Sie sind auf Seite 1von 29

60th Anniversary State-of-the-Art Reviews

Boussinesq Models and Their Application to Coastal


Processes across a Wide Range of Scales
Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

James T. Kirby, M.ASCE1


Abstract: In this paper, the development of a class of depth-integrated, phase-resolving models for surface wave propagation, known as
Boussinesq-type models (BTMs), is reviewed. This review concentrates on the extension of the leading order formulation for weakly dispersive waves to include a range of physical effects and considers model applications at a range of scales ranging from surf zone processes to
ocean basinscale tsunami propagation. A brief overview of the connection of BTMs to nonhydrostatic models, in either depth-integrated or
three-dimensional form, is included. DOI: 10.1061/(ASCE)WW.1943-5460.0000350. 2016 American Society of Civil Engineers.
Author keywords: Boussinesq equations; Nonhydrostatic models; Surf zone; Tsunamis.

Introduction
With the advent of signicantly more powerful computers in recent
decades, it has become possible to perform phase-resolving simulations of surface and internal wave motions over physical length
scales that are extensive enough to provide a comprehensive picture
of the processes under consideration. The main tool for performing
these studies in the context of relatively long waves (or waves having wavelengths that are long in comparison to the water depth) has
been a modeling technique based on the theory for weakly dispersive waves pioneered by Joseph Boussinesq in the late 1800s
(Boussinesq 1872). From their original niche as a tool for investigating the forms of surface waves propagating in shallow water, the
Boussinesq equations or their variants have been transformed into a
computational tool of great power and exibility. Although practical models began to be developed in the 1970s (Abbott et al. 1978),
application was generally limited by computer resources. This
began to change in the 1990s, and the growth in development and
application of Boussinesq-type models (or BTMs, as they are
referred to in the recent review by Brocchini 2013) has been explosive, with simulations of surface wind waves over domains with
dimensions of tens of kilometers, or of tsunami waves at global
scale, now being readily feasible. The Journal of Waterway, Port,
Coastal and Ocean Engineering has played a central role in the development of BTMs as a common tool in coastal applications.
Based on the Web of Science indexing for the journal, which
appears to start with the name of Journal of the Waterway, Port,
Coastal and Ocean Division in 1977, papers discussing
Boussinesq model development represent the top three most cited
papers in the journals history, led by the landmark paper by Nwogu
(1993) discussed later, with an additional fourth paper in the
remaining top 10.
Boussinesq equations have classically been derived using a double perturbation expansion in two parameters. One parameter m is
1

Professor, Dept. of Civil and Environmental Engineering, Center for


Applied Coastal Research, Univ. of Delaware, Newark, DE 19716.
E-mail: kirby@udel.edu
Note. This manuscript was submitted on February 23, 2016; approved
on May 2, 2016; published online on August 31, 2016. Discussion period
open until January 31, 2017; separate discussions must be submitted for
individual papers. This paper is part of the Journal of Waterway, Port,
Coastal, and Ocean Engineering, ASCE, ISSN 0733-950X.
ASCE

given by the ratio of a characteristic water depth h0 to a characteristic


horizontal length scale l 0, commonly represented by the inverse of
a wave number k0. This parameter may be understood to be a dispersion parameter, the connection being made clear by examining the
Taylor series expansion of the linear wave dispersion relationship
for kh about the shallow water limit kh = 0


1
2
3
5
2
v h=g kh tanh kh kh kh  kh kh :::
3
15
(1)
where model selection depends on the truncation of the series at a
xed order, together with optimization achieved by rearrangement
of the truncated series to achieve more accurate properties. Shallow
water theory results from the neglect of all but leading order terms,
giving a dispersion relationship

v 2 gk2 h

(2)
p
with a resulting phase speed c v =k gh that does not depend
on frequency v . The model equations that correspond to this limit
are referred to as the nonlinear shallow water equations (NLSWEs)
and describe the evolution of water depth and depth-uniform horizontal velocities in horizontal coordinates (x, y) and time t. The classic theory described by Boussinesq (1872) hinges on retention of
one further term and leads to models that are asymptotically equivalent at least to


1
2
(3)
v 2 gk2 h 1  kh
3
and that are referred to as being weakly dispersive. The development of the modern theory revolves around the effective rearrangement of the truncated series to improve accuracy (using techniques
such as Pade approximants), the extension of the series to higher
order, and the combination of both.
The second parameter d represents the ratio of a characteristic
wave amplitude a0 to depth h0, thus representing the degree of nonlinearity in the problem. Linearization results from neglecting terms
of O (d ) in comparison to leading order terms, whereas the classic
theory retains terms of O (d ) and further imposes the restriction
that d = m 2 O1. The resulting theory thus neglects terms of
Od 2 ; d m 2 ; m 4 and smaller. The development of the modern
theory often imposes no restriction on the size of d , leading to so-

03116005-1

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

called fully nonlinear Boussinesq models, in which d is simply a


placeholder in the nondimensional equations. This entire class of
models is referred to as BTMs rather than Boussinesq models, in
recognition of the fact that the initial context of Boussinesqs work
has been stretched a great deal to extend the approachs applicability to a wider range of problems.
The historical development of modern BTMs from the initial formulations of weakly nonlinear (Peregrine 1966, 1967) or fully nonlinear (Serre 1953; Green and Naghdi 1976) models for propagation
in constant depth or over simple slopes has proceeded with a range
of goals or intentions. Early reviews of these developments may be
found in Kirby (1997, 2003) and Madsen and Schffer (1999). The
resulting classical theory and modern extensions have played a key
role in the understanding of wave propagation and the development
of models that are useful for both scientic discovery and engineering application. A great deal of effort has gone into the extension of
the models to higher order in the parameter m 2 with the goal of
improving dispersion properties, resolution of the ow eld, and
representation of the nonlinear dynamics of complex wave elds.
This topic is reviewed extensively in Madsen and Fuhrman (2010)
and is covered only briey in this paper. Alternately, attempts to
improve and extend the basic theoretical framework for lower order
models continue as well. Finally, modern-day models incorporate a
wide range of extensions to account for additional physical effects
that lie outside the range of the classical formulation for inviscid
ow. Brocchini (2013) has provided a recent overview of a number
of aspects of the development and application of BTMs, including
the development of the theoretical basis since the 1990s, model
extensions to include a variety of physical effects, and the evolution
of numerical approaches. The resulting Boussinesq models are
highly evolved and are routinely applied to the study of fundamental
problems in ocean sciences in which this review is focused, as well
as to the applied engineering design of coastal infrastructure. There
is an extensive literature on the application of the model to coastal
engineering problems, such as harbor oscillation (Abbott et al. 1978;
Lepelletier and Raichlen 1987), morphology adjustment (Karambas
and Koutitas 2002; Xiao et al. 2010; Kim 2015), and breakwater and
reef overtopping and coastal inundation (Lynett et al. 2010a; Roeber
and Cheung 2012; Li et al. 2014).
This review concentrates on a description of recent developments and applications of leading order [i.e., O ( m 2)] BTMs,
which still represent the level of model development most frequently represented in application. The review is organized as
follows. There is a fairly general discussion of the development
of a basic O ( m 2) model to provide both a basis for present modeling approaches and connections to related developments in other
areas. Then, a brief discussion of the development of higher-order
versions of the theory is presented. Later, there is a brief overview
of numerical approaches, concentrating on the nite-volume
method, which is dominating the eld at present. Next, there is a
discussion of model extensions to incorporate additional physical
effects including wave breaking, boundary-layer driven turbulence and current shear, mixing and transport of scalars, and
applications to internal waves. The remaining sections describe
model applications in areas in which Boussinesq modeling has
provided a singular advance in understanding in recent years,
including applications to surf zone processes and tsunami generation, propagation, and inundation. There also is a discussion of
the appearance and rapid development of the so-called nonhydrostatic models (NTMs), which provide an avenue to development
of three-dimensional (3D) solutions for arbitrary free-surface
ows under fairly general conditions. Finally, the paper provides
suggestions for continued avenues for development.
ASCE

