ESOE 5136 2021 February 22 - Handout Lo

質量 守 恆 )
1. Conservation of mass (質
Mass cannot be created or destroyed. For water, we can assume it is incompressible,
so that its density is constant. Thus, “flow in” must equal “flow out”. Based on these
concepts, we can derive the “continuity equation” (連續方程), or the “mass equation” (質
量方程):
∂u ∂v ∂w
∇ · ~u = + + = 0, (1)
∂x ∂y ∂z
where u is the x−component of flow velocity, v is the y−component of flow velocity, and
w is the z−component of flow velocity.

速度 位 勢 )
2. Velocity potential (速
If the velocity, ~u = hu, v, wi, of a flow satisfies ∇×~u = 0, the flow is said to be “irrotational”
(非旋流), and a velocity potential, φ, can be defined, so that
∂φ ∂φ ∂φ
= u, = v, = w. (2)
∂x ∂y ∂z

拉普 拉 斯 方 程 )
3. Laplace equation (拉
If we consider an irrotational flow, so that a velocity potential exists: ~u(x, y, z, t) =
∇φ(x, y, z, t), the continuity equation (1) then becomes

∂2φ ∂2φ ∂2φ

∇ · ~u = ∇ · ∇φ = ∇2 φ = + 2 + 2 = 0, (3)
∂x2 ∂y ∂z
which is the Laplace equation.

全導 數 )
4. Total derivative (全
The total derivative,
also known as the material derivative (實質導數), of a function
f x(t), y(t), z(t), t is defined as
df Df ∂f ∂f dx ∂f dy ∂f dz ∂f ∂f ∂f ∂f
= = + + + = +u +v +w . (4)
dt Dt ∂t ∂x dt ∂y dt ∂z dt ∂t ∂x ∂y ∂z
The relation (u, v, w) = (dx/dt, dy/dt, dz/dt) is used to simplify the expression in Equation
(4).
The total derivative is used to calculate the rate
of change of f along a path, whose
location is described by (x, y, z) = x(t), y(t), z(t) . In fluid mechanics, we need to do
a lot of calculations for the water particles. Since water particles move along the flow,
we have to track them. To calculate the rate of change of something experienced by the
moving water particles, we have to use total derivatives.

動量 守 恆 )
5. Conservation of momentum (動
Sum of the forces equals to mass times acceleration, i.e., “F = M a”. To study water
waves (away from the boundary layers), we can assume it is inviscid; in other words, we
assume the viscosity of water to be zero. As a result, the shear stresses within the fluid are
zero. Based on these concepts, we can derive the “momentum equations”, or the “Euler
equations”:
Du 1 ∂p Dv 1 ∂p Dw 1 ∂p
=− , =− , =− − g, (5)
Dt ρ ∂x Dt ρ ∂y Dt ρ ∂z
where ρ is the density of the fluid (assumed to be constant; for water, generally ρ = 998
kg/m3 ), g is the gravitational acceleration (generally g = 9.81 m/s2 ) and p is the pressure
within the fluid.

1
ESOE 5136 2021 February 22 - Handout Lo

6. No-flow (or “no-penetration”) boundary condition

On a fixed wall, there should be no flow going across (perpendicularly to) the interface.
Letting n̂ denote the unit normal vector perpendicular to the wall, we write this require-
ment as
~u · n̂ = 0. (6)
7. Dynamic free surface boundary condition (DFSBC, 動 力自 由液面 邊界條 件)
The forces acting on the fluid at the free surface should be in equilibrium. Otherwise, the
free surface wouldn’t be there; it would move to somewhere else where the forces are in
equilibrium.
Put differently, the DFSBC can also be seen as a requirement for the pressure just above
the free surface to be the same as the pressure just under the free surface – to ensure the
continuity of pressure across the free surface.
In our idealized water wave problem, under the assumptions that water is incompressible
and the flow is irrotational, the Bernoulli equation,
∂φ 1 2 p
+ (u + w2 ) + + gz = C(t),
∂t 2 ρ
where C(t) is an arbitrary function that results from integration in time, can be used to
acquire the fluid pressure:
 ∂φ 1 
p = −ρ + (u2 + w2 ) + gz + ρC(t).
∂t 2
On the free surface z = η(x, t), the fluid pressure p should equal the atmospheric pressure
patm :
 ∂φ 1 
p = −ρ + (u2 + w2 ) + gz + ρC(t) = patm .
∂t 2
If we choose a convenient C(t) so that ρC(t) = patm , the DFSBC simplifies to
∂φ 1 2
+ (u + w2 ) + gz = 0, on the free surface z = η(x, t), (7)
∂t 2
For small-amplitude waves with no winds or other forcing mechanisms acting on the free
surface, this boundary condition can be further linearized to be
1 ∂φ
η=− , on the still water surface z = 0. (8)
g ∂t

