Contents

4 Waves in the ocean 3

4.1 Poincaré Waves . . . . . . . . . . . . . . . . . . . . . . . . . . 3
4.2 Kelvin Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
4.3 Planetary, or Rossby, waves . . . . . . . . . . . . . . . . . . . 9
4.3.1 Phase and group speeds . . . . . . . . . . . . . . . . . 11
4.3.2 Quasi-geostrophic Rossby waves . . . . . . . . . . . . 15
4.3.3 Rossby waves in observations and models . . . . . . 18
4.3.4 Computing Rossby wave phase speeds . . . . . . . . 20
4.4 Kelvin and Rossby waves in the general oceanic adjustment 25
4.4.1 Implications in climate change scenarios . . . . . . . 25

1
Page 2
Chapter 4
Waves in the ocean

In this chapter we describe the general solution and dynamics of shallow-

water waves and try to put them into a wider context, emphasizing their
role in the ocean circulation and the coupled ocean-atmosphere system.
We will look for wave solutions of the shallow-water equations on the
f-plane and β-plane. A few solutions will appear, some have already been
discussed in different contexts, and some are new.

4.1 Poincaré Waves

We start by linearizing our shallow-water equations for a fluid h = H + η
over a state at rest

u = u′ (4.1)
h = H+η (4.2)

so that our equations, after eliminating higher order terms, reduce to

∂u ∂η
− f0 v = −g (4.3)
∂t ∂x
∂v ∂η
+ f0 u = −g (4.4)
∂t ∂y
∂η
+ H∇ · u = 0 (4.5)
∂t
A dispersion relation can now be obtained by looking for wave solutions
of the type
(u, v, η ) = (u0 , v0 , η0 )ei(kx+ly−ωt) (4.6)

3
into our linearized equations:

−u0 iω − f 0 v0 = − gη0 ik (4.7)

−v0 iω + f 0 u0 = − gη0 il (4.8)
−η0 iω + H (iku0 + v0 il ) = 0 (4.9)
Non-trivial solutions of the system exist only if the determinant is equal
to zero, so that
−iω − f 0 gik
f 0 −iω gil = 0 (4.10)
ikH ilH iω
and this is true if
ω [ω 2 − f 02 − gH (k2 + l 2 )] = 0 (4.11)
Now, there are a few interesting possible solutions for the frequency ω.
The first case is
ω=0 (4.12)
This solution describes a time-independent flow and the equations de-
scribe a geostrophically balanced flow.
The second possible solution is if

ω 2 = f 02 + c2 (k2 + l 2 ) (4.13)

where c = ( gH )1/2 is the gravity wave phase speed. The dispersion re-
lation describes wave solutions of superinertial flow (ω > f 0 ) which are
called Poincaré waves. From this solution we can highlight three possible
limiting cases (see Fig. 4.1).
First, the limit of no rotation, when f 0 = 0. The solution reduces to
ω = c2 K2 and the frequency solutions are
2

ω = ±Kc (4.14)

where K2 = (k2 + l 2 ), which describe a classical gravity wave.

Second, the short wave limit, when K2 ≫ f 02 /( gH ), which gives

ω 2 = c2 K 2 (4.15)
again, the dispersion relation is that of the non-rotating case with phase
speed c. This is because

(2π )2 f2 2π f ( gH )1/2
≫ ≫ λ≪ 2π (4.16)
λ2 gH λ c f

Page 4
so that Ld ≫ λ, where Ld is the Rossby radius Ld = ( gH )1/2 / f . Basically,
this solution looks like a gravity wave in a rotating case.
Third, the long wave limit, when K2 ≪ f 02 /( gH ). In this case we have

ω 2 = f 02 (4.17)

and therefore the Rossby radius is much smaller than the wave length,
Ld ≪ λ. In this limiting case, there is no space dependency, k = l = 0, and
the surface elevation anomaly is also zero η = 0. The solution is
∂u
− fv = 0 (4.18)
∂t
∂v
+ fu = 0 (4.19)
∂t
and these are called inertial oscillations, circulating at the planetary fre-
quency ω = f .

Figure 4.1: Dispersion relation for Poincarè and Kelvin waves. The frequency is
scaled by f and the wavenumber by Ld . The black dot marks the inertial oscilla-
tions regime and the ω = 0 is the geostrophic case.