Overview of the Physical Basis for a Representative


O (l2) Model
Although the process for extending BTMs to great accuracy in
increasing water depth has met with great success, the resulting
models are complex and difcult to program and apply. Given the
development of the class of models now referred to as NHMs, it is
unclear if there is a niche for the continued development of highorder Boussinesq theory with pressure and velocities explicitly
developed in power series in z. However, there is a great deal of
ongoing use and development of low-order models for both scientic and practical application. This section provides a slightly different take on the derivation of such a model to provide some connection to the class of NHMs.
First, concentrate here on examining a consistent model for a
lowest order, fully nonlinear version of the Boussinesq model, following the results of Nwogu (1993) and Chen (2006). The derivation is presented in some detail to point out connections with developments in other areas that are not usually associated strictly with
the Boussinesq approach. Neglecting viscous stress terms, one
starts with the Euler equations in hx; t  z  h x; t
rh  u wz 0
ut u  rh u wuz

wt u  rh w wwz

(4)

1
rh p 0
r

(5)

1
pz g 0
r

(6)

where u and w = horizontal and vertical velocities; p = pressure; r =


uid density, and where still-water depth h and surface displacement h are single-valued functions of horizontal position x and
time t (thus allowing for the generation of wave motion through bottom motion). Density r is constant to eliminate the possibility of
vorticity generation through baroclinic effects. The motion is constrained by kinematic conditions

h t u  rh h w; z h

(7)

ht u  rh h w; z h

(8)

Finally, the pressure is equal to atmospheric pressure at the


surface
p pa x; t; z h

(9)

where again a variable pressure is allowed, as would be of interest,


for example, in the generation of meteotsunamis (Proudman 1929)
or in modeling of wave generation and modication by wind (Liu
et al. 2016).
The framework for developing the Boussinesq theory depends
on a scaling analysis. A scale h0 is adopted for depth and l 0 for horizontal variations, and let
u0

u
w
; w0
u0
w0

(10)

It is further assumed that u0 may be written as the product of a


phase speed c0 and a dimensionless parameter characterizing wave
height, which is denoted as d . In keeping with the idea that
our
p
theory will be a correction to the long wave limit, c0 gh0 .

03116005-2

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

Substituting Eq. (10) into Eq. (4) and forcing a leading order balance between the two terms gives w0 m u0 m d c0 , where m
h0 =0 characterizes the ratio of water depth to wavelength or other
horizontal scale. The shallow water limit follows from m ! 0,
whereas the Boussinesq theory follows from developing expansions
in powers of m  1. Turning to the momentum equations, it is recognized that the pressure must scale at leading order with the weight
of the water column, and let
p p0 p0 ; p0 r gh0

(11)

The resulting dimensionless forms of the governing equations


are given by (with primes dropped)

the size of d or further use of the parameter to determine where to


truncate the model equations. Subsequently, models of this type
will be referred to as fully nonlinear models, whereas models that
are truncated in powers of d with some underlying assumption
about the relative size of the Ursell number will be referred to as
being weakly nonlinear.
A weakly dispersive model following Chen (2006) is now
derived but with some generalization of the approach to draw parallels with other models that have not been placed in classical
Boussinesq form. First, the vertical momentum equation is integrated from an arbitrary elevation z to the surface dh to obtain
dh
pz d ~
Dwdz
(18)
p a  dh  z d m 2
z

rh  u wz 0

(12)

d ut d u  rh u wuz  rh p 0

(13)

d m 2 wt d u  rh w wwz  pz 1 0

(14)

The apparent smallness of the contribution of vertical acceleration to the pressure eld is clear in Eq. (14). Considering the kinematic surface boundary condition [Eq. (7)], h is characterized by an
amplitude a0 and obtains

h t d u  rh h w; z d h

~h t u  rh h d ~
h w; z h d ~
h

(16)

Likewise, if atmospheric pressure, pa, is constant, then its value


can be absorbed through the Bernoulli constant. Fluctuating pressure is then characterized using a0 as a scale for the pressure head,
leading to
(17)

In the absence of imposed forcing of either type, d is a free parameter constrained by initial or boundary data for the wave
motion.
The classic theory (Boussinesq 1872; Korteweg and de Vries
1895) corresponds to a double expansion in d and m 2 with each
assumed small and with the Ursell number d = m 2 O1. The
original Boussinesq or Korteweg-deVries (KdV) equations represent corrections to the linear, nondispersive shallow water theory,
which retains corrections to the leading order for weak O (d ) nonlinearity and weak O ( m 2) dispersive effects. This restriction on
the theory has long been superceded in practice due to the need, in
the case of short wind waves, to obtain a model that is valid for
larger values of m , as discussed later. It is now common to see the
development of models in which series expansions to much
higher powers of m are retained, together with no assumption on
ASCE

px; z; t z d p0 x; t m 2 p2 x; z; t

(19)

where p0 h represents a wave-induced hydrostatic pressure correction, and

(15)

where the choice d a0 =h0 has been introduced as the nondimensional parameter characterizing the motions amplitude. Turning to
the remaining conditions, it is noted that, if the motion is locally
forced by bottom displacement, then the magnitude of this motion
should serve as the scale for amplitude of the generated wave. (This
result is particularly clear in the usual approach to tsunami modeling
in which the initial static water displacement is set equal to the bottom displacement.) In this case, the depth should be separated into a
time-constant value hx scaled by h0 and a perturbation to this
value ~hx; t scaled by the amplitude a0, leading to

p d~
pa; z d h

where D t d u  rh wz  is a total derivative operator


following the ow. The only O (1) term in Eq. (18) represents the
increase in hydrostatic pressure down from the still-water level z =
0. The term in square brackets is only present in case of uctuating
air pressure. All terms scaled by d represent uctuating wave
motions relative to a constant background state. Neglecting atmospheric forcing, Eq. (18) can be written as

p2 x; z; t m

dh
2

Dwdz

(20)

represents a nonhydrostatic pressure correction. In the classical


theory, where terms at O ( m 2) are truncated at leading order,
D(w) would be replaced by wt in Eq. (20).
The classical Boussinesq theory would be further developed
here by replacing w in terms of u using the continuity equation [Eq.
(12)]. Integrating Eq. (12) from the bottom to elevation z and using
the kinematic boundary condition [Eq. (16)] gives
z
rh  udz
(21)
wz ht   uh  rh h 
h

which could be used in Eq. (20) immediately. Integrating up to dh


and using Leibnitzs rule then gives a depth-integrated volume conservation law
!
dh
udz 0
(22)
Ht rh 
h

where H h dh is the total local water depth, including any


effects of a moving bottom h(t).
Turning nally to the horizontal momentum equation, it is noted
that the crucial step in Chens development of a consistent theory at
leading order stemmed from requiring that the equation be integrated over the entire depth, as opposed to being evaluated at some
arbitrary level as had been done in Wei et al. (1995) or Chen et al.
(2003). Eq. (13) is approached by transforming the advective acceleration into a vorticity/Bernoulli head form to obtain


ut d Xiz  u n  wiz


1
rh d u  u w2  h m 2 p2 0
(23)
2
where n = horizontal vector component of vorticity; X = scalar vertical vorticity value, and where the terms involving squares of

03116005-3

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

velocities represent the Bernoulli head. The horizontal velocity may


then be expanded in powers of m 2 to obtain

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

ux; z; t u0 x; t m 2 u2 x; z; t O m 4

(24)

The choice of the depth-independent u0 is crucial to the accuracy


of the resulting model equations as depth increases. In the shallow
water limit for inviscid, irrotational ow, u0 is automatically identi , because terms of O ( m 2)
ed with the depth-averaged velocity u
vanish and the horizontal ow becomes depth uniform, with corresponding horizontal vorticity n 0.