8. Kinematic free surface boundary condition (KFSBC, 運 動自由 液面邊 界條件 )

The free surface can move in time, but there should be no flow that goes across (penetrates)
the free surface. Mathematically, since the free surface is located at z = η(x, t), we can
define the free surface to be F (x, z, t) = z − η(x, t) = 0. Imagine a point on the free
surface defined so that F (x(t), z(t), t) = 0. It should move with the flow and always stay
on the free surface. It does not fall under or fly over the free surface. Therefore, its total
derivative is required to be zero:
DF (x(t), z(t), t) ∂F ∂F ∂F ∂η ∂η
=0= +u +w =− −u +w
Dt ∂t ∂x ∂z ∂t ∂x . (9)
∂η ∂η
⇒ w= + u , on the free surface z = η(x, t)
∂t ∂x
For small-amplitude waves, this boundary condition can be further linearized to be
∂η
w= , on the still water surface z = 0. (10)
∂t

2
ESOE 5136 2021 February 22 - Handout Lo

9. Periodic boundary condition

If the waves are periodic in space and time, we know the solutions will just repeat in space
for every wavelength L, and in time for every wave period T . Thus, we can impose the
periodic boundary condition:
φ(x, z, t) = φ(x + L, z, t), φ(x, z, t) = φ(x, z, t + T ). (11)

10. Solution of the Laplace equation for small-amplitude periodic waves

控制 方 程 )
Governing equation (控
For an incompressible and irrotational flow, the Laplace equation
∂2φ ∂2φ
∇2 φ = + 2 =0 (12)
∂x2 ∂z
governs the 2D water flow under a wave.
Boundary conditions
We shall derive the solutions for small-amplitude periodic water waves in a constant water
depth of h. The problem setup is sketched in Figure 1. We seek the solutions for a periodic
wave of the form
η(x, t) = A cos(kx − ωt), (13)
where A is the wave amplitude, k is the wavenumber, and ω is the angular frequency.

Figure 1: Sketch of the problem setup for small-amplitude periodic water waves in constant
water depth, modified from Dean & Dalrymple (1991).

Since this problem is periodic in the x-direction, the “periodic boundary condition” should
be applied on boundary 1 and boundary 3. The “no-flow (no-penetration) boundary
condition” should be applied on boundary 2 because it is a solid wall that water cannot
go through. Boundary 4 is the free surface, so the linearized KFSBC and the linearized
DFSBC should be applied here.
Equations
To summarize, the governing equation is the Laplace equation:
φxx + φzz , for 0 ≤ x ≤ L, −h ≤ z ≤ 0. (14)
The boundary conditions are:


 periodic BC: φ(x + L) = φ(x) at x = 0 and x = L
no-flow BC: w=0 at z = −h

. (15)

 LKFSBC: w = ηt at z=0
LDFSBC: η = −φt /g at z=0


3
ESOE 5136 2021 February 22 - Handout Lo

General solution
We seek the solution of the form

φ(x, z, t) = X(x)Z(z)T (t)

so that the Laplace equation can be written as

X 00 Z 00
φxx + φzz = 0 ⇒ X 00 ZT + XZ 00 T = 0 ⇒ =− = −λ,
X Z
where λ is some constant to be determined.
We then have two ODEs:
 00 √
X + λX = 0 ⇒ r2 = −λ, r =√± −λ
.
Z 00 − λZ = 0 ⇒ r2 = λ, r = ± λ

• If λ > 0,
√ √ √ √
X = c1 cos( λx) + c2 sin( λx), Z = c3 e λz
+ c4 e− λz
.