Page 5
4.2 Kelvin Waves
Kelvin waves are a particular solution of the shallow water equations de-
scribing a gravity wave that exists in a rotating frame and with the help of
lateral boundaries. We could show Kelvin waves propagating in a chan-
nel, with two parallel boundaries, but for a start we will consider the case
of a single lateral boundary. The first assumption is that, if u = 0 at the
boundary, we could simply consider the zonal component of the velocity
zero everywhere. The meridional component is not zero at the boundary,
because the flow is frictionless. The linearized shallow water equations
are
∂η
− f0 v = −g (4.20)
∂x
∂v ∂η
= −g (4.21)
∂t ∂y
∂η ∂v
+H = 0 (4.22)
∂t ∂y

Continuity becomes, after differentiating with respect to y

∂η ∂2 v
= −H 2 (4.23)
∂t∂y ∂y

and using the momentum equation

∂2 v ∂2 v
= gH 2 (4.24)
∂t2 ∂y

which is the standard wave equation with phase speed c = ( gH )1/2 . The
solution to this is

v = V1 cos k (y − ct) + V2 cos k(y + ct) (4.25)

and the wave propagates along the meridional boundary. Substituting this
solution into the momentum equation we obtain a solution for η
c c
η = V1 cos k (y − ct) − V2 cos k (y + ct) (4.26)
g g

which describes a propagating wave in terms of surface elevation. The

solution has been found for both v and η with no Coriolis term: this has
the characteristics of a non-rotating shallow water wave. The velocity is

Page 6
 -y
+x -x

+x -x
-f +y -f
-f

Figure 4.2: For a system bounded to the west (x positive) the wave propagates in
the negative y direction, i.e. to the south. If x is negative this reverses so on the
eastern side of the basin the Kelvin wave propagates northwards. In the northern
hemisphere a Kelvin wave will keep the coast to its right as it pushed against it by
the Coriolis force.

in geostrophic balance with the pressure field, although it is a wave and

ω ∼ f.
The solutions in the x-direction are
V1 = e( f0 /c) x (4.27)
V2 = e−( f0 /c) x (4.28)
and remember that f 0 /c = L− 1
d . The first solution grows exponentially
for positive x away from the meridional boundary, which is not physically
possible. We are then left with the following set of solutions
v = e− x/Ld cos k (y + ct) (4.29)
u = 0 (4.30)
c
η = −e− x/Ld cos k(y + ct) (4.31)
g
= − ( H/g)1/2 e−x/Ld cos k(y + ct) (4.32)

Page 7
These are Kelvin waves. They are trapped by the meridional boundary
and decay exponentially away from it. The trapping spatial scale is given
by the Rossby radius, and for f 0 positive the boundary is at the right of
the wave propagation. Kelvin waves are balancing f against the wall,
which could be a topographic boundary or a waveguide such as the equa-
tor (Fig. 4.2).
Barotropic Kelvin waves are also tidal waves, propagating around an
amphidromic point .

Page 8
4.3 Planetary, or Rossby, waves
The time-dependent ocean circulation has an important impact on our cli-
mate due to the ocean large heat capacity. Any abrupt change, the intrinsic
variability and possible variations of the general circulation caused by the
atmospheric influence is fundamental in climate studies. Moreover, the
oceans are no longer considered passive in the atmosphere-ocean system,
but contribute to the production of the climate low-frequency variability
at interannual to decadal time scales (Talley, 1999; Dewar, 2001; Pierce et al.,
2001).
The discovery of Planetary waves by the solution of Laplace’s equation
as the second class waves dates back to the late nineteenth century by
Hough (1897). Later C.G. Rossby pointed out the characteristic of these
waves, hence they carry his name and are also called Rossby waves.
Since then, Rossby wave theory is well known (Gill, 1982; Dickinson,
1978; Leblond and Mysak, 1981) and is usually applied to an ocean at rest
with uniform depth. Rossby waves owe their existence to the meridional
variation of the Coriolis force (the β effect) and therefore propagate fol-
lowing an east-west waveguide, as the conservation of potential vorticity
is their restoring force. These kinds of waves, whose frequencies are con-
siderably lower than those of gravity waves and are subinertial (ω ≪ f ),
are also sometimes called quasigeostrophic waves, with a dynamic evolu-
tion depending on the departure from geostrophy.
The generation of these waves is still not completely understood but
the main forcing is wind stress and buoyancy forcing, though the latter is
thought to act in a minor way, and upwelling-downwelling on the eastern
boundary (Leblond and Mysak, 1981; Gill, 1982).
The oceans are forced at the surface by the wind frictional stress and
Rossby waves appear to play a fundamental role in redistributing and dis-
persing large-scale time-varying energy in the ocean. The propagation
of Rossby waves towards the ocean interior under the influence of wind
stress results in establishing a Sverdrup balance in the basin, accumulating
energy in the western boundaries and intensifying currents there (Ander-
son and Gill, 1975, 1979).
Due to the ubiquitous presence of Rossby waves in the world oceans
they influence ocean gyres and air-sea fluxes at all latitudes, affecting in
turn the atmospheric heat transport and circulation. They are believed
to provide teleconnections between the equatorial and middle latitudes
regions (Galanti and Tziperman, 2003) as well as transhemispheric and in-
terbasin communications (Cessi and Otheguy, 2003). Other major effects are
the maintenance and intensification of western boundary currents, trans-