In developing a fully nonlinear theory, Wei et al. (1995) effectively neglected V3 through the process of invoking a velocity
potential. Chen et al. (2003) closed the system by evaluating Eq.
(29) at z = za, neglecting the rst component of V3 because
u2 za 0. Chen (2006) closed the system by depth integrating Eq.
(29). After evaluating the Bernoulli head, higher-order acceleration,
and nonhydrostatic pressure, the resulting equations in a form
allowing for vertical motion of za is given by Shi et al. (2012)

h t rh  M 0

 3 O m 4
ua;t d ua  rh ua rh h m 2 V1 V2 V
(32)

Imposing n 5 0 : No Horizontal Vorticity


Boussinesq theory for irrotational waves is usually obtained by
imposing the restriction that the horizontal vorticity vanishes. This,
together with a sequential solution of the continuity equation and
bottom boundary condition for increasing powers of m , gives the
required expression for horizontal and vertical velocities as functions of z. The great contribution in the derivation of Nwogu (1993)
was in showing that choosing u0 ua uza to represent the horizontal velocity at an arbitrary elevation za z h led to a model
with great potential for optimization relative to the full linear
theory. Kennedy et al. (2001) further showed that allowing the reference elevation to move up and down with the free surface
za z h bh

where
A rh  hua;

where

z2a
h2
rh B zarh A  rh
Bt h At
V1
2
2
;t

A rh  ua

(27)

At leading order, w is given by wz m w1 z O m 3 , where


w1 z A zB

(28)


V2 rh za  h ua  rh A

(25)

further allowed for the freedom to optimize low-order nonlinear


properties of the model equations as well. These artices are now
commonly used in developing higher-order theories. (Allowing za
to move up and down is also critical to keeping the reference level
within the water column, which is not guaranteed in the case of a
xed elevation.) These choices lead to an expression for u2 z given
by

1
u2 z za  zrh A z2a  z2 rh B
(26)
2

1
A h B2
2

"

#!
 
p2
1 2
O m4
rh
d ua  u2 w1
2
r

(29)

where the rst three terms represent the depth-independent shallow


water momentum equation, and the remaining terms represent leading order dispersive effects and are still z dependent. The term V3 is
given by
V3 X0 iz  u2 X2 iz  ua

(30)

and represents the contribution at O ( m 2) to the cross-product of the


vertical vorticity and horizontal velocity.
ASCE


1 2
z  h 2 ua  rh B
2 a

V3 X0 iz  u2 X2 iz  ua

(33)

Optimization of the Basic Model


Nwogu (1993) showed that the choice of reference level za could be
used to optimize Eqs. (31) and (32) with respect to prediction of the
linear phase speed. Linearizing the model and applying it to a single
wave component in constant water depth yields the dimensional dispersion relationship


1
2
1  a kh
3
v 2 gk2 h
(34)
2
1  akh
where
 
1 za 2 za
a

2 h
h

Introducing the expansion X X0 m 2 X2 and using Eqs.


(24)(28) in Eq. (23) nally gives
ua;t d ua  rh ua rh h m 2 u2;t d V3

(31)

(35)

Madsen et al. (1991) achieved the same end result by manipulating linear operators in equations derived using depth-averaged velocity as the dependent variable. In either case, the choice a
2=5 reduces Eq. (34) to the (2, 2) Pade approximant, which is
accurate to O ( m 6). Nwogu (1993) suggested that the range of
model applicability could be extended to deeper water by adopting
a criterion for choosing a based on minimizing an error measure
over some range of kh values. Nwogu (1993) chose to minimize the
error in phase speed over the range 0  kh  3. Results for the ratio
of model/exact phase speed are shown in Fig. 1, in which the
dashed-dotted line represents a model with depth-averaged velocity as the dependent variable (Peregrine 1966, 1967), the dotted
line represents the Pade approximant, and the dashed line represents Nwogus optimization, giving a value of a 0:39 or
za 0:53h.

03116005-4

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

1.2

1.1

c/cexact

0.9

0.8

0.7

0.6
0

Fig. 1. Phase speed estimates relative to full linear theory: O ( m 2) Boussinesq based on depth-averaged velocity (dashed-dotted line); O ( m 2)
Boussinesq (data from Nwogu 1993) (dotted line); O ( m 4) Boussinesq (adapted from Gobbi et al. 2000, with permission) (dashed line)

120

h= 2.5cm
h= 5.0cm
100

h= 7.5cm
h=10.0cm
80

Surface elevation (cm)

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

h=12.5cm
h=15.0cm
60

h=17.5cm
h=20.0cm
40

h=25.0cm
h=30.0cm
20

h=35.0cm
h=47.0cm
0

20

22

24

26

28

30

32

34

36

38

40

Time (sec)

Fig. 2. Sample water surface displacements at wave gauge locations in Mase and Kirby (1992), Run 2: measured (solid lines); Nwogu optimization
(dashed lines); and Simarro optimization (dotted lines) (reprinted from Choi et al. 2015b, with permission)

The choice of optimization strategy used to pick a value of a is


not unique, and model Eqs. (31) and (32) can be further manipulated
to improve model accuracy. Schffer and Madsen (1995) showed
ASCE

that Nwogus weakly nonlinear model could be further manipulated


by operator rearrangement to give a model with a Pade (4,4) dispersion relationship of the form

03116005-5

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

1
1
2
kh4
1 kh
10
9
945
v gk h
Okh
4
1
2
4
1 kh kh
9
63

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

(36)

The resulting model has a signicantly improved range of accuracy in both phase speed and group velocity. No error minimization
strategy was used to try to improve the results further. Schffer and
Madsen (1995) also advocated the use of the linear shoaling gradient a5 to assess the accuracy of the model for waves over a sloping
bed, dened by
Ax
hx
a5
A
h

(37)

where A = linear wave amplitude. The complete linear theory gives


a5 as
1
1 G1  cosh 2kh
2
a5 G
;
1 G2

2kh
sinh 2kh

Ht rh  H
u 0
  rh u
 grh h
t u
u

1
p2 h
rh p2 h
rh h  h 0
2r
rH

(38)
t
w

Working backward from an O ( m 4) model of Gobbi and Kirby


(1999) described later, Kennedy et al. (2002) showed that coefcients could be chosen in the linear approximation in a manner that
eliminated all O ( m 4) terms without sacricing model accuracy.
This result is equivalent in spirit to the ndings of Schffer and
Madsen (1995). The derivations used to obtain the extensions to O
( m 4) accuracy in models expressing only O ( m 2) terms are quite
complex, and the results have not been used in any of the popular
lower order models that are presently in wide use.
Other authors (Simarro 2013; Simarro et al. 2013; Galan et al.
2012) have advocated optimization strategies based on a number of
criteria, including accuracy in phase speed, group velocity, local
shoaling gradient, and integrated energy ux measures. The question of whether the choice of optimization would improve results in
relatively shallow water was raised recently by Simarro (2015)
in connection with a study of surf zone currents (Choi et al. 2015a).
In response, Choi et al. (2015b) showed calculations for shoaling
and breaking random waves in the one-dimensional (1D) laboratory
experiment of Mase and Kirby (1992). The results, partially illustrated in Fig. 2, compare water surface elevations for model results
based on Nwogus optimization in comparison to results based on
the criteria of Simarro (2013). The results show that the predictions
of the two models are closer to each other than either is to the laboratory data. This result holds for a range of quantities including
wave height, power spectra, and third-moment statistics. This demonstration is not meant to imply that there would not be a range of
depths in which the optimizations could be distinct, particularly in
deep water. However, it is apparent that errors caused by a less than
perfect model setup and specication of boundary conditions can
overwhelm considerations of variations between models. This
result is likely to be even more pronounced in the eld in which
uncertainties in bathymetry, tidal and wind-driven ows, and incident wave conditions would likely combine to create greater variations than would be obtained due to differences in the models.
The Other Approach: Maintaining Nonhydrostatic
Pressure p2 as a Distinct Effect
Before moving on to higher-order Boussinesq models in the more
classical setting, consider briey the development of a model
obtained by depth integrating Eq. (29), with p2 retained and determined externally using some approximate form of Eq. (20). Models
ASCE

of this type, which will be referred to as depth-integrated NHMs,


may be thought of as a variant either of the low-order Boussinesq
models considered in this section or of the 3D NHMs discussed
later. Models of this type have appeared recently, both as alternate
approaches to modeling dispersion in surface waves, and as a means
for representing large vertical accelerations in depth-integrated, deformable owing bodies. Two prominent examples are the hydrodynamic model NEOWAVE, developed by Yamazaki et al. (2009),
and the model for avalanches based on saturated debris ow proposed by Denlinger and Iverson (2004). Yamazaki et al. (2009)
integrated the governing equations over depth and wrote them in
 , leading to
terms of depth-averaged velocities 
u; w