• If λ = 0,
X = c1 + c2 x, Z = c3 + c4 z.
• If λ < 0,
√
−λx
√ √ √
X = c1 e + c2 e− −λx
, Z = c3 cos( −λz) + c4 sin( −λz).

Since we are looking for waves periodic in the x-direction, we immediately see that only
the case with λ > 0 is possible!
The waves should repeat themselves for every wavelength, i.e.,
√ √ √  √ 
X(x+L) = X(x) ⇒ c1 cos( λx)+c2 sin( λx) = c1 cos λ(x+L) +c2 sin λ(x+L) .
(16)
The trigonometric identities (三角恆等式) state that
 √ √ √ √ √ √
cos(√ λx + √ λL) = cos(√ λx) cos(√ λL) − sin(√ λx) sin(√ λL)
, (17)
sin( λx + λL) = sin( λx) cos( λL) + cos( λx) sin( λL)

so Equation (16) can be written as

√ √ √ √ √ √
c1 cos( λx) + c2 sin( λx) = c1 cos(√ λx) cos(√ λL) − c1 sin(√ λx) sin(√ λL)
+ch2 sin( λx) cos( λL) + c2 cos(
i λx) sin( λL)
√ √ √ ,
= c1 cos( λL) + c2 sin( λL) cos( λx)
h √ √ i √
+ − c1 sin( λL) + c2 cos( λL) sin( λx)

from which we have

 √ √ √ √
c1 = c1 cos( √λL) + c2 sin( √λL) ⇒ cos( √
λL) = 1, sin( √λL) = 0
.
c2 = −c1 sin( λL) + c2 cos( λL) ⇒ cos( λL) = 1, sin( λL) = 0
Since λ > 0, we then see that
√ √ 2nπ
λL = 2nπ ⇒ λ= = nk, n = 1, 2, · · · ,
L

4
ESOE 5136 2021 February 22 - Handout Lo

where k = 2π/L is the wavenumber. We are to derive the solutions for a wave of the form

H
η(x, t) = cos(kx − ωt).
2
To examine the behavior of a periodic water wave of
√ this form with a known wavenumber
k, we need to focus only on the case n = 1, where λ = k.
The X equation is therefore

X(x) = c1 cos(kx) + c2 sin(kx). (18)

Next, we consider the Z equation, which has the solution form

√ √
Z = c3 e λz
+ c4 e− λz
.

The no-flow boundary condition at the bottom boundary requires that

w = φz = 0 = XZ 0 T at z = −h ⇒ Z 0 (−h) = 0.

Calculate

Z 0 (z) = c3 kekz − c4 ke−kz ⇒ Z 0 (−h) = c3 ke−kh − c4 kekh = 0 ⇒ c4 = c3 e−2kh ,

so
 
Z(z) = c3 ekz + c3 e−2kh e−kz = c3 e−kh (ekh+kz + e−kh−kz ) = 2c3 e−kh cosh k(h + z) .
(19)
Since our solution should be periodic in time with an angular frequency of ω, we assume
the solution form to be
T (t) = c5 cos(ωt) + c6 sin(ωt). (20)

We can now assemble the complete solution as

φ(x, hz, t) = X(x)Z(z)T (t) =ih

ih i
c1 cos(kx) + c2 sin(kx) 2c3 e−kh cosh k(h + z) c5 cos(ωt) + c6 sin(ωt)
h i .
⇒ K1 cos(kx) cos(ωt) + K2 sin(kx) sin(ωt) + K3 cos(kx) sin(ωt) + K4 sin(kx) cos(ωt)


cosh k(h + z)

The trigonometric identities, (17), can again be used to rewrite the solution as
h i

φ = C1 sin(kx − ωt) + C2 cos(kx − ωt) + C3 sin(kx + ωt) + C4 cos(kx + ωt) cosh k(h + z) .
(21)
We see that the general solution consists of all possible sine and cosine waves with the
wavenumber k and the angular frequency ω: a sine wave and a cosine wave going to the
right, and a sine wave and a cosine wave going to the left. In linear theory, we can add
any of these waves together to obtain a wave of an arbitrary shape!
Specific solution
Since we wish to examine the wave of the form

η(x, t) = A cos(kx − ωt), (22)

5
ESOE 5136 2021 February 22 - Handout Lo

we will further simplify the solution obtained in (21).