Page 9
port of a large amount of heat and, because of their time-scale, they play a
key role in the climate system.
Rossby waves are very long waves so that the f -plane is not a good
approximation anymore and we will build our solutions on the β-plane.
The frequency is going to be subinertial, ω ≪ f , and so they are close to
geostrophy.
Our set of equations is
∂u ∂η
− ( f 0 + βy) v = − g (4.33)
∂t ∂x
∂v ∂η
+ ( f 0 + βy) u = − g (4.34)
∂t ∂y
 
∂η ∂u ∂v
+H + = 0 (4.35)
∂t ∂x ∂y

Given that ω ≪ f , ∂t∂ ≪ 1 and βL/ f 0 ≪ 1 we can approximate the mo-

mentum equations to a geostrophic flow
∂η
− f0 v = −g (4.36)
∂x
∂η
f0 u = −g (4.37)
∂y
and adding these geostrophic solutions to the shallow water equations

∂η g ∂η g ∂2 η
− f 0 v = − g + βy + (4.38)
∂x f 0 ∂x f 0 ∂y∂t
∂η g ∂η g ∂2 η
f0 u = −g + βy − (4.39)
∂y f 0 ∂y f 0 ∂x∂t
The first part of the momentum equations is that of a geostrophic flow
and the remaining is the small contribution from variations induced by
the ageostrophic component. The last terms will be responsible for the
propagation of Rossby waves.
Using continuity and (4.38)-(4.39) we arrive to
∂η ∂η
− L2d ∂t ∇2 η − βL2d =0 (4.40)
∂t ∂x
which is a leading order approximation to the potential vorticity equation
describing a quasi-geostrophic flow
  ∂η
∂ t ∇2 η − L −
d
2
η +β =0 (4.41)
∂x

Page 10
 Now we can look for Fourier type solutions in the form η = η0 ei(kx+ly−ωt)

βk
ω=− (4.42)
( k2 + l 2 ) + L−
d
2

or alternatively
k
ω = − βL2d (4.43)
1+ L2d (k2 + l2)
Evidently, on the f -plane (β = 0) the solution reduces to a geostrophic
flow and no wave is allowed to propagate. The meridional gradient in f
is thus the restoring force for Rossby waves.
Two possible cases can be envisaged, setting l = 0.
First, that of short waves, where L ≤ Ld and therefore kLd ≥ 1, for L a
typical scale of the wave length and k a typical scale of the wave number.
In this case the dispersion relation reduces to

k β
ω = − βL2d =− = − βL (4.44)
L2d (k2 ) k

Given that we are on the β-plane approximation, βy ≪ f 0 → βL ≪ f 0 →

ω ≪ f 0 , confirming the subinterial period.
Second, waves could have very long wave length L ≥ Ld or kLd ≤ 1

L2d k2 β
ω = − βL2d k = − β ≪ − = − βL (4.45)
k k
and therefore ω ≪ f 0 . The period of Rossby waves is always subinertial.

