J. Fluid Mech. (2004), vol. 513, pp. 161170.

c
2004 Cambridge University Press
DOI: 10.1017/S0022112004000369 Printed in the United Kingdom
161
Explicit solutions for a steady
vortexwave interaction
By DARREN G. CROWDY
Department of Mathematics, Imperial College of Science, Technology and Medicine,
180 Queens Gate, London SW7 2AZ, UK
(Received 12 March 2004 and in revised form 2 June 2004)
A class of equilibria of the Euler equations is derived describing a vortex dipole
of non-zero circulation interacting with the interface between a uniform shear layer
and an irrotational region. The ow eld and the shape of the deformed shear
layer prole are given in terms of explicit mathematical formulae. Properties of the
solutions are discussed. They share many qualitative features with a new translating
quasi-geostrophic V-state solution recently found by McDonald (2004).
1. Introduction
In geophysical uid dynamics there is much interest in nding stable self-
propagating vortexwave solutions where a localized vortex structure interacts with
some interface of vorticity (or potential vorticity) associated with, for example, an
escarpment or an interface separating two shear ows of piecewise-constant vorticity.
One aim of such studies is understanding the existence of long-lived geophysical and
astrophysical phenomena such as Jupiters Great Red Spot and other well-dened
vortical structures in zonal jet ows. Bell (1990) investigated a basic model where a
point vortex interacts with a horizontal interface separating two regions of uniform
potential vorticity. Related studies are by Dunn, McDonald & Johnson (2001) and
McDonald (2002).
This paper presents a new class of equilibria of the Euler equations representing a
steady interaction between a localized vortex structure and a vortical interface. This
investigation has been motivated by the new travelling V-state solution recently found
by McDonald (2004). McDonalds solution consists of two line vortices interacting
with an interface representing a jump in potential vorticity caused by the presence
of an escarpment. Since his solution involves combinations of vortex patches and
point vortices and also turns out to have zero net circulation, McDonald intuited the
possible relevance of a new analytical construction technique for vortical equilibria
of the Euler equations originally developed in Crowdy (1999) (and subsequently
generalized in a number of directions, e.g. Crowdy 2002a, b). This approach involves
hybrid combinations of point vortices and vortex patches.
In similar spirit to McDonalds paper, the ow conguration considered here is a
basic model of a geophysical ow. To motivate the conguration, it is rst observed
that the pair of point vortices in McDonalds solution essentially lends the embedded
vortex a dipolar character possessing a non-zero net circulation. Second, it is known
(e.g. Bell 1990; McDonald 2004) that the barotropic quasi-geostrophic equations with
step topography are dynamically equivalent to the two-dimensional Euler equations
with a piecewise-constant shear velocity playing the role of topography in providing a
jump in the background vorticity distribution. Motivated by these two facts, the model
162 D. G. Crowdy
Shear layer
Stagnant flow region Vortex dipole embedded in the
shear layer (shown as white dot)
Figure 1. Schematic illustrating the vortexwave. An innite shear layer with uniform vorticity
containing a vortex dipole with non-zero net circulation sits below a stagnant (irrotational)
ow region. The conguration is assumed to be in steady equilibrium.
system to be considered here comprises a singular vortex dipole with non-zero net
circulation sitting in a shear ow and interacting with a vorticity jump separating this
shear ow region and an irrotational region. An attractive feature of the mathematical
solutions is that they are completely explicit. In order to derive such solutions, it is
necessary to assume that the upper region of irrotational ow is stagnant. While
this idealization limits the physical relevance of the solutions, it is expected that more
general solutions exist in which the ow above the interface is non-trivial. It is likely
that such generalized solutions, which will be more physical, can be found using
perturbation or numerical methods based on the class of explicit solutions derived
herein.
2. Mathematical formulation
We seek solutions for a dipolar vortex with circulation near an innite interface
separating a region of irrotational ow from a uniform shear ow. The ow is assumed
to be incompressible. As x , the interface is supposed to become at and tend to
y =0. Far from the interface, as y , the shear ow is assumed to have the form
(u, v) =(y, 0). (2.1)
The vorticity in the region of shear is taken to equal 1 everywhere. It will be
assumed that the dipolar vortex sits inside this rotational region. Let D denote the
rotational region and D the interface. Attention is restricted to a class of solutions
in which the uid above the interface is stagnant (and therefore trivially irrotational).
Figure 1 shows a schematic of the ow conguration.
In terms of the standard complex variables z = x +iy and z =x iy, the velocity
eld (u, v) we seek is of the complexied form
u iv =
_
1
2
i ( z F(z)) , z D
0, z / D
(2.2)
where F(z) must have a second-order pole at the point in space corresponding to the
vortex dipole. The function F(z), which is an analytic function of z (except for the
isolated pole), represents the irrotational component of the velocity eld. In order
that this vortex has a net circulation, the residue of F(z) at this point must not vanish.
The condition that, as y , the velocity has the form (2.1) is equivalent to the
requirement that
F(z) z +O(z
1
) as z . (2.3)
Explicit solutions for a steady vortexwave interaction 163
There is a kinematic and dynamic boundary condition to be satised on the
interface. The kinematic condition is that D must be a streamline while the dynamic
condition that the uid pressures must be continuous across the vortex jump is known
to be equivalent to continuity of velocity on D (Saman 1992). Continuity of velocity
immediately implies, using (2.2), that
z =F(z) on D. (2.4)
Equivalently, in the terminology of Crowdy (1999), dening the Schwarz function
S(z) of the curve D to be the unique function, analytic in a strip-like neighbourhood
containing D, that is equal to z everywhere on D, then
F(z) =S(z). (2.5)
It remains to satisfy the kinematic condition on the interface. However, this follows
automatically since it is equivalent to , the streamfunction associated with the
incompressible ow, being constant on the interface and
d =
z
dz +
z
d z =
1
4
( z F(z)) dz +
1
4
_
z F(z)
_
d z = 0 on D (2.6)
where the last equality follows from (2.4) and we have used u iv =2i
z
. F(z)
denotes the complex conjugate of F(z). Finally, there is an additional constraint on
the solution (dictated by the Helmholtz vortex theorems Saman 1992) that the
non-self-induced contribution to the velocity eld at the vortex dipole position must
vanish in order for the solution to be in consistent hydrodynamic equilibrium.
It follows that
(z, z) =
_