The linearized DFSBC requires that
1
η = − φt , at z = 0. (23)
g
We calculate φt to be

h i
φt = ω cosh k(h+z) −CA cos(kx−ωt)+CB sin(kx−ωt)+CC cos(kx+ωt)−CD sin(kx+ωt) ,
(24)
so
h i
φt (z = 0) = ω cosh(kh) −CA cos(kx−ωt)+CB sin(kx−ωt)+CC cos(kx+ωt)−CD sin(kx+ωt) .
(25)
The DFSBC gives
1 1 h i
η = − φt |z=0 = A cos(kx − ωt) = − ω cosh(kh) − CA cos(kx − ωt) + · · · . (26)
g g
We see that the constants
gA
CB = CC = CD = 0, CA = . (27)
ω cosh(kh)

For the small-amplitude periodic water wave of the form

η(x, t) = A cos(kx − ωt), (28)
we found the velocity potential to be
 
gA cosh k(h + z)
φ(x, z, t) = sin(kx − ωt). (29)
ω cosh(kh)

We recall that the flow velocity under this wave can be calculated as
φx = u, φz = w. (30)

Dispersion relation
Lastly, we shall apply the linearized KFSBC:
w = φz = ηt , at z = 0.
We calculate   

 gkH sinh k(h + z)
φz = sin(kx − ωt)

2ω cosh(kh) .
 ωH
sin(kx − ωt)

 η =
t
2
The linearized KFSBC then requires
gkH ωH
φz |z=0 = ηt = tanh(kh) sin(kx − ωt) = sin(kx − ωt).
2ω 2
We then see that
gkH ωH
tanh(kh) = ⇒ ω 2 = gk tanh(kh),
2ω 2
which is the dispersion relation for small-amplitude water waves.

6
ESOE 5136 2021 February 22 - Handout Lo

11. Small-amplitude water wave solutions

To summarize, for a small-amplitude water wave of the form
η(x, t) = A cos(kx − ωt), (31)
we found   

 gA cosh k(h + z)
φ(x, z, t) = sin(kx − ωt)






 ω cosh(kh)



  



 cosh k(h + z)

 u(x, z, t) = φx = Aω cos(kx − ωt)
sinh(kh)






  
sinh k(h + z) . (32)
sin(kx − ωt)


 w(x, z, t) = φz = Aω



 sinh(kh)



  



 cosh k(h + z)
p(x, z, t) = −ρgz + ρgη(x, t)


cosh(kh)








 2
ω = gk tanh(kh)
Although we derived the solutions for just one specific periodic wave, similar solutions
with slight modifications can be derived for rudimentary periodic waves of different forms
– such as η2 = A cos(kx + ωt), η3 = A sin(kx − ωt), and η4 = A sin(kx + ωt).
12. Shallow water approximations
When the wavelength L of a wave is much larger than the water depth h, i.e., O(h/L)  1,
we can simplify the solutions in (32) and obtain the shallow water approximations.
To the leading order, the shallow water approximations are obtained by taking the limit
of the solutions as h/L → 0, or kh = (2π/L)h → 0. For kh → 0, the hyperbolic functions
can be approximated as
cosh(kh) ' 1, sinh(kh) ' kh, tanh(kh) ' kh. (33)

Also, since −h ≤ z ≤ 0, we have the relations

0 ≤ (h + z) ≤ h ⇒ 0 ≤ k(h + z) ≤ kh ⇒ k(h + z) → 0 when kh → 0. (34)

In the shallow water limit kh → 0, the solutions in (32) simplify to


gA
φ(x, z, t) ' sin(kx − ωt)


ω








 Aω

 u(x, z, t) ' u(x, t) = cos(kx − ωt)
kh





Aω
. (35)
 w(x, z, t) ' k(h + z) sin(kx − ωt)  1
kh







p(x, z, t) ' −ρgz + ρgη(x, t)








 ω ' k pgh ⇒ c = pgh


Some important observations of the shallow water approximations include:

7
ESOE 5136 2021 February 22 - Handout Lo

(a) The horizontal flow velocity is uniform in depth – it does not change in the vertical
direction.
(b) The vertical flow velocity is one order of magnitude smaller than the horizontal flow
velocity.
(c) The fluid pressure is quasi-hydrostatic – it depends only on the total water depth
above the reference location.
√
(d) The wave speed is simply c = gh.