4.3.1 Phase and group speeds

Keeping l = 0 and using the following scaling

ω = ω̂βLd (4.46)
κ̂
κ = (4.47)
Ld

The dispersion relation takes the form

κ̂/Ld
ω̂βLd = − βL2d 2 (4.48)
1 + L2d Lκ̂ 2
d

Page 11
 κ̂/Ld κ̂
ω̂ = − Ld = − (4.49)
1 + κ̂ 2 1 + κ̂ 2
and for |κ̂ | = −1 the frequency takes the value |ω̂ | = −0.5 (see Fig. 4.3).
The phase speed of Rossby waves is easily computed (with l = 0),
using (4.42)
ω −β
cp = = (4.50)
k k2 + L−
d
2

it is always negative and larger for long waves .

For long Rossby waves, the phase velocity is approximated by

c p = ω/κ = − βL2d (4.51)

which is strictly westward even if l ̸= 0.

Figure 4.3: Rossby wave dispersion relation, phase and group speeds. Phase
velocities, C p , are always westward. Group velocities, Cg , are westward for long
waves (κ < −1) and eastward for short waves (κ > −1). The cutoff frequency is
set at βLd /2.

Page 12
 The phase speed is always negative but is the energy flux always di-
rected westward? this does not seem possible.
The group velocity is obtained by differentiating the dispersion rela-
tion
∂ω ∂ω
cg = ( , ) = β (κ 2 − l 2 − L −
d
2
, 2κl )/(κ 2 + l 2 + L−
d
2 2
) (4.52)
∂κ ∂l
or, by setting l = 0
βκ 2 − βL−
d
2
cg = (4.53)
(κ 2 + L − 2 2
d )
By setting k = 0, group and phase velocities are now equal c p = c g =
− βL2d (see Fig. 4.3), and long waves are therefore non-dispersive.
If l ̸= 0 the dispersion relation takes the form of the dispersion diagram
in Fig. 4.4. The group velocity, the gradient of the frequency in wavenum-
ber space, is normal to the contours and inversely proportional to the spac-
ing between contours. The hyperbola separating waves with eastward and
westward group velocity is shown by the dashed line and is κ 2 = l 2 + L− 2
d .
Frequency contours reduce to a single point when ω = 0.5βLd and κ = Ld .

Page 13
Figure 4.4: Rossby wave dispersion diagram. Contours are of frequency in units
of β Ld . The group velocity, the gradient of the frequency in wavenumber space,
is normal to the contours and inversely proportional to the spacing between con-
tours. The hyperbola separating waves with eastward and westward group veloc-
ity is shown by the dashed line and is κ 2 = l 2 + L− 2
d . Frequency contours reduce
to a single point when ω = 0.5βLd and κ = Ld .

Page 14
4.3.2 Quasi-geostrophic Rossby waves
In order to obtain and describe the Rossby wave solutions, we consider the
linearised quasi-geostrophic (QG) potential vorticity equation (Pedlosky,
1987):
∂t qi + J (ψi , qi ) = 0, (4.54)
where J ( a, b) = a x by − ay bx is the Jacobian and ψ the stream function.
Introducing a plane wave solution of the type ψ = Ψei(kx+ly−σt) into (4.54)
we naturally obtain the dispersion relation for Rossby waves, showing
their basic characteristics (Leblond and Mysak, 1981; Gill, 1982)

βk
ω=− ,
( k2 + l 2 ) + L−
d
2

where ω is the frequency, k and l are the horizontal wavenumber, β is the

meridional variation of the Coriolis parameter and Ld the Rossby radius
(C2 / f 2 ). It is clear that Rossby waves have westward phase velocities (of
the order of a few cm/s) and that these are increasing toward the equator
(where equatorial wave theory holds) with a maximum speed C p = βL2d .
Group velocities, Cg , in the case of long waves, are westward and the
waves are nondispersive (Cg = C p ), while short waves propagate east-
wards but with very slow speeds.
Another remarkable feature of the planetary wave dispersion relation
is that not all frequencies exist, with a cutoff frequency at 12 βLd .
Besides the horizontal problem, the vertical one is of great importance.
Using a normal mode representation (Leblond and Mysak, 1981), separating
the vertical and horizontal structure, we find an infinite set of solutions
(or normal modes). The zeroth is the barotropic one, almost vertically in-
dependent and very rapid; the other solutions, or modes, are called baro-
clinic with decreasing phase speeds and increasing oscillation in the ver-
tical. A first-mode baroclinic Rossby wave takes months to years to cross
an ocean basin, depending on the latitude.