p2 h
rH

(39)

 is obtained by linearizing D(w) in the


where the last equation for w
expression [Eq. (20)] for p2 and neglecting all nonlinear terms arising while using Liebnitzs rule on the integral. The development of
the second equation depends in turn on the same linearization and
the assumption that 
p 2 1=2p2 h, which is suspect because p2
itself depends on w, which in turn varies linearly over depth.
Yamazaki et al. (2009) proceeded by solving the hydrostatic portion
of the problem using an appropriate NLSWE solver and then used
an approach based on Poissons equation to determine the pressure
correction, after which the velocity eld is again updated. A different approach for obtaining the nonhydrostatic correction is
described by Lu et al. (2015), in which the Poisson problem is not
used in the correction.
For the case of a granular debris ow, Denlinger and Iverson
(2004) proposed a similar approach but retained a more complete
version of the nonlinear expression for pressure, leading to an
expression for a reduced gravity g0 given by
g0 g D
w

(40)

which is then used directly in a modied hydrostatic pressure term.


 is dened as before and is
The depth-averaged vertical velocity w
evaluated directly using the kinematic boundary conditions.
Denlinger and Iverson (2004) showed that the resulting formulation
preserves the slope-oriented solution for the ow of a layer of uniform thickness, indicating that the model is valid for vertical velocities of the same order as horizontal velocities.
As pointed out by Castro-Orgaz et al. (2015), there is a close
connection between these approaches and the Boussinesq theory,
although the connection is not maintained adequately in all cases. In
particular, a careful derivation of the depth-integrated equations
shows that, for a linear variation of w over depth, the vertical variation of p2 is quadratic, which is at odds with the assumptions used in
Yamazaki et al. (2009) and indicates that their pressure correction is
incorrect at leading order. The structure of models of this type and
their relationship to the usual Boussinesq theory needs further consideration in the future. An interesting example of a procedure for
obtaining an accurate representation of the nonhydrostatic pressure
term, using Boussinesq scaling, may be found in Donahue et al.
(2015).

03116005-6

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Comparison of (a) horizontal and (b) vertical velocity proles at O ( m 2) and O ( m 4) (reprinted from Gobbi et al. 2000, with permission)

Evolution of the Theoretical Framework


The development of fully nonlinear Boussinesq models and methodologies for optimizing model properties has led to a continuing
effort to develop improved models, using both extensions of the series expansions to higher power in m 2 and the use of a ner vertical
subdivision of the water column. These extensions are reviewed
briey here. A more complete review of models based on higherorder series may be found in Madsen and Fuhrman (2010).
Higher-Order Series
Two early examples of the extension of models to higher order in m
include the work of Gobbi et al. (2000) and Agnon et al. (1999).
Proceeding in a fairly ad hoc manner, Gobbi et al. (2000) used a
weighted average of two velocity potentials dened at separate reference elevations and then used expansions to O ( m 4), obtaining a
fully nonlinear model with linear dispersion properties corresponding to the (4,4) Pade approximant. No further attempts at model
ASCE

optimization were made. The resulting model provided a fairly robust estimation of internal ow kinematics compared with the full
linear theory, as illustrated in Fig. 3. Gobbi and Kirby (1999) developed a variable depth version of the model in one horizontal dimension; this was subsequently extended to two horizontal dimensions
by Zhou and Teng (2010). This approach has not been extended to
higher order, as the extension to the strategy for choosing weighted
averages for potentials is not clear.
The work of Agnon et al. (1999) established the basis for a continued effort in developing higher-order models. Agnon et al.
(1999) rst developed a power series solution for Laplaces equation referenced to values at the still-water level, and then used operator methods to enhance the form of truncated representations of
horizontal and vertical velocities. The resulting model was truncated at leading order in bottom slope terms, which proves to be a
drawback in practice. Madsen et al. (2002, 2003) modied the procedure of Agnon et al. (1999) by referencing the series expansion to
a xed level in the water column rather than the still-water level,
and chose model coefcients by minimizing errors in predictions of

03116005-7

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

linear velocity proles. Madsen and Agnon (2003) examined the accuracy of the expansions in terms of radius of convergence.
The expansion strategies developed in this series of work can be
extended to high order, with the caveat that numerical implementation becomes difcult due to the presence of high-order spatial
derivatives. The model with truncation at O ( m 6) provides accurate
reproduction of horizontal velocities up to kh  25 and vertical
velocities up to kh  12. Fuhrman et al. (2004, 2006) have investigated the evolution of wave instabilities in deepwater wave patterns
and have demonstrated that the high-order model is a faithful representation of the full water wave problem.
Multilayer Approaches
Models based on extended higher-order series expansions are difcult to implement as a result of the presence of higher and higher
order spatial derivatives as model accuracy increases. Pursuing a
different strategy, Lynett and Liu (2004) suggested dividing the
water column into multiple layers and then formulating the O ( m 2)
in each layer, with the layers coupled by appropriate matching conditions. Lynett and Liu (2004) worked from the Euler equations in
each layer (which opens up the possibility of treating rotational
ows). The derivation is closed by imposing irrotationality. For the
case of two layers, it is possible to eliminate the lower layer velocity, and the resulting model is then a two-equation model for surface
displacement and horizontal velocity in the upper layer. Lynett and
Liu (2004) allowed for arbitrary, moving reference levels in each
layer, and chose the levels based on an optimization for combined
errors in phase speed, group velocity, and shoaling coefcient. The
model provided accurate phase speed up to kh  6 and group velocity up to kh  4. More importantly, the division of the velocity proles into matched segments eliminated the development of reversed
ows or inection points at large kh values, as seen in single-layer
models such as Gobbi et al. (2000). A slightly different version of
the two-layer model, developed to O ( m 2) based on Laplaces equation, may be found in Liu and Fang (2015), which retains the lower
layer velocity as a dependent variable and achieves greater accuracy
in linear dispersion.
Further examples of multilayer approaches include the work of
Chazel et al. (2009), in which the higher-order expansion (Agnon
et al. 1999; Madsen et al. 2003) is applied to the irrotational problem
in each of two layers, and Bai and Cheung (2013), in which a procedure based more closely on the NHM formulation is applied to multiple layers.
Integral Transform Methods
The problem of linear wave propagation in a region of constant
depth can be fully described using an integro-differential equation
with a kernel based on the Fourier transform of the phase velocity.
Karambas and Memos (2009) used this approach to develop a fully
dispersive BTM for weakly nonlinear waves over a mild slope. The
model is seen to be quite accurate compared with laboratory examples, in spite of the limitation of weak nonlinearity and the fact that
the convolution integral is only strictly accurate in constant depth.
The second restriction ends up not being important because the integration kernel dies off rapidly with distance. Further enhancements
of this model are described in Memos et al. (2016).

Numerical Approaches
The review of Brocchini (2013) provided an extensive overview
of numerical methods used in Boussinesq models, including
ASCE

nite-difference, nite-volume, and nite-element models.


Only a few aspects of this topic are touched on here.

Finite Difference
Wei and Kirby (1995) developed a nite-difference scheme for the
weakly nonlinear equations of Nwogu (1993), using a mixed-order
centered difference scheme in space and the fourth-order AdamsBashforth-Moulton (ABM) scheme in time. The ABM scheme has
remained a popular choice for solving Boussinesq-type problems
and has been used as the basis for a number of codes, along with
higher-order Runge-Kutta methods. The centered difference
scheme for spatial derivatives proved to be noisy in practice, requiring lters to suppress noise near shorelines and in locations with
rapidly changing solutions, such as breaking wave crests. The use
of staggered grid schemes proved to generally be more workable,
but to a large extent the approach has been replaced by widespread
use of nite-volume schemes.

