Atmosphere, Ocean and Climate Dynamics

Answers to Chapter 6

1. Consider the typical, zonally averaged flow, u, shown in Fig.5.20. Con-

centrate on the vicinity of the subtropical jet near 30 o N in winter
(DJF). If the x-component of the frictional force per unit mass is

Fx = ν∇2 u ,

where the kinematic viscosity coefficient is ν = 1.34 × 10−5 m 2 s −1 and

∇2 ≡ ∂ 2 /∂x2 + ∂ 2 /∂y 2 + ∂ 2 /∂z 2 . Compare the magnitude of this east-
ward force with the northward or southward Coriolis force and thus
convince yourself (and me!) that the frictional force is negligible. [ 10o
of latitude ' 1100km; the jet is at an altitude of about 10km. You
should find that an order-of-magnitude calculation will suffice to make
the point unambiguously.]
Frictional force is
U
Fx = ν∇2 u ∼ ν
H2
where H is the scale over which the zonal wind of speed U varies.
The Coriolis force is
C ∼ fU

Thus the ratio of the two terms is:

frictional ν HU2 ν
∼ = 2 = 10−9
Coriolis fU H f

if H = 10km. Thus the frictional term is utterly negligible compared

to Coriolis.

2. Using only the equation of hydrostatic balance and the rotating equa-
tion of motion, show that a fluid cannot be motionless unless its den-
sity is horizontally uniform. (Do not assume geostrophic balance, but
you should assume that a motionless fluid is subjected to no frictional
forces.)

1
 The eq. of motion in a rotating frame is
Du 1
z × u = − ∇p − gb
+ fb z+F .
Dt ρ
If friction vanishes in a fluid that is everywhere motionless then, for
u = 0, the horizontal gradients of pressure must be zero everywhere,
since it would otherwise be the only nonzero term in the horizontal
components of the eq. of motion. But, from hydrostatic balance
∂p
= −gρ ,
∂z
∂
∇p = −g∇ρ .
∂z
So, if the horizontal gradient of pressure is everywhere zero, the hor-
izontal gradient of density must also be zero. So a motionless fluid
requires no horizontal gradients of density; conversely, a fluid cannot
be motionless unless density is horizontally uniform.
3. a. What is the value of the centrifugal acceleration of a particle fixed
to the earth at the equator and how does it compare to g? What is the
deviation of a plumb line from the true direction to the centre of the
earth at 45◦ N?
Ω

A 45

g γ

ο
45

The centrifugal acceleration at the equator is just Ω2 a, where Ω =

2π/86400 = 7. 27 × 10−5 s−1 is the Earth’s rotation rate and a is the
Earth radius, which we here take to be 6370 km. Then the cen-
trifugal acceleration at the equator is just Aeq = Ω2 a = 0.034 ms−2 ,
a factor 3. 5 × 10−3 smaller than gravity. At 45o N, the centrifugal
acceleration is A45 = Ω2 r45 , where r45 = a cos (450 ) is the distance

2
from the surface to the rotation axis. Then A45 = 0.024 m s−2 , di-
rected outward, normal to the rotation axis, as illustrated in the figure.
This acceleration has a vertical component (normal to the surface) of
A45 cos (45o ) = 0.017ms−2 and a horizontal component (directed equa-
torward) of A45 sin (45o ) = 0.017 ms−2 . Hence the net (gravity + cen-
trifugal) acceleration is directed at an angle γ to the downward vertical,
where
A45 sin (45o )
tan γ = = 0.0017 .
g − A45 cos (45o )
Since this number is small, we may use the small angle approximation,
so that γ = 0.0017 rad = 0.097o .
b. By considering the centrifugal acceleration on a particle fixed to
the surface of the hearth,iobtain an order-of-magnitude estimate for the
earth’s ellipticity r1r−r
0
0
, where r1 is the equatorial radius and r0 is the
polar radius. As a simplifying assumption the gravitational contribution
to g may be taken as constant and directed toward the centre of the
earth. Discuss your estimate given that the ellipticity is observed to be
1
297
. You may assume that the mean radius of the earth is 6000 km.
The total potential Φ (incorporating both gravity and centrifugal ef-
fects) for a spherical earth is given by Eq.(6.30):
1
Φ = gz − Ω2 r2 ,
2
where r is the distance from the rotation axis and z the height above
the spherical reference surface of radius rref . Note that we have here
assumed that gravity is directed toward the earth’s center, with a con-
stant magnitude g. Now, we can estimate the shape of the earth by
assuming the actual surface to be one of uniform potential. Assum-
ing variations of surface height to be small, r ' rref cos ϕ, where ϕ is
latitude. Then the surface is defined by
1
gz(ϕ) = C + Ω2 rref 2
cos2 ϕ ,
2
where C is a constant. Hence the height difference between the equator
(cos ϕ = 1) and poles (cos ϕ = 0) is, using the answer to part (a),
2
1 2 2 (7.27 × 10−5 × 6 × 106 )
r1 − r0 = Ω rref = = 9698 m
2g 2 × 9.81