A 3-layer model
In the case of a 3-layer ocean, the potential vorticities are given by

q1 = ∇ψ1 + βy − F11 (ψ1 − ψ2 )

q2 = ∇ψ2 + βy − F21 (ψ2 − ψ1 ) − F22 (ψ2 − ψ3 )
q3 = ∇ψ3 + βy − F32 (ψ3 − ψ2 ),

Page 15
where Fm,n = f 02 /( Hm gn′ ) and gi′ and Hi are the reduced gravities and layer
depths respectively.
For this 3-layer system, substitution of a plane wave solution leads to
a generalised eigenvalue problem of the form AΨ = ωBΨ, or explicitly:
     
β1 0 0 ψ1 − G1 1 0 ψ1
 0 β 2 0   ψ2  = ω  G2 − G3 1   ψ2  ,
0 0 β3 ψ3 0 1 − G4 ψ3

where β 1 = (kβ)/F11 , β 2 = (kβ)/F22 , β 3 = (kβ)/F32 and G1 = (K2 +

F11 )/F11 , G2 = F21 /F22 , G3 = (K2 + F21 + F22 )/F22 , G4 = (K2 + F32 )/F32 ,
where K2 = k2 + l 2 .
The solution of the system is plotted in Fig.4.5 and it describes the ba-
sic properties of Rossby wave propagation. In fact, for the 3-layer system,
the dispersion relation is found on the upper panel and both phase and
group velocities on the bottom panel of Fig.4.5. We can distinguish the
barotropic mode with increasing frequencies towards long wavelengths,
very fast phase speeds and positive (eastward) group velocities. The baro-
clinc modes have smaller frequencies, their phase velocities are always
westward but their group velocities turn from westward to eastward at
the point of maximum frequency

kLd = |1| and ω ( βLd )−1 = |0.5| (4.55)

where the group velocity is zero. Therefore, long baroclinic waves direct
their energy westward while short waves direct it eastward. This means
that, in the limit of long wavelengths, the phase and group speeds are
the same and the waves are nondispersive. On the other hand, for short
waves phase and group speeds differ and the waves are dispersive. The
maximum group and phase velocity (C p = Cg = − βL2d ) are attained for
long waves, they are to the west and can be found on the axis origin of the
dispersion relation.
The system could be extended to an N-layer or even to a continuously
stratified ocean. In every case, the solutions obtained are one barotropic
and N-1 baroclinic modes of decreasing phase speeds. This method of
analysis is called the normal modes method, in which the ocean is decom-
posed into an infinite set of solutions (or modes): one barotropic (or exter-
nal) and the remaining baroclinic (or internal).

Page 16
 2
1.5
1
Frequency

0.5
0
−0.5
−1
−1.5
−2
−4 −3 −2 −1 0 1 2 3 4

2
Phase & Group Speed

1.5
1
0.5
0
−0.5
−1
−1.5
−2
−4 −3 −2 −1 0 1 2 3 4
Wavenumber
Figure 4.5: Upper panel: the dispersion relation for the barotropic and
first two baroclinic modes of the 3-layer QG ocean. Shown are values
of both positive and negative wavenumbers. The wavenumber is scaled
by the deformation radius Ld and the frequency by βLd ; the meridional
wavenumber l is set to zero. The first baroclinic mode frequency reaches
a maximum at ωmax = βLd /2, i.e. ωmax = |0.5|. Lower panel: phase (solid
lines, C p = ω/k) and group (dashed lines, Cg = ∂ω/∂k) velocities of the
barotropic and first two baroclinic modes, scaled by βL2d .