Finite Volume
The vast majority of BTMs in use today are based on nite-volume
schemes. This approach has become dominant largely because the
robustness of numerical schemes developed for the shallow water
equations (Toro 2001; Leveque 2002) carries over to the
Boussinesq application. Starting with the work of Erduran et al.
(2005) and Cienfuegos et al. (2006), the approach has largely supplanted the use of nite-difference models. The shock-capturing
capabilities of the codes, when applied to the NLSWE, has motivated the use of hybrid BTM/NLSWE models in which the shockcapturing properties are allowed to take over and control the wave
breaking process by suppressing the dispersive portion of the governing equations, as discussed later.
Various nite-volume formulations have been used. A number
of models (Erduran et al. 2005; Kim et al. 2009; Tonelli and Petti
2009; Shi et al. 2012; Kazolea et al. 2012) use the fourth-order
MUSCL-TVD scheme to reconstruct velocity variables, and ux
computations are often based on the HLL scheme. The choice of
time-stepping algorithms is fairly evenly split between the ABM
explicit/implicit scheme and various versions of Runge-Kutta
schemes. There are a number of alternative choices that have been
made in all these categories in various models. The models, overall,
are characterized by extreme robustness when compared with their
nite-difference predecessors, and they provide a natural framework for handling two of the problem areas that generally were
troublesome in nite-difference calculations: breaking wave formation, which is treated naturally by the shock-capturing capabilities
of the numerical schemes, and wetting-drying boundaries, which
are handled straightforwardly when the information needed to specify shoreline movement is already carried in the Riemann variables
computed as part of the solution.

FEMs
The FEMs for BTMs have been developed beginning with the
work of Walkley and Berzins (1999, 2002). Modern developments have focused on using the discontinuous Galerkin method
and are reviewed in Brocchini (2013). A recent addition in this
area is the discontinuous Galerkin model of Panda et al. (2014)
applied to the fully nonlinear model of Zhang et al. (2013).

03116005-8

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

ADV A
3

U (m/ s)

2
1
0

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

-1
-2
0

10

15

20

25

t(s)

30

35

40

45

Fig. 4. Comparison of measured and modeled horizontal velocity at


ADV A for the run-up experiment of Lynett et al. (2010b); results computed using FUNWAVE-TVD (data from Shi et al. 2012); solid line:
data; dashed line: eddy viscosity breaking model; dash-dot line: hybrid
model

Additional Physics
Wave Breaking
Most aspects of explicit models for breaking wave dissipation,
including roller models (Schffer et al. 1993), detailed vorticity
models (Veeramony and Svendsen 2000), or eddy viscosity models
(Zelt 1991; Kennedy et al. 2000) were already established at the
time of the review of Kirby (2003) and were discussed there. Of
these methods, the eddy viscosity method is still widely used.
Progress in this area since 2003 includes the development of the relative trough Froude number (RTFN) as a breaking criterion for use
in either eddy viscosity or roller models (Okamoto and Basco
2006), and the breaking celerity index (BCI) (DAlessandro and
Tomasicchio 2008), which combines the RTFN criterion with the
earlier criterion developed by Kennedy et al. (2000). Each of these
sees occasional use, along with the original version. Cienfuegos
et al. (2010) further added terms representing breaking effects to
both the mass and momentum equations. Each term is in the form of
a diffusion operator, with diffusivities determined from calibration
against laboratory data.
After the amount of effort devoted to the development of explicit
expressions for wave breaking effects in the 1990s and early 2000s,
it is somewhat surprising that most recent progress has been made
using a hybrid approach in which dispersive effects are turned off
when some criterion is reached, reducing the model to the
NLSWEs. Already steep wave crests rapidly evolve toward discontinuous jump solutions, which are characteristic of the NLSWE,
and lead to energy dissipation rates found in hydraulic jumps or
bores with comparable crest-to-trough surface elevation changes.
The approach, which appears to have rst been introduced by
Tonelli and Petti (2009), goes hand in hand with the adoption of the
nite-volume method as the dominant computational approach,
because these methods had already evolved to provide robust
shock-capturing capabilities for the NLSWE (Toro 2001). Because
the consequence of turning off dispersion and shifting to the
NLSWE form is basically the same in any robust solution method
for the NLSWE, the differences between implementations of the
strategy basically boils down to choosing criteria for (1) when to
turn dispersion off and (2) when to turn it back on.
Tonelli and Petti (2009) used a Froude number criterion based
on the ratio of crest-to-trough wave height to water depth in front of
ASCE

the crest. The criterion is used to choose between NLSWEs if the


critical value is exceeded and the Boussinesq model if the criterion
is below the critical value. The same approach is used in Tonelli and
Petti (2010) and Shi et al. (2012). Tonelli and Petti (2011, 2012)
adopted a lower value of the criterion for switching from the
NLSWEs back to the Boussinesq form to improve model stability.
Roeber and Cheung (2012) used a different criterion based on
the gradient of momentum. The measure is more local than the
Froude number criterion of Tonelli and Petti (2009, 2010, 2011),
and often allows the neglect of dispersion and resulting dissipation
to be localized on the front face of breaking wave crests. Tissier
et al. (2012) used a very different strategy for handling breaking in
their splitting scheme. In the rst step of the scheme, which solves
the NLSWEs, the wave eld is interrogated for the presence of
energy dissipation spikes, which are indicative of the formation of
shocks in the nondispersive solution step. If a measure of normalized dissipation is below a certain value, the second, dispersive
model step regularizes the solution, suppressing breaking. Beyond a
critical level, the wave crest is examined to test for exceedance of a
local surface slope criterion. If this second criterion is exceeded,
dispersion is turned off locally in the second solution step, preserving the dissipation predicted by the NLSWEs. This approach is
adopted in the unstructured grid nite-volume model of Kazolea
et al. (2012) as described in Kazolea et al. (2014).
There has not been a great deal of testing of how different numerical implementations perform when using different approaches
to breaking. Fig. 4 shows a comparison of an eddy viscosity calculation (Kennedy et al. 2000) and hybrid calculation (Tonelli and Petti
2009) in which both are implemented in FUNWAVE-TVD (Shi
et al. 2012). These data are three components of velocity collected
at an ADV located in front of the shoal in the experiment of Lynett
et al. (2010b); there is also a sequence of snapshots of the event later
(see Fig. 17). The two breaking models were run using standard
choices of parameters, and the results are basically indistinguishable. Kazolea et al. (2014) showed similar results for solitary wave
run-up on a plane slope. On the other hand, results for periodic surf
zone waves shown by Kazolea et al. (2014) show a greater difference between the approaches, with the hybrid model apparently performing better.
Turbulence, Vorticity, Boundary Layers, and Mixing
Very few BTM models address the presence of turbulence and vorticity in a direct manner. All BTM models with advective acceleration effects in the horizontal ow eld can advect vertical vorticity
once it is generated but do not address horizontal vorticity associated with turbulent velocity proles or interaction between the components. Lateral mixing effects resulting from turbulence are often
handled using Smagorinsky-like diffusion operators, as is common
in ocean modeling, and bottom friction is treated using standard
quadratic formulations based on depth-averaged velocity with calibrated friction coefcients.
Kim et al. (2009) have developed a more comprehensive treatment of the depth-averaged ow eld, taking into account the presence of turbulence and vorticity. Kim and Lynett (2011) developed
depth-integrated scalar transport equations along similar lines.
They also added a stochastic component to the expression for turbulent stresses, in recognition of the fact that the model failed to develop energetic mixing layers between coowing streams over at
bottoms. As mentioned by the authors, this component of the model
is less crucial for ows over variable bathymetry, in which instabilities and resulting mixing occur more readily. Kim (2015) used this
model together with an erosion/deposition model to model suspended sediment ux and resulting morphology adjustment. Kim