3
 and so our estimate of the earth’s ellipticity is (r1 − r0 ) /r0 ' (r1 − r0 ) /rref '
9698/ (6 × 106 ) ' 1. 6×10−3 . This is less than one half of the observed
value of 1/297 = 3. 4×10−3 . [We have left out a number of factors, the
most important being gravity feedback: the distortion of the gravity
field associated with the departure of the earth’s shape from a perfect
sphere is a significant factor, and intensifies the equatorial bulge beyond
what our simplified calculation predicts.]
4. A punter kicks a football a distance of 60m on a field at latitude 45◦ N.
Assuming the ball, until being caught, moves with a constant forward
velocity (horizontal component) of 15ms −1 , determine the lateral de-
flection of the ball from a straight line due to the Coriolis effect. [Neglect
friction and any wind or other aerodynamic effects.]
Neglecting all other forces, then if u is the velocity component in the
direction it is kicked, and v the component normal to this (to the right
of u), then, neglecting other forces,
∂v
= fu
∂t
where f is the Coriolis parameter at 45◦ N. Given that u = 15ms−1 is
constant,
v = fut ,
and so the displacement, to the right of the kicking direction is
1
y = fut2 .
2
Since the travel time t = L/u, where L is the distance traveled,
1 f L2
y= .
2 u
With the given numbers,
1.03 × 10−4 × 602
y= = 0.012m = 1.2cm.
2 × 15
NOTE: (1)The ratio of the lateral displacement to the distance traveled
is just
y 1 fL
= ,
L 2 u
4
 i.e., just one half of the inverse of the Rossby number based on the
kicking velocity and distance traveled. (2) We have here neglected
the action of Coriolis forces of the vertical component of the motion.
This is not really valid–the scaling argument we applied to a shallow
atmosphere does not apply to the ball’s trajectory–but, in fact, the
net effect is zero.
5. Imagine that Concord is (was) flying at speed u from New York to
London along a latitude circle. The deflecting force due to Coriolis is
toward the south. By lowering the left wing ever so slightly the pilot
(or perhaps more conveniently the computer on board) can balance this
deflection. Draw a diagram of the forces – gravity, uplift normal to
the wings and Coriolis – and use it to deduce that the angle of tilt, γ,
of the aircraft from the horizontal required to balance the Coriolis force
is
2Ω sin ϕ × u
tan γ = ,
g
where Ω is the Earth’s rotation, the latitude is ϕ and gravity is g.
If u = 600 m s−1 , insert typical numbers to compute the angle. What
analogies can you draw with atmospheric circulation? [Hint: cf Eq.(7.8).]

N MAcor
S

Mg

Consider the figure. The southward Coriolis acceleration (at latitude

ϕ = 45o N), with u = 600 ms−1 , is just Acor = 2Ω sin ϕ × u = 0.062
ms−2 , much less than gravity. The vertical lift force acting on the air-
craft required to balance gravity is just Mg, where M is the aircraft’s
mass. Tilting the aircraft at an angle γ produces a northward hori-
zontal component of this force equal to Mg tan γ, which is required to

5
 balance the Coriolis force MAcor . Hence g tan γ = Acor , or
2Ω sin ϕ × u
tan γ = .
g

With the given numbers, tan γ = 0.062/9.81 = 6.3×10−3 , or (since this

angle is small) γ = 6.3 × 10−3 rad = 0.36o . Note that this equation
for the required slope directly parallels that of pressure surfaces in
geostrophic balance with the zonal wind u: from Eq.(7.8), we have
∂z 2Ω sin ϕ
=− u,
∂y g
and the meridional slope of the pressure surfaces, γ p , is such that
tan γ p = −∂z/∂y.