Page 17
4.3.3 Rossby waves in observations and models
Chelton and Schlax (1996) presented for the first time the results of these
observations identifying clear Rossby waves signals (Fig.4.6) and common
features like the increase of phase speed in the western basin, the effect
of bottom topography, eastward propagating equatorially trapped Kelvin
waves and pulses related to El Niño events.
As anticipated, the advent of satellite altimetry brought a powerful
tool to describe Rossby waves in the real ocean. The TOPEX/POSEIDON
(T/P) altimeter is able to detect long baroclinic planetary waves unam-
biguously over the entire world ocean (Fig. 4.7).
The T/P altimetry data reveal the sea surface height anomalies (SSHA)
and to analyse this data time-longitude plots, known as Hovmöller dia-
grams, are used, which clearly show Rossby waves as diagonal alignments
of crests and troughs moving westward. An example of this is given in
Fig.4.7, where SSHA data from the Indian Ocean are plotted for the lati-
tude 20o S from 1993 till May 2005; in the left panel the row data are plotted
while in the right panel the data have been filtered with a westward filter
to better show Rossby wave propagation.
By this technique, Rossby waves are detected in all basins and altime-
try has been used also in the Southern Ocean (Hughes, 1995) where two
dynamical systems were found, a supercritical and a subcritical one with
respect to Rossby waves, the first one being able to advect the waves east-

Figure 4.6: Sea surface height anomalies showing the propagation of plan-
etary waves in the Pacific Ocean. Also clear is the β-effect inducing larger
phase speeds towards the equator [from Chelton and Schlax (1996)].

Page 18
 SSHA at 20° S, unfiltered SSHA at 20° S, westward-filtered

2004 2004
0.24 0.24

0.2 0.2

2002 0.16 2002 0.16

0.12 0.12

2000 0.08 2000 0.08

0.04 0.04

0 0
1998 1998

-0.04 -0.04

-0.08 -0.08
1996 1996
-0.12 -0.12

-0.16 -0.16

1994 1994
-0.2 -0.2

60 80 100 60 80 100

Figure 4.7: Time-longitude plot of the sea surface height anomalies (in me-
ters) in the Indian Ocean at 20o S. On the left panels, the original altimeter
data. On the right panel, the corresponding westward-filtered signature.
There is a clear evidence of crests and troughs propagating westward with
a biannual period (Courtesy of P. Cipollini).

Page 19
ward.
Rossby waves are also detected by other sensors like the Along-track
Scanning Radiometer (ATSR) in sea surface temperature (SST) and, re-
cently, SeaWiFS in ocean colour.
As an example, Hill et al. (2000) used a SST record to compute Rossby
wave phase speeds finding good agreement with Killworth et al. (1997).
They were also able to detect topographic effects such as those predicted
by Killworth and Blundell (1999).
One of the latest applications has been using ocean colour. Cipollini
et al. (2001) found for the first time Rossby waves in SeaWiFS datasets,
although they are neither very clear nor ubiquitous. A preliminary expla-
nation for this detectability was in term of the vertical displacements of the
thermocline associated with the Rossby wave and subsequently changes
in the nutrient upwelling.

4.3.4 Computing Rossby wave phase speeds

In order to compute the gravity wave phase speeds and Rossby radii of de-
formation we need to solve the generalized eigenvalue problem of Sturm-
Liouville form:
d2 ϕ N 2 ( z )
+ ϕ=0 (4.56)
dz2 C2
subject to the following boundary conditions
ϕ=0 at z = 0, − H (4.57)
where H is the local mean water depth and N 2 is the Brunt-Väisälä fre-
quency, computed from the potential density method as outlined in Chel-
ton et al. (1998). Solution of the system (4.56)-(4.57) leads to an infinite set
of eigenvalues Cm −2 , the baroclinic gravity wave phase speeds, and corre-

sponding eigenfunctions ϕm .
However, Chelton et al. (1998) showed that a WKB approximation of
the gravity wave speed is generally in good agreement with the solution
given by the system (4.56)-(4.57), and this is:
Z 0
WKB −1
Cm ≈ Cm = (mπ ) N (z)dz, m ≥ 1. (4.58)
−H

Then, within the extratropical regions, the Rossby radii of deformation

are simply found by applying
Cm
Lm
d = . (4.59)
| f (θ )|

Page 20
We will be focusing on extratropical regions only (|θ | ≥ 10◦ ), leaving the
equatorial wave dynamics response aside.
Now, we can compute the unperturbed long Rossby wave speeds
2
cm = − βCm / f 2. (4.60)