03116005-9

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

and Lynett (2013) instead used a s coordinate transport model in


three dimensions (3D), with the upper boundary determined by the
BTM. This conguration becomes more like a 3D NHM formulation (Ma et al. 2013a) but with the velocity eld determined semianalytically from the depth-integrated BTM.
This sequence of papers represents a concerted effort to inject
missing physics (turbulence, vorticity, and mixing) into the BTM
framework, which is usually based on the inviscid Euler equations
as a starting point, with additional effects added in an ad hoc fashion. It has not been clearly demonstrated, however, whether the
extra care leads to an obviously superior model in terms of predictive capabilities. Direct one-to-one comparisons between models or
against data are difcult, because many of the processes being studied are highly irregular or chaotic, such as the generation of large
eddies during strong ows through harbor entrances. The evaluation
of these extended models continues, and it will for some time before
the end-user community is ready to adopt them as the basis for a
standard BTM form for real-world geophysical ows.
Waves on Sheared Mean Currents
Waves in coastal settings may encounter strong currents that are
sheared in the vertical, either through the effects of bottom stress
and turbulence, or in response to inhomogeneities, such as density
stratication. A few attempts have been made to develop
Boussinesq equations that are generalized to take the presence of
strongly sheared mean ows into account. Two early examples
are the work of Shen (2001) and Rego et al. (2001). Rego et al.
(2001) used an approach that was limited to one horizontal
dimension and used a scalar stream function to represent the rotational part of the wave and current eld. The resulting equations
are similar to the model of Veeramony and Svendsen (2000) for
wave breaking but allow for nonzero vorticity at the bottom in
recognition of the dominance of bottom stress in determining the
mean ow prole. Shen (2001) provided a different formulation
based on the vertical velocity w as the main dependent variable,
which is a natural choice in developing wave-current models
because it generalizes the stream function problem to three
dimensions in a straightforward way (McWilliams et al. 2004).
Both derivations work only with the inviscid Euler equations;
thus, details about the vorticity distribution have to be imposed
externally, either directly (Shen 2001), or through specication of
a mean current prole (Rego et al. 2001). In the latter case, the
vertical structure of wave motion was determined in the linear
limit using the Rayleigh equation and used to test the accuracy of
model formulations. The dispersion relationship resulting from
the model by Rego et al. (2001) may be written as


g 1 k2 g 2
2

s s s s g 0 kh  s s k
2
1  akh


2
1  a 1=3kh


gk  g 0 s s kh
(41)
2
1  akh
where s s v  kUs = intrinsic frequency at the mean water surface; and g 0  g 2 are integrals of the current vorticity over water
depth. Fig. 5 shows plots of the linearized model phase speed cb normalized by the exact linear solution ce for two assumed current proles and for a range of kh = m values. The model accuracy is comparable to the formulation of Nwogu (1993) and reduces to it in the
limit of zero vorticity.
ASCE

More recently, Son and Lynett (2014a) have followed up on this


work and have generalized the rotational Boussinesq model of Kim
et al. (2009) to include the presence of a sheared, turbulent mean
ow. The model also depends on the specication of an imposed
mean current, but the model physics allow for waves to modify the
vertical structure of the mean ow. Figs. 6 and 7 show a comparison
of model results to laboratory measurements of mean current proles obtained by Kemp and Simons (1982, 1983) for the case of following and opposing currents, respectively. These results indicate
that following currents experience a drop in current velocity at the
surface, whereas opposing currents become more strongly sheared.
These results are well understood and have been reproduced by 3D,
wave-averaged models. The power of the Boussinesq approach lies
in the use of a depth-integrated model, with the mean current represented as a polynomial with coefcients determined as part of the
solution.
All of these treatments depend on the specication of a current
eld with which the waves then interact. This implies that, in a realistic setting, one would need to run a circulation model to provide
the current and then run the Boussinesq model with circulation
model results provided as input. The Boussinesq model results
could, in principle, be used to develop wave forcing for the circulation model, and the two could be run in a fully coupled fashion. The
newly available NHMs, reviewed briey in the next section, largely
eliminate this necessity by providing a unied simulation of the
entire ow eld at resolutions that would be comparable to usual
choices for the Boussinesq model.
Wind Effects on Waves
Douglass (1990) conducted a set of laboratory experiments that
showed offshore- or onshore-directed winds could have a dramatic
effect on the shoaling evolution and location of wave breaking, raising the possibility that models that do not account for these effects
could provide inaccurate predictions of surf zone width and wave
height distribution, leading in turn to lower quality predictions of
nearshore circulation and sediment transport. These conclusions
were supported by further experiments of Feddersen and Veron
(2005), who showed that wind could have a dramatic effect on
wave shape and, by extension, the skewness of the wave orbital velocity. Although this problem has still not been pursued in a comprehensive way, some attempt has been made to add wind effects
(beyond simple application of a spatially uniform surface stress) to
Boussinesq models. Chen et al. (2004) added a modied surface
stress term to the Boussinesq model, which incorporated a dependence on surface slope in the determination of the drag coefcient
CD in a conventional wind stress formulation. The resulting model
was tested against data for wave growth with fetch, and showed
some skill in predicting the change in wave height, but failed to predict concurrent increase in wave period. Liu et al. (2016) introduced
a new source formulation based on a wave-induced pressure perturbation (Jeffreys 1925) instead of wind stress, and further developed
a spectral model to examine the effect of wind on nonlinear triad
interaction and recurrence, but did not provide comparisons with
data for the new model or provide any relative evaluation of the two
formulations or any comparison of effects on local surface geometry. Work in this area is still in its infancy, with much to be done to
arrive at a point in which an understanding of wind effects on the
breaking process is in hand.
Internal Waves
Internal waves, supported by vertical density stratication in the
coastal or upper deep ocean, often fall into a regime in which the

03116005-10

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Normalized phase velocity (model velocity cb divided by the exact phase velocity calculated from the Rayleigh equation, ce) and associated
current proles: (a) 1/7 power law; (b) cubic polynomial (reprinted from Rego et al. 2001, ASCE)

wavelength is long relative to either the overall depth or, in the case
of distinct density layers, to the depth of either the upper or lower
layer. The case in which both layers are shallow is naturally treated
by models developed in the Boussinesq framework, and there are a
number of instances in the literature of idealized treatments using
KdV-type equations in one horizontal dimension (Benjamin 1966),
K-P equations for weakly two-dimensional (2D) problems (Kirby
1988; Pierini 1989), or Boussinesq-type equations for fully 2D
problems (Tomasson and Melville 1992). Extensions to fully nonlinear and higher-order models for waves in two-layer systems continue to be developed as well (Choi and Camassa 1999; Debsarma
et al. 2010). However, implementations of models for general 2D
geometries are relatively rare. One example is that of Lynett and
Liu (2002b), who used a procedure similar to that used in Lynett
and Liu (2004) to eliminate the velocity variable in one layer and
obtain a set of two coupled equations in the interface displacement
and a single momentum ux variable. The model was implemented
and applied to several idealized cases involving shoaling through a
transition from a deeper to a shallower lower layer (in which the
sign of quadratic nonlinear effects changes), and internal wave passage through sudden expansions. Fig. 8 shows an example for an
idealized geometry corresponding to the Strait of Gibraltar, together
with a satellite image of the same location. There is room for continued development of this type of model for internal waves; however,
discretely layered systems are not always an appropriate model for
ocean stratication, and it is likely that future developments would
be more likely to occur in the context of the 3D NHMs, which are
discussed next.