13. Examples of shallow water flows

As listed in the textbook by Vreugdenhil (1994), some examples of shallow water flows
include: atmospheric flows, tidal flows, storm surge, river flows, dam break, coastal flows,
tsunamis, lake flows (seiche), and internal flows/waves.

14. Mass equation for shallow water waves

The mass equation for water waves in the shallow water limit can be derived by considering
that a net flow into a water column would result in a rise of the water column height, as
illustrated in Figure 2. By using the notation of “depth-averaged velocities”, we have the
mass equation for shallow water waves as
∂H ∂HU ∂HV
+ + = 0, (36)
∂t ∂x ∂y
where 

 H(x, y, t) = h(x, y) + η(x, y, t)



1 η

 Z

U (x, y, t) = u(x, y, z, t)dz

H −h . (37)



1 η

 Z

 V (x, y, t) = v(x, y, z, t)dz


H −h

Figure 2: If there is a net flow into the water column, the height (η) of the water column
would increase.

Alternatively, the mass equation for shallow water waves can be derived purely mathemat-
ically. The 3D mass equation with constant density is
∂u ∂v ∂w
+ + = 0, (38)
∂x ∂y ∂z

8
ESOE 5136 2021 February 22 - Handout Lo

where u(x, y, z, t), v(x, y, z, t), and w(x, y, z, t) are the flow velocities in the x-, y-, and z-
directions, respectively. Let us integrate the equation over the water depth:
Z η(x,y,t) 
∂u ∂v ∂w 
+ + dz = 0
−h(x,y) ∂x ∂y ∂z
Z η(x,y,t)  Z η(x,y,t) 
∂  ∂ 
= u(x, y, z, t) dz + v(x, y, z, t) dz + w(x, y, η, t) − w(x, y, −h, t).
−h(x,y) ∂x −h(x,y) ∂y
(39)
For the first two terms on the right hand side of (39), we need to apply the Leibniz
integral rule (積積分符 號內取 微分, 萊布尼 茨積分 法則), which states that
Z b(x) Z b(x)
d 
d
d d
f (x, z)dz = f x, b(x) · b(x) − f x, a(x) · a(x) + f (x, z)dz. (40)
dx a(x) dx dx a(x) dx

Applying (40), we can write

 Z η
∂  η
Z
∂u  ∂η ∂h

 dz = udz − u(x, y, η, t) − u(x, y, −h, t)
 −h ∂x ∂x −h ∂x ∂x


. (41)
Z η Z η

 ∂v ∂   ∂η ∂h
dz = vdz − v(x, y, η, t) − v(x, y, −h, t)


∂y ∂y ∂y ∂y

−h −h

For the third term on the right hand side of (39), we need to apply the Kinematic Free
Surface Boundary Condition:
∂η ∂η ∂η
w(x, y, η, t) = + u(x, y, η, t) + v(x, y, η, t) . (42)
∂t ∂x ∂y

For the fourth term on the right hand side of (39), we need to apply the no-flow boundary
condition on the bottom boundary. Defining the bottom boundary as a surface,

F (x, y, z) = z + h(x, y) = 0, (43)

we then write the no-flow condition as

∇F (x, y, z)
~u · n̂ = 0 = hu(x, y, −h, t), v(x, y, −h, t), w(x, y, −h, t)i ·
|∇F (x, y, z)|

1 ∂h ∂h
= hu(x, y, −h, t), v(x, y, −h, t), w(x, y, −h, t)i · h , , 1i
|∇F (x, y, z)| ∂x ∂y
. (44)
1  ∂h ∂h 
= u(x, y, −h, t) + v(x, y, −h, t) + w(x, y, −h, t) = 0
|∇F (x, y, z)| ∂x ∂y