6. Consider horizontal flow in circular geometry in a system rotating about

a vertical axis with a steady angular velocity Ω.
Starting from Eq.6.29 of our notes, show that the equation of motion for
the azimuthal flow in this geometry is, in the rotating frame (neglecting
friction and assuming 2-dimensional flow)
Dvθ ∂vθ ∂vθ vθ ∂vθ vθ vr 1 ∂p
+ 2Ωvr ≡ + vr + + + 2Ωvr = − (1)
Dt ∂t ∂r r ∂θ r ρr ∂θ
where (vr , vθ ) are the components of velocity in the (r, θ) = (radial,
azimuthal) directions (see figure). [Hint – write out Eq.6.29 in cylin-
drical coordinates, noting that vr = Dr
Dt
; vθ = r Dθ
Dt
and that the gradient
∂ 1 ∂
operator is ∇ = ( ∂r , r ∂θ )]

6
The horizontal component of Eq.6.29 is:
D 1
uh + ∇h p = − (2Ω × u)h .
Dt h ρ
The azimuthal component of the above is:
∂vθ 1 ∂
+ [uh · ∇uh ]bθ + p = −2 [Ω × uh ]b
∂t rρ ∂θ θ

∂ 1 ∂
where uh = (vr , vθ ). Noting that ∇ = ( ∂r , r ∂θ ) in cylindrical coordi-
nates, and being careful to note that ∇ may operate on unit vectors as
well, we have:
·µ ¶ ¸
∂ vθ ∂
[uh · ∇uh ]bθ = vr , uh
∂r r ∂θ b θ

Now uh = vr rb + vθb
θ and so:
· ¸ ·µ ¶ ´¸
∂ vθ ∂ ∂ vθ ∂ ³
(vr , )uh = vr , vr rb + vθb
θ
∂r r ∂θ b
∂r r ∂θ b
θ
θ

∂b b
Noting that ∂b
r
∂θ
=b
θ; θ
∂θ
r; ∂b
= −b r
∂r
= 0; ∂∂rθ = 0 we get:
D ∂vθ ∂vr ∂br vθ ∂vr vθ vr ∂b
r ∂vθ b
vθ = + vr rb + vr2 + rb + + vr θ+
Dt h ∂t ∂r }
| {z ∂r}
| {z |r {z
∂θ } r ∂θ ∂r
not in b
θ direction b
∂r
∂r
=0 not in b
θ direction

∂b
θ vθ2 ∂b
θ vθ ∂vθ b
vθ vr + + θ
| {z∂r} |r{z∂r} r ∂θ
∂b
θ
=0 not in b
θ direction
∂r

Thus:
D ∂vθ vθ vr ∂b
r ∂vθ vθ ∂vθ
vθ = + + vr +
Dt h ∂t r ∂θ ∂r r ∂θ
Since [(2Ω × u)h ]b = −2Ωvr , we arrive at the answer! A pretty in-
θ
volved calculation, the complications arising from the fact that ∇ acts
on the unit vector.

7
 (a) Assume that the flow is axisymmetric (i.e., all variables are in-
dependent of θ). For such flow, angular momentum (relative to
an inertial frame) is conserved. This means, since the angular
momentum per unit mass is

m = Ωr2 + vθ r , (2)

that

Dm ∂m ∂m
≡ + vr =0. (3)
Dt ∂t ∂r
Show that Eqs.(1) and (3) are mutually consistent for axisymmet-
ric flow.
If the flow is axisymmetric, then:
∂vθ ∂vθ vθ ∂vθ vθ vr 1 ∂p
+ vr + + + 2Ωvr = −
∂t ∂r r ∂θ r ρr ∂θ