Modeling Surf Zone Flows


Within the limitations imposed by prescribed and oversimplied
vertical ow structure, BTMs represent a comprehensive approach
for studying wave-driven circulation in the surf zone and nearshore.
BTMs describe the loss of wave energy at large scale due to breaking and cross-shore wave height decay, and they provide details of
this process down to the scale of individual breaking wave crests.
ASCE

They can be extended to include detailed formulations for turbulent


processes and mixing (Kim and Lynett 2011), although it is not yet
clear that such extensions are needed once the ow eld is sufciently perturbed (Kirby et al. 2006).
The ability of BTMs to describe the cross-shore decay of wave
height due to breaking has been well established. This section
reviews recent results aimed at describing the large- and ne-scale
structure of surf zone currents arising from a range of forcing
mechanisms.
Surf Zone Currents
Given appropriate boundary conditions and within the limitations
of the depth-averaged ow eld description, Boussinesq models
provide a calculation of the entire ow eld. Yoon and Liu (1989)
rst derived a coupled wave-current system based on Boussinesq
scaling. Equivalent systems of equations can be obtained directly
from a Boussinesq model, but the model must be of a fully nonlinear
type. Comparable derivations, starting from weakly nonlinear models and using a decomposition of velocity into wave and current
components, lead to defective sets of coupled equations, indicating
that the wave-driven mean ows may be of suspect accuracy. There
are, however, no detailed model intercomparisons of weakly nonlinear versus fully nonlinear models and comparison with data that
would substantiate this claim, and it is assumed for now that weakly
nonlinear and fully nonlinear BTMs are equally valid predictors of
surf zone dynamics.
BTMs have been extensively used to study the large-scale structure of surf zone ows. Chen et al. (1999) conducted a study of rip
currents generated by segmented laboratory bars. Chen et al. (2003)
developed an improved representation of modeled vertical vorticity
effects, and applied the model in a study of alongshore currents during the Duck Experiment on Low-Frequency and Incident-Band
Longshore and Across-Shore Hydrodynamics (DELILAH) eld
experiment at Duck, North Carolina. This study revealed the
growth to nite amplitude and breakdown of shear waves and
showed that alongshore pressure gradients can play a signicant
role in the alongshore momentum balance over beaches with

03116005-11

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

0
Model
Experiment
Currentonly

0.02

0.04

0.08

0.08
depth(m)

0.06

depth(m)

0.06

0.1

b=0.5

0.1
0.12

0.12

0.14

0.14

0.16

0.16

0.18

0.18
0.2

0.2

0.05
0.1
0.15
Timeaveraged horizontal velocity(m/s)

0.05
0.1
0.15
Timeaveraged horizontal velocity(m/s)

0.2

0
Model
Experiment
Currentonly

0.02
0.04

b=0.8

0.04

0.08

0.08
depth(m)

0.06

0.1

0.12

0.14

0.14

0.16

0.16

0.18

0.18
0.05
0.1
0.15
Timeaveraged horizontal velocity(m/s)

0.2

b=1.2

0.1

0.12

Model
Experiment
Currentonly

0.02

0.06

0.2

0.2

depth(m)

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

0.04

Model
Experiment
Currentonly

0.02

b=0.2

0.2

0.05
0.1
0.15
Timeaveraged horizontal velocity(m/s)

0.2

Fig. 6. (Color) Mean velocity proles of combined waves and currents from experiments of Kemp and Simons (1982); waves on following currents
(reprinted from Coastal Engineering, Vol. 90, Son and Lynett, Interaction of dispersive water waves with weakly sheared currents of arbitrary prole, pp. 6484, Copyright 2014, with permission from Elsevier)

fairly weak alongshore variations. Additional comparisons


of wave-averaged circulation and model-predicted wave statistics
may be found in Feddersen et al. (2011), Geiman et al. (2011),
Feddersen (2014), and Choi et al. (2015a). BTMs provide an
accurate model of bulk surf zone properties, indicating their
potential suitability for studies of ner scale processes.
Undertow: The Missing Wave Roller
One deciency in early BTMs, recognized during the early development of explicit breaking models, was the absence of a roller
effect and a resulting decit of cross-shore Eulerian mean return
ow, or undertow, under shoreward propagating breaking waves.
This decit was noted by Madsen et al. (1997) and arises because
the usual volume conservation equation, which is capable of correctly describing Stokes drift effects, is not capable of describing
roller volume ux unless details of the roller structure are provided explicitly. This effect is not often incorporated in waveresolving calculations. (A counterexample may be found in Long
2006 in which instantaneous roller volume ux is accounted for
ASCE

in both continuity and momentum equations.) The consequence


of this neglect can be seen in Fig. 9, which shows comparisons of
Eulerian mean ow from a BTM in standard form (red dashed
lines) to measured undertow proles (Lynett 2006; Cox et al.
1995). The discrepancy between the two is quite pronounced in
the outer surf zone in which wave height is still large and roller
volumes, which scale with the square of wave height, are largest.
Lynett (2006) developed an enhanced-breaking model, which
provides a vertical structure for the velocity correction associated
with the roller together with expressions for the additional uxes
appearing in the mass and momentum equations. Results for the
case in Fig. 9 are shown as the blue line and show a signicant
improvement in reproducing undertow. The model is not included
in the BTM equations themselves, it is only applied in a postprocessing sense. As a result, modications that could arise in quantities, such as cross-shore bottom stress, as a result of changes in
the near-bed Eulerian velocity eld, are not incorporated. The
problem of properly incorporating roller volume and momentum
uxes and obtaining more correct estimates of Eulerian near-bed
velocities still appears to be largely open.

03116005-12

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. (Color) Mean velocity proles of combined waves and currents from experiments of Kemp and Simons (1983); waves on opposing currents
(reprinted from Coastal Engineering, Vol. 90, Son and Lynett, Interaction of dispersive water waves with weakly sheared currents of arbitrary prole, pp. 6484, Copyright 2014, with permission from Elsevier)

Fig. 8. (a) Numerical snapshot of internal wave passing through the Strait of Gibraltar and (b) satellite image of the same location (reprinted
from Wave Motion, Vol. 36, Lynett and Liu, A two-dimensional, depth-integrated model for internal wave propagation over variable bathymetry, pp. 221240, Copyright 2002, with permission from Elsevier)

ASCE

03116005-13

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. (Color) Comparison with undertow data of Cox et al. (1995) [Note: Experimental undertow values are shown with dots; breaking-enhanced
Boussinesq results are shown by solid lines; and unmodied Boussinesq results are shown by the dashed-dotted lines (reprinted from Lynett 2006,
with permission)]

Short-Crested Incident Waves


Since the work of Peregrine (1998), who studied the impulsive generation of vertical vorticity occurring along a breaking wave crest
with along-crest wave height variation, it has become clear that a
major source for vorticity generation in the wave-driven ocean is
associated with detail at the scale of the wave crests. In the deep
ocean, this additional contribution to forcing is associated with the
pattern of whitecaps on the surface. This contribution can be successfully treated using stochastic representations (Sullivan et al.
2007) because the spatial coverage is uniformly distributed and the
scales of variation of the wind forcing are large relative to the scales
associated with distance between whitecap events. The addition of
this stochastic component of forcing leads to signicant changes in
levels of upper ocean turbulence, transport by Langmuir circulation,
and mixed layer deepening.
In contrast, the nearshore environment is complex in ways that
would likely defy the successful development of a purely stochastic
forcing model. The surf zone scale is small relative to the scale of
the pattern of breaking dissipation. The location and structure of
breaking can be controlled as much by the pattern of inhomogeneities in depth as by the intrinsic short-crestedness in the waves themselves [Fig. 10(a)], which shows an aerial view of breaking events
during an experiment at the United States Army Corps of Engineers
Field Research Facility at Duck, North Carolina. Clark et al. (2012)
have performed a detailed study of vorticity generation at the scale
of breaking wave crests, verifying that the predictions of Peregrine
(1998) are correct. However, the likelihood of developing a simple
stochastic extension to the smoother wave forcing elds that
are characteristic of wave-averaged models seems remote given the
inevitable interplay between the intrinsic short-crestedness of the
ASCE

incident wave eld and its interaction with a similarly variable


underlying bathymetry. The wave-resolving nature of the BTM is
thus crucial for the specication of this variability and the localization of vorticity forcing.
A number of studies have used BTMs to examine small-scale
forcing and resulting ow eld turbulence and mixing effects. In
a pioneering study, Johnson and Pattiaratchi (2006) simulated
directional random waves incident on a plane beach. They
clearly demonstrated the presence of vorticity generation during
the passage of breaking wave crests, and they further demonstrated the tendency for ne-scale vorticity, accumulated in the
nearshore, to organize into larger scale structures and nally to transient rip currents (Fig. 11). Similar numerical experiments were performed by Spydell and Feddersen (2009), who concentrated on the
evaluation of Lagrangian particle statistics and the estimation of surf
zone diffusivity. Feddersen et al. (2011) and Clark et al. (2011)
extended the comparison of BTM results and eld observations to
the case of obliquely incident waves to obtain estimates of both
cross-shore and alongshore diffusivities. Feddersen (2014) and Choi
et al. (2015a) examined strong alongshore ows observed during the
SandyDuck experiment in 1997. Feddersen (2014) compared results
of a BTM model with resolved short-scale forcing and a waveaveraged circulation model forced smoothly by radiation stress
gradients derived from the incident wave conditions. Both studies
showed the presence of a strongly energetic ow eld in frequency/alongshore wave-number space lying along a line that represents the local mean alongshore current velocity, representing
the alongshore transport of eddy structures by the mean current
(Figs. 12 and 13). Feddersen (2014) showed that the smoothly
forced model did not generate a comparably energetic ow eld.