Since model data provide potential density ρθ , we can compute the

stratification directly from the potential density method of the non-equispaced
vertical levels k:
h i
N 2 (z) = − g/ρ0 ρθ (z) − ρθ (z + 1)/(δk (z) − δk (z + 1)) (4.61)

The gravity wave speed can be obtained from two different method.
First, as a good approximation, we can infer it from the WKB method as
suggested in Chelton et al. (1998):
Z 0
−1
Cm = (mπ ) N (z)dz, m ≥ 1. (4.62)
−H

After obtaining N 2 and Cm , the Rossby radii of deformation are readily

computed as

Cm
Lm
d = , | θ | ≥ 5◦ (4.63)
| f (θ )|
Cm
Lm
d = , | θ | ≤ 5◦ . (4.64)
2| β(θ )|

or, for the extratropical band: Lm −1 0 N ( z )dz

R
d = (| f | mπ ) −H
For the linear, long and extratropical waves, we can simply compute
the Rossby wave phase speed as

cm = − β( Lm
d)
2
(4.65)

Page 21
Figure 4.8: A global contour map of the baroclinic gravity wave phase speed [from
Chelton et al., 1998] and its zonal mean.

Page 22
Figure 4.9: A global contour map of the baroclinic Rossby radius of deformation
and its zonal mean. [data from Chelton et al., 1998]

Page 23
Figure 4.10: A global contour map of the baroclinic Rossby wave phase speed and
its zonal mean. [data from Chelton et al., 1998]

Page 24
4.4 Kelvin and Rossby waves in the general oceanic
adjustment
The importance of Rossby waves in the spinup of the ocean and in the
adjustment of the ocean interior was also recently shown by Johnson and
Marshall (2002). They proposed a theory for surface Atlantic response to
thermohaline variability; in their work they study the reaction of the ocean
to a perturbation of the rate of deep water formation at high latitudes.
These changes initiate Kelvin waves which propagate along the western
boundary, in a similar response of that demonstrated by Kawase (1987),
and then cross the basin as equatorial Kelvin waves until they reach the
eastern boundary where they propagate northwards and southwards. The
final part of the response is the radiation of Rossby waves from the east-
ern boundary, communicating the thermocline displacement to the ocean
interior which is clearly illustrated with a series of snapshots (Fig. 4.11).

4.4.1 Implications in climate change scenarios

Saenko (2006) recently showed that, within the IPCC models, there is clear
evidence of an increase of the first baroclinic Rossby radius with increas-
ing oceanic stratification in the warmer climate. The changes range from
15 to 20% depending on the model and latitude. This would imply a
greater length scale for mesoscale eddies and modified characteristics for
oceanic Rossby waves, whose speed is proportional to the squared baro-
clinic Rossby radius of deformation. Also, the adjustment time scale in the
ocean would decrease as well as in any ocean-atmosphere climate variabil-
ity process where Rossby waves set the dominant period. Equally impor-
tant, if not more in certain basins, is the change in the background baro-
clinic mean flow and its subsequent effect on the propagation of Rossby
waves. This effect was not considered in Saenko (2006).
Modifications to the background stratification and mean flows are ob-
served between pre-industrial and climate-change runs in the GFDL CM4
model (Fig.4.12). However, the question of the quantification of these ef-
fects on the Rossby wave activity, as well as the changes induced by a
modified background mean flow, is still unanswered. We expect to show
considerable alterations to the Rossby wave phase speeds at different lat-
itudes, leading to important changes in the ocean adjustment time-scale
and coupled ocean-atmosphere interactions where Rossby waves set the
clock (Fig.4.13 and Fig.4.14).

Page 25
Figure 4.11: Surface layer thickness after a thermohaline overturning of 10 Sv
is switched on at time t = 0 in the northwest corner of an ocean initially at rest.
There is no wind forcing, and the surface layer is initially 500 m deep. The contour
interval is 2 m, and thicknesses less than 499 m are shaded. Note that the thickness
anomaly on the western boundary is much greater than that in the interior. [from
Johnson and Marshall, JPO2022]

Page 26
Figure 4.12: The baroclinic gravity wave phase speed computed from the GFDL-
CM4.0 model under historical conditions for years 2010-2014 (in black) and for
the future scenario SSP585 (in red).

Page 27
Figure 4.13: The Rossby radius computed from the GFDL-CM4.0 model under
historical conditions for years 2010-2014 (in black) and for the future scenario
SSP585 (in red).