_
1
4
_
z z
_
z
S(z

) dz

_
z
S(z

) dz

_
, z D
0, z / D
(2.7)
where the conjugate function

S(z) is dened as S(z) = S( z). The functional form of
is the same as the class of solutions presented in Crowdy (1999).
3. Conformal mapping
It remains to establish whether a domain D exists having an associated Schwarz
function with the required properties. We have found that they do exist. As in Crowdy
(1999), a convenient way to construct them is to consider a conformal mapping from
a parametric -plane. Let the interior of the unit -circle map to the unbounded
domain D with the circle | | = 1 mapping to the interface. Let z( ) be the conformal
map. Since the interface is of innite extent, z( ) must have a simple pole singularity
on the unit circle. This will be taken to be at =1. The vortex dipole is assumed
to be situated on the imaginary axis and the displaced interface will be taken to be
reectionally symmetric about this axis. This means that we can expect the point
=1 to correspond to the point of the interface on the axis of symmetry and
that complex-conjugate points and

in the pre-image plane will correspond to
reectionally symmetric points in the physical plane. Thus, if z
a
corresponds to z(
a
)
then z
a
will be given by z(

a
) so that z
a
=z(
a
) =( z
a
) = z(
a
) leading to the
conclusion that
z( ) = z( ) for all . (3.1)
164 D. G. Crowdy
Introduce the class of conformal mappings given by
z( ) =iR
_
1
1
+a +b
2
+c
_
(3.2)
where, in order that the map satises (3.1), the parameters a, b, c and R are taken to
be real. It is assumed that a, b, c and R can be found such that (3.2) is a one-to-one
map from the interior of the unit -circle to the domain D. Equation (3.2) will be
shown to be the required class of conformal maps for appropriate choices of the
parameters.
Now, the Schwarz function S(z) is given by
S(z) = z =z( ) = z(

) = z(
1
) (3.3)
where we have used the fact that

=
1
on | | = 1. On use of (3.2), we obtain
S(z( )) =
iR
1

iaR

ibR

2
iRc (3.4)
or, on rearrangement,
S(z( )) =iR
_
1 +
1
1

2
c
_
. (3.5)
While (3.5) is valid on | | = 1 it is also valid o this circle by analytic continuation.
Using (3.2), (3.5) can be written as
S(z( )) =z +iR iaR
_
+
1

_
ibR
_

2
+
1

2
_
2iRc. (3.6)
where we have eliminated iR( 1)
1
. Note that as 1, z and
S(z( )) z +iR(1 2a 2b 2c) +O(( 1)). (3.7)
Combining (2.5) and (3.7) it is seen that F(z) satises the required far-eld condition
(2.3) provided that
R
_
a +b +c
1
2
_
=0, (3.8)
which constitutes our rst constraint on the parameters.
It is noted that (3.4) has a second-order pole at = 0 which maps to a point inside
D. Since z( ) is a one-to-one map this implies that S(z) also has a second-order pole
inside D. For general a, b and c, this pole of F(z) will have non-zero residue and will
correspond to the desired point vortex dipole with non-zero net circulation. Therefore,
let z
d
=z(0) be the position of this dipole in D. Near z
d
, the complex velocity eld
inside D will have the general form
u iv =