∂h ∂h
⇒ w(x, y, −h, t) = −u(x, y, −h, t) − v(x, y, −h, t)
∂x ∂y

Substituting the expressions in (41), (42), and (44) into the integrated mass equation (39),
we obtain
∂  η ∂  η
Z Z
∂η  
+ udz + vdz = 0 (45)
∂t ∂x −h ∂y −h

9
ESOE 5136 2021 February 22 - Handout Lo

Noting that
∂H ∂h(x, y) ∂η(x, y, t) ∂η
= + = (46)
∂t ∂t ∂t ∂t
and Z η Z η
udz = HU, vdz = HV, (47)
−h −h
we can further simplify the mass equation for shallow water waves to
∂H ∂HU ∂HV
+ + , (48)
∂t ∂x ∂y
which is the same as (36).
15. Momentum equations for shallow water waves
For an inviscid and incompressible flow with constant density, the 3D momentum equations
are the Euler equations:
Du 1 ∂p Dv 1 ∂p Dw 1 ∂p
=− , =− , =− − g. (49)
Dt ρ ∂x Dt ρ ∂y Dt ρ ∂z

In the shallow water limit, the fluid pressure is hydrostatic; i.e.,

p(x, y, z, t) = ρg(η − z). (50)

With the hydrostatic pressure assumption, the momentum equation in the x-direction then
simplifies to
Du ∂u ∂u ∂u ∂u ∂η
= +u +v +w = −g . (51)
Dt ∂t ∂x ∂y ∂z ∂x
We will use a trick here to simplify the expression. Noting that the mass equation requires
∂u ∂v ∂w
u( + + ) = 0, (52)
∂x ∂y ∂z
we write
∂u ∂u ∂u ∂u ∂u ∂u ∂u ∂u ∂u ∂v ∂w
+u +v +w = +u +v +w + (u +u +u )
∂t ∂x ∂y ∂z ∂t ∂x ∂y ∂z ∂x ∂y ∂z
. (53)
∂u ∂u2 ∂uv ∂uw ∂η
= + + + = −g
∂t ∂x ∂y ∂z ∂x
Integrating (53) over the water depth, we obtain
Z η Z η Z η
∂u ∂u2 ∂uv ∂η
dz + dz + dz + (uw) − (uw) = −gH . (54)
−h ∂t −h ∂x −h ∂y z=η z=−h ∂x

The Leibniz integral rule can again be used to find

 Z η
∂h η
Z
∂u i ∂η ∂h

 dz = udz −u −u
−h ∂t ∂t −h z=η ∂t z=−h ∂t






 Z η
∂u2 ∂ h η 2 i
 Z
 ∂η ∂h
u dz − u2 − u2

dz =

∂x ∂x −h z=η ∂x z=−h ∂x . (55)
 −h


 η ∂uv ∂ h η
 Z Z
 i ∂η ∂h
dz = uvdz − (uv) − (uv)


∂y ∂y −h z=η ∂y z=−h ∂y


 −h



10
ESOE 5136 2021 February 22 - Handout Lo

Since h = h(x, y), ∂h/∂t = 0. In addition, the no-flow bottom boundary condition (44)
requires
∂h ∂h
w +u +v = 0. (56)
z=−h z=−h ∂x z=−h ∂y
The KFSBC requires
∂η ∂η ∂η
w − −u −v = 0. (57)
z=η ∂t z=η ∂x z=η ∂y

These conditions allow us to cancel many terms in (55), and (54) becomes
∂ h η 2 i ∂ h η
Z Z
∂HU i ∂η
+ u dz + uvdz = −gH . (58)
∂t ∂x −h ∂y −h ∂x
Rη Rη
Without additional assumptions, the integral terms −h u2 dz and −h uvdz cannot be fur-
ther simplified. Since our goal is to solve for depth-averaged velocities, let us decompose
each velocity component (u or v) into the depth-averaged value (U or V ) and the fluctu-
ation (u0 or v 0 ):
1 η
 Z
0

 u(x, y, z, t) = U (x, y, t) + u (x, y, z, t) = u(x, y, z, t)dz + u0 (x, y, z, t)
H −h



Z η . (59)

 0 1 0
 v(x, y, z, t) = V (x, y, t) + v (x, y, z, t) = v(x, y, z, t)dz + v (x, y, z, t)


H −h
An illustration of this decomposition process is shown in Figure 3.