Page 28
Figure 4.14: The baroclinic Rossby wave phase speed computed from the GFDL-
CM4.0 model under historical conditions for years 2010-2014 (in black) and for
the future scenario SSP585 (in red).

Page 29
Page 30
Bibliography

Anderson, D. L. T., and A. E. Gill, Spin-up of a stratified ocean, with ap-

plication to upwelling, Deep-Sea Res., 22, 583–596, 1975.
Anderson, D. L. T., and A. E. Gill, Beta-dispersion of inertial waves, J.
Geophys. Res., 84, 1836–1842, 1979.
Cessi, P., and P. Otheguy, Oceanic teleconnections: Remote response to
decadal wind forcing, J. Phys. Oceanogr., 33, 1604–1617, 2003.
Chelton, D. B., and M. G. Schlax, Global observations of oceanic rossby
waves, Science, 272, 234–238, 1996.
Chelton, D. B., R. A. deSzoeke, M. G. Schlax, K. E. Naggar, and N. Siwertz,
Geographical variability of the first baroclinic rossby radius of deforma-
tion, J. Phys. Oceanogr., 28, 433–460, 1998.
Cipollini, P., D. Cromwell, P. G. Challenor, and S. Raffaglio, Rossby waves
detected in global ocean colour data, Geophys. Res. Lett., 28, 323–326,
2001.
Dewar, W., On ocean dynamics in midlatitude climate, J. Climate, 14, 4380–
4397, 2001.
Dickinson, R. E., Rossby waves-long period oscillations of oceans and at-
mosphere, Annu. Rev. Fluid Mech., 10, 159–195, 1978.
Galanti, E., and E. Tziperman, A midlatitude-enso teleconnection mech-
anism via baroclinically unstable long rossby waves, J. Phys. Oceanogr.,
33, 1877–1888, 2003.
Gill, A. E., Atmosphere-Ocean Dynamics, International Geophysics Series, vol.
30, 662 pp., Academic Press, 1982.

31
Hill, K. L., I. Robinson, and P. Cipollini, Propagation characteristics of ex-
tratropical planetary waves observed in the atsr global sea surface tem-
perature record, J. Geophys. Res., 105, 21,927–21,945, 2000.

Hough, S., On the application of harmonic analysis to the dynamical the-

ory of the tides, part i, on laplace’s oscillations of the first species, and on
the dynamics of ocean currents, Phil. Trans. Roy. Soc. Lond., 189, 201–257,
1897.

Hughes, C. W., Rossby waves in the southern ocean: A comparison of

topex/poseidon altimetry with model predictions, J. Geophys. Res., 100,
15,933–15,950, 1995.

Johnson, H. L., and P. Marshall, A theory for the surface atlantic response
to thermohaline variability, J. Phys. Oceanogr., 32, 1121–1132, 2002.

Kawase, M., Establishment of deep ocean circulation driven by deep-

water production, J. Phys. Oceanogr., 17, 2294–2317, 1987.

Killworth, P. D., and J. R. Blundell, The effect of bottom topography on

the speed of long extratropical planetary waves, J. Phys. Oceanogr., 29,
2689–2710, 1999.

Killworth, P. D., D. B. Chelton, and R. A. de Szoeke, The speed of observed

and theoretical long extratropical planetary waves, J. Phys. Oceanogr., 27,
1946–1966, 1997.

Leblond, P. H., and L. A. Mysak, Waves in the Ocean, 1st ed., 602 pp., Else-
vier Oceanography Series, 1981.

Pedlosky, J., Geophysical Fluid Dynamics, 2nd ed., Springer-Verlag, 724 pp,
1987.

Pierce, D., T. P. Barnett, N. Schneider, R. Saravanan, D. Dommenget, and

M. Latif, The role of ocean dynamics in producing decadal climate vari-
ability in the north pacific, Climate Dyn., 18, 51–70, 2001.

Saenko, O. A., Influence of global warming on baroclinic rossby radius in

the ocean: a model intercomparison, J. Climate, 19, 1354–1360, 2006.

Talley, L., Simple coupled midlatitude climate models, J. Phys. Oceanogr.,

29, 2016–2037, 1999.

Page 32