reduces to:
∂vθ ∂vθ vθ vr
+ vr + + 2Ωvr = 0
∂t ∂r r
Expanding out Eq.(3), ∂m ∂t
+ vr ∂m
∂r
= 0, using the definition of m,
Eq.(2), we arrive at the above.
(b) When water flows down the drain from a basin or a bath tub, it
usually forms a vortex. It is often said that this vortex is anti-
clockwise in the northern hemisphere, and clockwise in the south-
ern hemisphere. Test this saying by doing the following.
Fill a basin or a bath tub (preferably the latter–the bigger the bet-
ter) to a depth of at least 10cm, let it stand for a minute or two,
and then let it drain. When a vortex forms1 , estimate, as well as
you can, its angular velocity, direction, and radius (floating some
small floats, such as pencil shavings, will help to see the flow).
Hence calculate the angular momentum per unit mass of the vor-
tex.
Now, suppose that, at the instant you opened the drain, there was
1
A clear vortex (with a “hollow” center) may not form. As long as there is an identifiable
swirling motion, you will be able to proceed; if not, try repeating the experiment.

8
 no motion (relative to the rotating Earth). Now if only the vertical
component of the Earth’s rotation matters, calculate the angular
momentum density due to the Earth’s rotation at the perimeter of
the bath tub or basin. [Your tub or basin will almost certainly not
be circular, but assume it is, with an effective radius R such that
the area of your tub or basin is πR2 in order to determine m.]
Suppose, in a bathtub of dimension 1.8 × 0.8 m, the flow rotation
period is found to be about 2 s (so the rotation rate is ω = 3.1
s−1 ) at a radius of r = 2 cm. The angular momentum per unit
mass of this flow is therefore ωr2 ' 3.1 × 0.02 × 0.02 ' 1.2 × 10−3
m2 s−1 . (Since ω is so much grater than the earth’s rotation Ω,
we can here neglect the contribution from the latter.) The area of
the bath is 1. 44 m2 , corresponding (if it were circular) to a radius
of R = 0.68 m. Assuming no motion in the initial state (when
the plug is removed) other than the earth’s rotation, the absolute
angular momentum per unit mass at the perimeter would be just
ΩR2 = 7.27 × 10−5 × 0.682 = 3.4 × 10−5 m2 s−1 .
(c) Since angular momentum should be conserved, then if there was
indeed no motion at the instant you pulled the plug, the maxi-
mum possible angular momentum per unit mass in the drain vortex
should be the same as that at the perimeter at the initial instant
(since that is where the angular momentum was greatest). Com-
pare your answers and comment on the importance of the Earth’s
rotation for the drain vortex, and hence comment on the validity
of the saying.
If angular momentum were conserved, and if the water in the bath
were completely motionless (relative to the earth) then the angu-
lar momentum of the flow could not exceed that value associated
with the earth’s rotation, evaluated at the perimeter (since that
is where the maximum angular momentum, with respect to the
drain, would be found). But we have seen (at least in the case
considered here) that the value at the drain far exceeds the ex-
pected value. Since frictional effects will reduce, and not enhance,
the angular momentum as the water flows in towards the drain,
the only reasonable explanation is that the water was not sta-
tionary when the plug was removed. (In fact, a value of angular
momentum equal to that observed near the drain would require

9
 an initial azimuthal flow velocity v, at the perimeter, such that
(v + ΩR) R = 1.2 × 10−3 m2 s−1 whence v ' 2 mm s−1 .)
(d) In view of your answer to (c), what are your thoughts on Perrot’s
experiment, GFD Lab VI?
Obviously, the key to executing Perrot’s experiment successfully
is to take all precautions to eliminate relative motion in the tank,
to ensure that the absolute angular momentum of the fluid is
dominated by the earth’s rotation. This means, e.g., that the net
swirl velocity v near the edge much be small compared with ΩR;
for R = 0.68 m, this requires |v| ¿ 5 × 10−5 m s−1 . Achieving
such a state is a very difficult task.