Figure 3: An illustration of the velocity decomposition u(x, y, z, t) = U (x, y, t) + u0 (x, y, z, t),

where U (x, y, t) is the depth-average value, and u0 (x, y, z, t) is the fluctuation.

With the decomposition (59), (58) can be written as

∂HU 2 ∂HU V ∂ h η 02 i ∂ h η 0 0 i
Z Z
∂HU ∂η
+ + + u dz + u v dz = −gH . (60)
∂t ∂x ∂y ∂x −h ∂y −h ∂x
In the shallow water limit, the horizontal flow velocities are uniform in depth.
Therefore, the fluctuation terms are zero – u0 = 0 and v 0 = 0. On the other hand,
if we assume the velocity profiles (such as a parabolic shape) and modify the equations
correspondingly (including the fluid pressure distribution), we obtain the general “long-
wave equations”, which may allow weak depth-variation of the horizontal flow velocities.
A similar procedure can be performed on the y-momentum equation to obtain
∂HV 2 ∂ h η 0 0 i ∂ h η 02 i
Z Z
∂HV ∂HU V ∂η
+ + + u v dz + v dz = −gH . (61)
∂t ∂x ∂y ∂x −h ∂y −h ∂y

11
ESOE 5136 2021 February 22 - Handout Lo

16. Nonlinear shallow water wave equations (NSWE)

Combining the mass equation and the momentum equations for shallow water waves, we
have a system of equations to be solved, commonly referred to as the nonlinear shallow
water wave equations (NSWE):

∂H ∂HU ∂HV
+ + =0


∂t ∂x ∂y







∂HU 2 ∂HU V

 ∂HU ∂η
+ + = −gH (+ other driving forces) , (62)

 ∂t ∂x ∂y ∂x



∂HV 2

∂HV ∂HU V ∂η


+ + = −gH (+ other driving forces)



∂t ∂x ∂y ∂y

where H = H(x, y, t) is the total water depth, η = η(x, y, t) is the free surface elevation,
U = U (x, y, t) is the depth-averaged flow velocity in the x-direction, and V = V (x, y, t) is
that in the y-direction.
Since the vertical dimension (z) is not resolved in these equations, shallow water equations
like (62) are commonly referred to as 2DH (two dimensional in the horizontal directions).
The objective of this course is to solve equations like these, either numerically or analyti-
cally, in 1DH and in 2DH.

17. Linear shallow water wave equations (LSWE)

NSWE (62) was derived for nonlinear waves and the water depth was allowed to vary
in space. To linearize the equations for small-amplitude waves, we assume that O(η) '
O(U ) ' O(V )  1. Keeping only the first-order terms, we have the linear shallow water
wave equations (LSWE) as

∂η ∂U h ∂V h

 + + =0
∂t ∂x ∂y







∂U ∂η

= −g , (63)


 ∂t ∂x


∂V ∂η


= −g



∂t ∂y

where h = h(x, y) is the still water depth.

In constant water depth, namely, when h is a constant that does not change in space, (63)
can be combined into
∂2η ∂2η ∂2η
C 2( 2 + 2 ) = 2 , (64)
∂x ∂y ∂t
√
where C = gh. (64) is the classical “wave equation” that arises in many other fields.

正規化 ) / nondimensionalization (無
18. Normalization (正 無因 次 化 )
To linearize the NSWE, it is common to start by normalizing (also called nondimension-
alizing) the governing equations. Firstly, we need to define the characteristic scales (特徵
尺度) in this problem. The choice of the scales is based on experience or prior knowledge
of the problem. For the water wave problem, we choose:

12
ESOE 5136 2021 February 22 - Handout Lo

wave depth: h0
wave amplitude: A0
wavelength: L
√0
wave speed: gh√
0
wave period: L0 / gh0
wave nonlinearity parameter: √= A0 /h0
flow velocity:  gh0

We then normalize (denoted with an asterisk superscript ∗ ) the variables as

x∗ = x/L0 ⇔ x = L0 · x∗
y ∗ = y/L0 ⇔ y = L0 · y ∗
z ∗ = z/h0√ ⇔ z = h0√· z ∗
U ∗ = U/√gh0 ⇔ U = √gh0 · U ∗
V ∗ = V / gh0 ⇔ V =  gh0 · V ∗
η ∗ = η/A0 ⇔ η = A0 · η ∗
h∗ = h/h0 ⇔ h = h0 · h∗

