a cftrd representation ofthe main o.eani. thermocline.)

(n

3.3

ghzl2)l:

For a shallow water fluid the energy equation, (3.88), has the form arlat + tt(E + O. But fot a.ompressible fluid, the corresponding energy equation, (l.187), has

Notes and

Problems

I5l

the fom aElaf + v [r(I + p)] = 0 ln a shaLlow water fluid P +gh':l2 at a point sothe5e (Hint: What is equations are supefficially different Explain this and reconcile the two forms pressure in a fluid coLumn?) the averaqe 3.4 a can the shallow water equations for an incompressible fluid be derived by wav of an asymptotic expantion in the aspect ratio? lf so, do it That is, without assuminq hvdrostasv ab initio, expand the Euler equations with a free surface in a small parameter equal to the ratio ofthe depih ofthe fluid lo the horizontal scale ofthe motion, and so obtain the shallow water equations, 1.5 The inviscid shallow water equauons, whether rotating or not, can support gravity waves of arbitrarilyshort wavelenqths. For suffcientlv high wavenumber, the wavelength will be shorter than the depth ofthe fluid. ls this consistentwith an asvmptoiic nature ofthe shallowwater equations? Discuss. 3.6 Show that the venical velocity within a shallow water svstem is qiven by

'=
3.7 What
is the appropriate qeneralization

z-nrDh h Dr'

Dn, Dr

(rl.r)

lntrpret this result, showing that it qivs sensible answers at the top and bottom ofthe fluid

of(3.97) to two dimensionsT Suppose lhat attime I = 0 what theheightfieldisgivenbyaGaussiandktributionh'=Aer/"',where/'?=x':+it': is the subsequent evolution of this, in the linear approximation? Show that the distribution rema ns Gausslan, and that its w dth increases at speed 1'?F where n s the mean depth of
potentiaL
(P3.2)

1.8 ln an adiabat c shallow water fluid ln a rotating reference frane show that the
vorticitv conservation law is

'

!!!=o Dtn hb

where is the height ofth fiee surface and referenced to the same flat surface

h, c the height ofthe bottom topography' both

(4) A cyLirdricalcolumn ofalr at 30'latilude with radius 100 km expands hor zontallv to twice its originalradius. lfthe air is lnitially a! rest, what is the mean targentialvelociiv at the perimeter after ihe expansion? (r) An air colu m n at 60' N wlth zero re lative vorticill/ (( = 0) stretches frorn the su rface to the tropopause, which we assume k a rigid lid, at lokrn The air column moves zonallv on to a plateau 2,5 km hi9h. what is its relative vorticitv? suppose it then moves southwards to 30'N. What is its vor! city? (Assume that the densitv is constart.)

3.9 a

ln the long wave limit of Poincar waves, fluid parcels behave as free-aqentsl that is, ike free so id particles moving in a rotatlng frame unencumbered bv prssure forces Why lhen, is their frequencv qiven by o = = 2r) where C ls the rotation rate ofth coordinate system' and not by.o itsel? Do particles that are stationary or move in a straight line ir the nertlal frame of refrence satisfy the dispe6ion relationship for Poincar waves in this lirnit? Exp ain' lsee also Duran (1993), E99er (1999) and Phillips (2000) l

l.l o

Linearize the J-plane shallow water system about a state of iritialdisturbance qiven in the qeneralform n=

rest

Suppose that there is an

lJa.,

"'*"'at

ar,

(P3-3)

where 4 is the deviation surface heiqht and the Fourier coeffitjerc initialve ocity is zero.

irr

are given' and that the

162

Chapter

3.

Shallow Water Systems and lsentropic Coordinates

fr)
(r)

Obtain the seopotential field at the completion ofgeostrophic adjustment, and showthat the defomation scale is a natural lenqth scale in th problem. Show that the change in total energy d uring the ad.iustment is always less than or eq ual to zero. Neglect any initial divergence. N.B. Becalse the problem is linear, the Fou er modes do not inreract.

3.ll

One way of assimilatins obsewatjons into numerical models isviaa method known as 'nudgins' or 'robust diagnosrics'. ln this method, a relaxation tem is added to rhe right hand side of an volution equation, relaxins the fild back to observations. Thus, for example, DO/DI = othertrms + l(dobs - 0). lf 0 is potentialtmperature, is this mthod likely to work better in th ocean or in the atmosphere, as regards reproducing the large scale mean potential temperature and velocity fields? What steps might be taken to improve the method?

1.12 Geosttophtc adpstnent of d vetocityjunp Consider the evolution ofthe linarized J-plane shallowwater quations in an infinitedomain. Suppose that initially the fluid surface is flat, the zonalvelociry is zero and the meridionat
velocity is given by

?(t)

= 1,ossn(x)

(?3.4)

(a) Find the equilibium height and velocity filds at t = 6. fA what are the initial and final kinetic and potential energies?
The potential vorticity is 4 =

j04lH,

so that th initial and finat state are

(P3.s)

q = zl,a 6(x).

(why?) rhe final stare srreamlun.$on ir thus given bv (d' .A" L, q' . 4. wrlh solurion ) q = qoexp(xlLd) and q' = qoexp( xlla) fot x <0andr. > 0, where qd =rr'uo (whvt, and n = lov lg.lhe enetgy is E = I\Hu, + grtr)l2dt The initiatKE is infinite, the initiat PE is zero, and the final state has PE = KE = gLdntl4 that is, the energy is equipartitioned
between kinetic and potential energies.

3.13

ln the shallowwater equations show that, ifth flow is approximately gostrophicatty balanced, the energy at large scales is predominanrly potential enrgy and that nergy at smalt scates is predominantly kinetic energy. Dfrne precisely what 'large scat' and 'smalt scat' mean in scates the velocity

3.14 ln the shallowwatr geostrophic adjusrment problem, show rhar at large 3.15

adjusts to th heightfreld, and that at small scales the height fietd adjusts to the vetocity fietd.

Consider the problem of minimizing the full energy li.e., I(hrr)+gar)dtl, given rhe potential vofticity field 4(x,y) : (( + f)lh- Show thar the batance retations anatogous to (3.8.5) ar r,h = a@q t) lay and uh - a@4 r)/ax where B is the Bernou i function B = gtt +v212. Show thar steady flow does not necssarily satisf,/these equations. Discuss.
Using realistic values for temperature! velocity, etc.,.alculate approxinatevalues for the total potential energyi the available potentialenrgy and the kinetic energy, ofeither a hemisphere in the atmosphere or an ocean basin.

3.l6