2(z z
d
)
2

i
2(z z
d
)
+V +O((z z
d
)) (3.9)
for some constants , and V. In order that the solution is a consistent steady
solution of the Euler equation, the solution must be such that V = 0 since V is the
non-self-induced velocity at the dipole. Here has an interpretation as the vortex
dipole strength while is its net circulation.
Some straightforward (but somewhat lengthy) algebraic manipulations using the
Taylor expansion of z( ) about = 0 leads to the equations

2
=
bR
3
(a 1)
2
2
,
i
2
=
iR
2
(a(a 1) + 2b(b 1))
2
(3.10)
Explicit solutions for a steady vortexwave interaction 165
while the condition that V = 0 is found to be equivalent to
b
3
2b
2
b(a
2
3a + 1) +(a 1) =0. (3.11)
Equation (3.11) is a second constraint on the conformal mapping parameters.
We choose to normalize the class of solutions so that the vortex dipole is situated
at z =i. This corresponds to
z
d
=i =iR(c 1) (3.12)
which is a third constraint on R, a, b and c. To impose that the interface tends to
y =0 as x note that, as 1,
z z
M
( ) +O( 1). (3.13)
where
z
M
( ) =
iR
1
+iR(a +b +c). (3.14)
is a Mobius map. The condition that z
M
( ) takes | | =1 to the line y =0 is found to
be the same as condition (3.8).
With a as parameter, (3.11) gives a cubic polynomial equation for b. Once b is
determined, (3.8) and (3.12) provide equations for c and R. Since the equation for b
in terms of a is a cubic polynomial, the solution can be written down, leading to a
completely explicit class of solutions with parameters b = b(a), c = c(a) and R = R(a)
available as known functions of a. Indeed, by completing the cube and making an
identication with the trigonometric identity cos 3 =4 cos
3
3 cos , some algebra
leads to the following formula for b:
b(a) =
2
3
+(a) cos
_
1
3
cos
1
(4(a)/(a)
3
)
_
(3.15)
where
(a) =2a
2
/3 +3a 61/27, (a) =2(a
2
3a +7/3)
1/2
/

3. (3.16)
The three possible solutions of the cubic (3.11) are encapsulated in the multivaluedness
of the inverse cosine function. Of the three generally distinct solutions, only one has
been found to yield a one-to-one conformal mapping and is therefore the only one
that is physically admissible. Simple algebraic manipulations then lead to
c(a) =1/2 a b(a), R(a) =(1/2 + a + b(a))
1
, (3.17)
while the vortex dipole strength and circulation are
(a) =R
3
(a)b(a)(a 1)
2
, (a) =R
2
(a)[a(a 1) + 2b(a)(b(a) 1)]. (3.18)
The (complex) velocity eld in the shear layer is also available as an explicit function
of ,

:
u iv =
R(a)
2
_
1

1
+a

+b(a)

b(a)

2
_
. (3.19)
Finally, we remark that (3.2) can be rearranged to form a cubic equation for with
coecients depending on z and the conformal mapping parameters. The solution of
this cubic can be written down (e.g. using Cardans formula) leading to an explicit
formula for the inverse map (z). Thus, the complex velocity eld (3.19) can, in
principle, be written down explicitly in terms of z and z; however the resulting
expression is complicated. The conformal mapping approach provides a much more
convenient parametrization of these non-trivial solutions.
166 D. G. Crowdy
0
1
2
0
1
2
6 4 2 0 2 4 6
0
1
2
6 4 2 0 2 4 6
0
1
2
6 4 2 0 2 4 6
6 4 2 0 2 4 6
(a)
(b)
(c)
(d)
Figure 2. Equilibrium solutions for a = a
crit
=0.9632 and a =1.5, 3, 10 ((a)(d)).
A black dot shows the position of the vortex dipole.
4. Characterization of the solutions
Solutions have been found to exist for all values of a in the interval
a (, a
crit
] (4.1)
where a
crit
=0.9632. Figure 2 shows the four representative interface proles for
a =0.9632, 1.5, 3 and 10. The two limiting states a = a
crit
and a are
of interest. When a = a
crit
=0.9632 an inward-pointing cusp develops on the line
of symmetry leading to a double hump prole. Another allowable singularity is a
90