7. We specialize Eq.(6.44) to two-dimensional, inviscid ( F = 0) flow of

a homogeneous fluid of density ρref thus:
Du 1 ∂p
+ − fv = 0
Dt ρref ∂x
Dv 1 ∂p
+ + fu = 0
Dt ρref ∂y
D ∂ ∂ ∂
where Dt
= ∂t
+ u ∂x + v ∂y and the continuity equation is
∂u ∂v
+ = 0.
∂x ∂y
(a) By eliminating the pressure gradient term between the two momen-
tum equations and³ making use ´of the continuity equation, show
∂v
that the quantity ∂x − ∂u
∂y
+ f is conserved following the mo-
tion: i.e. µ ¶
D ∂v ∂u
− + f = 0.
Dt ∂x ∂y
Taking
· ¸ · ¸
∂ Dv 1 ∂p ∂ Du 1 ∂p
+ + fu − + − fv = 0
∂x Dt ρref ∂y ∂y Dt ρref ∂x
differentiating products and collecting terms in the total derivative
expressions, noting that f = f (y) and that ∂u ∂x
+ ∂v
∂y
= 0, yields
the result that we seek.

10
(b) Convince yourself that
∂v ∂u
b
z.∇ × u = −
∂x ∂y
∂v
(see Appendix A.2.2), i.e. that ∂x − ∂u
∂y
is the vertical component
of a vector quantity known as the vorticity, ∇ × u, the curl of the
velocity field.
Forming the curl of the velocity field we have:
¯ ¯
¯ x z ¯¯ µµ ¶ µ ¶ µ ¶¶
¯ b∂ y b b
∂w ∂v ∂u ∂w ∂v ∂u
¯ ∂
∇×u = ¯ ∂x ∂y ∂z ¯ =∂ ¯
− , − , −
¯ u v w ¯ ∂y ∂z ∂z ∂x ∂x ∂y

∂v
and so we see that the b z component is ∂x − ∂u
∂y
, known as the
relative vorticity.
∂v
The quantity ∂x − ∂u∂y
+ f is known as the ‘absolute’ vorticity and
is made up of ‘relative’ vorticity (due to motion relative to the
rotating planet) and ‘planetary’ vorticity, f , due to the rotation
of the planet itself.
(c) By computing the ‘circulation’ – the line integral of u about the
rectangular element in the (x, y) plane shown in Fig.6.22 – show
that:
circulation
= average normal component .
area enclosed
of vorticity

The velocity at the center of the element is u = u (x, y). Thus

evaluating the circulation by summing the contributions along the
southern, eastern, northern and western edges we obtain:
µ ¶ µ ¶
∂u δy ∂v δx
δx u − + +δy v + +
∂y 2 ∂x 2
| {z } | {z }
south east
µ ¶ µ ¶
∂u δy ∂v δx
−δx u + + − δy v −
∂y 2 ∂x 2
| {z } | {z }
north west
µ ¶
∂v ∂u
= − δxδy
∂x ∂y

11
 Figure 1: Circulation integral schematic.

Dividing through by the area of the element, δxδy, yields the

answer we seek.
Hence deduce that if the fluid element is in solid body rotation
then the average vorticity is equal to twice the angular velocity of
its rotation.
The above result implies that the vorticity of a fluid in solid body
rotation with angular velocity ω is:

2πrV 2πr (rω)

ω= = = 2ω
πr2 πr2
twice the angular velocity of rotation.
6
(d) If the tangential velocity in a hurricane varies like v = 10r m s−1
where r is the radius, calculate the average vorticity between an
inner circle of radius 300 km and an outer circle of radius 500 km.
Express your answer in units of planetary vorticity f evaluated at
20o N. What is the average vorticity within the inner circle?
The average vorticity between the inner and outer circle is exactly
zero because vr = constant is an example of irrotational motion.
6
2πr× 10 m s−1
The vorticity within the inner circle is: π×(300
r
km)2
= 2. 22 ×
o
10 s . The value of f at 20 N is 2 × 7.27 × 10 s sin 20◦ = 4.
−5 −1 −5 −1

9 × 10−5 s−1 . Thus the hurricane has an average vorticity of 0.45

times the local Coriolis parameter.

12