It is hoped that with the chosen characteristic scales, each normalized variable is of the
order one, O(1).
Take the mass equation in (62) for example and write H as H = η + h:
∂η ∂ηU ∂hU ∂ηV ∂hV
+ + + + = 0. (65)
∂t ∂x ∂x ∂y ∂y
Using the normalization factors listed in the tables above, we get
hA p i ∂η ∗ h A p i ∂η ∗ U ∗ h A p i ∂h∗ U ∗
0 0 0
gh0 +  gh0 + gh 0
L0 ∂t∗ L0 ∂x∗ L0 ∂x∗
. (66)
hA p i ∂η ∗ V ∗ h A p i ∂h∗ V ∗
0 0
+  gh0 + gh0 =0
L0 ∂y ∗ L0 ∂y ∗
h √ i
Dividing by the common factor (A0 /L0 ) gh0 results in

∂η ∗ ∂η ∗ U ∗ ∂h∗ U ∗ ∂η ∗ V ∗ ∂h∗ V ∗
+  + +  + = 0. (67)
∂t∗ ∂x∗ ∂x∗ ∂y ∗ ∂y ∗

For small-amplitude water waves, the nonlinearity parameter  = A0 /h0 is small. Neglect-
ing higher-order, O(), terms in (67) yields
∂η ∗ ∂h∗ U ∗ ∂h∗ V ∗
+ + = 0. (68)
∂t∗ ∂x∗ ∂y ∗

Or, dimensionally (有因次的),

∂η ∂hU ∂hV
+ + = 0, (69)
∂t ∂x ∂y
which is the linearized mass equation for shallow water waves.
Similarly, for the x-momentum equation in (62), we first use H = η + h to write

∂ηU ∂hU ∂ηU 2 ∂hU 2 ∂ηU V ∂hU V ∂η ∂η

+ + + + + = −gη − gh . (70)
∂t ∂t ∂x ∂x ∂y ∂y ∂x ∂x

13
ESOE 5136 2021 February 22 - Handout Lo

Then, we normalize the equation as

hA i ∂η ∗ U ∗ hA i ∂h∗ U ∗ hA i ∂η ∗ U ∗2
0 0 0 2
gh0 + gh0 +  gh0
L0 ∂t∗ L0 ∂t∗ L0 ∂x∗
hA i ∂h∗ U ∗2 h A i ∂η ∗ U ∗ V ∗ h A i ∂h∗ U ∗ V ∗
0 0 2 0 . (71)
+ gh0 +  gh0 + gh0
L0 ∂x∗ L0 ∂y ∗ L0 ∂y ∗
hA i ∂η ∗ h A i ∂η ∗
0 0
=− gh0 η ∗ ∗ − gh0 h∗ ∗
L0 ∂x L0 ∂x
h i
Again, dividing by the common factor (A0 /L0 )gh0 results in

2 2
∂η ∗ U ∗ ∂h∗ U ∗ ∗ ∗
2 ∂η U ∂h∗ U ∗ ∗ ∗ ∗
2 ∂η U V ∂h∗ U ∗ V ∗ ∗ ∂η
∗
∗ ∂η
∗
 + +  +  +  +  = −η − h .
∂t∗ ∂t∗ ∂x∗ ∂x∗ ∂y ∗ ∂y ∗ ∂x∗ ∂x∗
(72)
Retaining only the first-order, O(1), terms gives

∂h∗ U ∗ ∗ ∂η
∗
= −h , (73)
∂t∗ ∂x∗
Or, dimensionally,
∂hU ∂η ∂U ∂η
= −gh ⇒ = −g , (74)
∂t ∂x ∂t ∂x
since h(x, y) does not vary in time. This is the linearized mass x-momentum equation for
shallow water waves.
The y-momentum equation for shallow water waves can be linearized in the same manner
to be
∂V ∂η
= −g . (75)
∂t ∂y
We have thus derived the linearized system of equations, LSWE, as given in (63).

14