corner (see Overman 1986) and such limiting solutions on uniform vortex layers
have been found by Broadbent & Moore (1985) in the case of uniformly travelling
waves on a layer of nite depth. The solution here appears to be the rst example
of a cuspidal limiting state on a uniform vortex layer. It corresponds to a zero of
z

meeting the unit circle at =1. The critical parameter a

crit
=0.9632 is the
simultaneous solution of (3.11) and z

(1) =0, or
a 2b(a) 1/4 = 0. (4.2)
As a gradually decreases, the prole becomes smoother and develops a single hump
prole. The horizontal extent of the prole is greatest when a is close to a
crit
. As
a the edges of the hump draw inwards so that the prole begins to adopt a
near-circular shape. Figure 3 shows the prole in the case a =300 with the circle
|z i| =1 superposed for comparison. The interface develops two points of very high
curvature where the near-circular vortex sits on the near-at shear layer.
Figure 4 shows graphs of the dipole strength and the circulation as functions
of a. Note rst that is real so the dipole is always aligned with the horizontal
axis. The circulation of the dipole is greatest in the critical double-hump state
Explicit solutions for a steady vortexwave interaction 167
Figure 3. Solution for a =300. The circle |z i| =1 is superposed (dotted line)
for comparison.
20 15 10 5 0
6
5
4
3
2
1
0
a

20 15 10 5 0
3
4
5
6
7
8
9
10
11
a

Figure 4. Graphs of dipole strength (a) and circulation (a) against a.

with a = a
crit
. It is also in this state that the dipole strength has its greatest
magnitude. On the other hand, as a , while 0. Since the prole in
this limit tends to a near-circle with unit radius, the vortical conguration moves
close to a situation where an area- Rankine vortex of uniform vorticity 1 with
a superposed point vortex at its centre of equal and opposite circulation sits
on top of a near-at shear layer. Such a limiting conguration is consistent with
ideas presented in Crowdy (2002b) concerning the possibility of constructing vortical
equilibria with complicated geometry by superposing, and merging, shielded Rankine
vortices. Indeed, the solutions here are smoothed-out equilibria where a shielded
Rankine vortex merges with a uniform shear layer.
168 D. G. Crowdy
Figure 5. Streamlines for the solutions with a =1.5 and a =3. Note the presence of a
large counter-rotating recirculation region below the vortex dipole.
McDonald (2004) nds a solution in which a pair of point vortices is situated
below the height of an undisturbed vortical interface. In the solutions above, the
vortex dipole is always above the far-eld interface height and we have not found
any solutions for vortex dipoles situated below it. However, the streamline patterns
shown in gure 5 for a =1.5 and 3 display interesting features. Most noticeable
is the presence of a large counter-rotating recirculation region below the dipole. It is
straightforward to show that there always exists a stagnation point on the imaginary
axis just below the dipole. For a given a, the pre-image of such a stagnation point is
a real solution for between 1 of the nonlinear equation
1
1
+a +b(a)
2
=

1
+
a

+
b(a)

2
. (4.3)
The vertical position, as a function of a, of the stagnation point below the dipole is
plotted in gure 6; it is clear that while the vortex singularity is above the far-eld
interface height, the centre of the region of uid recirculating against the shear is
always below it.
McDonald (2004) also nds that his equilibrium solution has zero net circulation.
Consider the net circulation associated with the uniform vorticity of magnitude 1
which has crossed the undisturbed prole y =0 into y >0. The area that has crossed
y =0 is given by the formula

1
2i
_
| |=1
z(
1
)z

( ) d
1
2i
_
| |=1
z
M
(
1
)z
M
( ) d

. (4.4)
An exercise in residue calculus shows that the area (4.4) is R(a)
2
(a(a 1) +
2b(a)(b(a) 1)) which, by (3.18), equals (a). Hence, like the solution of McDonald
(2004), the exact solutions share the property of having zero net circulation for all
values of a.
Explicit solutions for a steady vortexwave interaction 169
30 25 20 15 10 5 0
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
P
o
s
i
t
i
o
n

o
f

l
o
w
e
r

s
t
a
g
n
a
t
i
o
n

p
o
i
n
t
a
Figure 6. Height of the lower stagnation point (i.e. the centre of the lower recirculation
region) as a function of a. It is always situated below the far-eld height of the interface
(which is at y =0).
5. Discussion
Signicantly, there is no limit of the above solutions in which the interface tends
to the linear at state as the dipole strength and circulation vanish. The two limiting
states are highly nonlinear structures, as are all the intermediate solutions. As such, we
expect that they might be important in providing non-trivial initial states from which
a numerical continuation procedure can be used to construct steadily translating
equilibria not necessarily continuously connected to the at state. Since they have a
dipolar character, there may well exist continuations of these equilibria that are
steadily translating. This is based on experience with the class of non-rotating
multipolar equilibria derived in Crowdy (2002a) where it is found using numerical
simulations that, when slightly perturbed, stable rotating structures can result. It
may also be possible to use perturbation theory about the exact solutions to nd a
semi-analytic representation for generalized vortexwaves that have small non-zero
translational velocities.
Modied solutions in which the point-vortex dipole is replaced by a point-vortex
pair or desingularized to a uniform vortex patch are reasonable candidates for
generalized equilibria. The exact solutions are likely to provide good initial estimates
for the circulations of these modied models. There is also no nite value of a for
which either or becomes zero: both the circulation and dipolar character of the
vortex are crucial for equilibrium. As mentioned earlier, in the solutions found here
the dipole is always above the mean far-eld height of the vortical interface. In the
numerical algorithm of McDonald (2004), the two point vortices are constrained to
be below the far-eld interface height. The present results suggest that McDonalds
solution class may well be generalizable if this constraint is relaxed.
Perhaps the most unphysical aspect of the solutions is the fact, observable in
gure 5, that the streamlines intersect the vortical interface. This is a result of
the assumption that the upper uid is stagnant so that the vortical interface is a
stagnation line where the local direction of the ow is not uniquely determined. We
expect, however, that there exist generalized equilibria in which both the interface
170 D. G. Crowdy
shape and dipole strength and circulation are close to those of the exact solutions
but where the motion of the upper uid is non-trivial leading to a more physically
realistic ow. Perturbation theory or numerical methods might be used to uncover
such generalized solutions.
The solutions here share many qualitative features with a class of wave solutions on
a layer of constant vorticity computed numerically in Vanden-Broeck (1994, 1995). He
nds steadily translating solutions whose limiting forms consist of arbitrary numbers
of circular Rankine vortices sitting on top of a region of uniform shear. This is
reminiscent of the limiting solution here where a shielded Rankine vortex sits on
top of a at shear layer. This strongly suggests that there might exist a class of
intermediate, steadily translating, vortex-wave solutions (not necessarily describable
in exact mathematical form) where the circulation of the dipole diers from that of
the above solutions. Vanden-Broeck (1995) also remarks on the fact that his solutions
do not bifurcate from the uniform shear layer solution.
It should be possible to study the linear stability of the solutions as a function
of a using complex-variable methods based on perturbed conformal maps as in
Crowdy (2002a). The robustness of the structures to nonlinear perturbations could
be examined using modied contour dynamics/surgery codes (Dritschel 1988).
REFERENCES
Bell, G. I. 1990 Interaction between vortices and waves in a simple model of geophysical ow.
Phys. Fluids A 2, 575586.
Broadbent, E. G. & Moore, D. W. 1985 Waves of extreme form on a layer of uniform vorticity.
Phys. Fluids 28, 15611562.
Crowdy, D. G. 1999 A class of exact multipolar vortices. Phys. Fluids 11, 25562564.
Crowdy, D. G. 2002a Exact solutions for rotating vortex arrays with nite-area cores. J. Fluid
Mech. 469, 209235.
Crowdy, D. G. 2002b The construction of exact multipolar equilibria of the two-dimensional Euler
equations. Phys. Fluids 14, 257267.
Dritschel, D. G. 1988 Contour surgery: a topological reconnection scheme for extended
integrations using contour dynamics. J. Comput. Phys. 77, 240266.
Dunn, D. C., McDonald, N. R. & Johnson, E. R. 2001 The motion of a singular vortex near an
escarpment. J. Fluid Mech. 448, 335365.
McDonald, N. R. 2002 Steady, nonradiating geophysical ow past a cylinder with circulation.
Phys. Fluids 14, 30183027.
McDonald, N. R. 2004 A new translating quasigeostrophic V-state. Eur. J. Mech. B/Fluids
23, 633644.
Overman, E. A. 1986 Steady-state solutions of the Euler equations in two dimensions. SIAM
J. Appl. Maths 46, 765800.
Saffman, P. G. 1992 Vortex Dynamics. Cambridge University Press.
Vanden-Broeck, J.-M. 1994 Steep solitary waves in water of nite depth with constant vorticity.
J. Fluid Mech. 274, 339348.
Vanden-Broeck, J.-M. 1995 New families of steep solitary waves in water of nite depth with
constant vorticity. Eur. J. Mech. B/Fluids 14, 761774.