03116005-14

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

14 presents results for a case with relatively weak alongshore ow


and particle trajectories generally trapped in rip recirculation cells.
Results for this case indicate that apparent diffusivity is basically
the same for the wave-averaged and wave-resolving models, with
both in general agreement with eld observations.
Clearly, there is something going on here that is very different
from the case of waves incident on a featureless beach in which the
spatial scale associated with the short-crestedness of waves is the
only scale imposed on the problem. For rip channels, the channels
appear to provide the dominant scale, with differences between
smooth and irregular wave forcing lost in the noise. These results are
distinct and point to some possibly interesting middle ground in
which controlling scales supplied by different mechanisms are comparable in importance and are able to interact in a more balanced
way, with an unforeseeable outcome. BTM models are perfectly
adequate for this type of study and are likely to remain a valuable
tool for studying wave-driven surf zone ows in the near future.

Tsunamis: Generation, Propagation, and Inundation

Fig. 10. (Color) (a) Photograph of breaking waves (propagating toward the shore from lower right to upper left) showing the triangular
patches of residual white foam marking the location in which breaking
occurred; (b) schematics (looking down from above) of negative and
positive vorticity generated by left-handed and right-handed ends of
breaking waves and possible variations in positioning of breaker termination relative to sensing array (dark circles) (reprinted from Clark
et al. 2012, with permission)

All of these studies, in which waves are propagating and breaking over fairly featureless bathymetry, show ow complexity and
mixing processes in wave-resolved calculations that are consistent
with observations and much stronger than predicted by waveaveraged models. [Indeed, in the case of normally incident waves
(Johnson and Pattiaratchi 2006; Spydell and Feddersen 2009), a
wave-averaged model would simply predict setup with no resulting
mean ow.] Geiman et al. (2011) performed a similar model/
model/data comparison for waves incident over a complex beach
incised by rip channels located reasonably regularly in the alongshore direction. For each model [FUNWAVE as the BTM model
and Delft3D (Lesser et al. 2004) as the wave-averaged model],
Lagrangian particle trajectories were computed using modelgenerated ow elds and then used to compute particle separation
statistics, which is a precursor to deriving diffusion estimates.
Model-generated statistics were compared with each other and to
similar statistics derived from drifter trajectories in the eld. Fig.
ASCE

In recent years, Boussinesq models have come into play as a frequently used component in the process of modeling tsunamis. The
choice of BTMs over NLSWE models for modeling tsunamis has
been somewhat controversial, because the problem is thought by
many to be essentially nondispersive and thus not requiring the
input of the more expensive BTMs. Initially, the use of BTMs was
often a matter of preference, with investigators who were involved
in developing or using BTMs for surface wave applications choosing to use the models in tsunami applications. Subsequent work has
revealed a number of areas in which the capabilities of BTMs relative to NLSWEs come into play in possibly signicant ways.
Tsunami hydrodynamic modeling may be thought of as being divided into three categories: generation due to ground motion, propagation over local to ocean basin scales, and shoreline inundation. These
aspects of the problem may be treated separately in separate model
runs, or comprehensively in a unied, nested treatment, depending on
the choice of model and preferences of the investigator.
Tsunami Generation
Dynamic tsunami generation using BTMs has largely been limited
to modeling the response to an imposed time-dependent ground
motion. Both seismic subduction zone events and smaller scale submarine mass failures (SMFs) have been simulated this way. Lynett
and Liu (2002a) described a study of SMF tsunami generation in
one dimension, using a Boussinesq model extended to account for
time-dependent bottom motion. A BIEM model was used to provide
a reference solution for evaluation of the Boussinesq model, and the
Boussinesq model was seen to differ from the full solution, particularly in details of the maximum drawdown above the moving slide.
Lynett and Liu (2005) extended the modeling to two horizontal
dimensions and studied the generation of edge waves resulting from
slides occurring on planar slopes.
Zhou and Teng (2010) extended the O (m 4) model of Gobbi and
Kirby (1999) to two horizontal dimensions and incorporated a moving
bottom. They performed 1D laboratory experiments for wave generation by a sliding solid wedge and measured waveforms as well as velocity proles on vertical transects. Model-data comparison showed
that there was not much difference between the O ( m 4) model results
and results computed using the equations of Lynett and Liu (2002a),
with the difference between model results being smaller than the differences between either model and their measurements. The O ( m 4)
model provided consistently better estimates of the velocity eld,

03116005-15

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. (Color) Generation of local vorticity patches at ends of breaking wave crests (reprinted from Johnson and Pattiaratchi 2006, with permission)

Fig. 12. Horizontal distributions of vorticity uctuations for two peak wave periodaveraged velocity at (a) t = 30 Tp, (b) t = 410 Tp, (c) t = 420 Tp,
and (d) t = 430 Tp in the present simulation of the SandyDuck experiment (reprinted from Choi et al. 2015a, with permission)

however, which could be expected due to the greater level of detail on


the ow structure preserved in the higher-order models formulation.
Fuhrman and Madsen (2009) used the higher-order Boussinesq
model (Madsen et al. 2002, 2003) extended to retain higher-order
bottom slope effects (Madsen et al. 2006) and incorporate bottom
ASCE

motion. Among a number of tests, Fuhrman and Madsen (2009)


repeated the cases studied by Lynett and Liu (2002a) to compare
results with the BIEM results from the earlier study. Comparisons
with the BIEM were greatly improved and indicate the higher-order
models apparent approach to convergence to a complete solution.

03116005-16

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 13. Wave number-frequency spectra obtained using measurements of (a) cross-shore and (b) longshore velocity and the computations of (c)
cross-shore and (d) longshore velocity for the longshore array at x = 160 m and y = 704,906 m [Note: The logarithmic gray scale indicates energy density; the bold lines denote mode-0 edge waves propagating upcoast and downcoast with a small wavenumber offset (0.0015 m1) (reprinted from Choi
et al. 2015a, with permission)]

Studies of real-world SMF events performed using BTMs have


appeared in the literature (Geist et al. 2009; Parsons et al. 2014) but
have usually used fairly idealized approaches to specifying bottom
motion. The greatest strides remaining to be made in this area will
result from full coupling of the hydrodynamic models (in either
BTM or NHM form) to more complete models for slide dynamics,
including initiation of motion, slide deformation, and cessation of
motion (Kirby et al. 2016).
Propagation Modeling
Along with the usual impact of bottom friction damping for progressive long waves, tsunamis propagating at ocean basin scale are
weakly affected by a number of processes including nonlinearity,
frequency dispersion, Coriolis effects, and compressibility/deformability of the water column and underlying earth crust. All of these
effects are very weak in a classical scaling sense, but can accumulate to recognizable amounts given the long propagation distances
that are possible for transoceanic events. In addition, locally generated tsunamis resulting from landslides and other geotechnical failures are much shorter in period and can exhibit distinctly dispersive
effects over moderate distances.
The use of BTMs for ocean basinscale propagation problems is
still relatively uncommon. A body of evidence indicating the importance of dispersion for tsunami propagation has only been recently
developed and is often based on comparison of model predictions
within the same model with dispersive effects either included or
neglected. This sort of comparison presupposes the existence of the
dispersive model in the rst place; thus, it is reasonable to review
ASCE

the available approaches prior to reviewing the conclusions drawn


from their application.
Dispersive tsunami models in common use include Boussinesq
models in standard form with higher-order derivatives expressed
and depth-integrated models with nonhydrostatic pressure identied and determined outside of the horizontal momentum equations.
In addition, the tsunami community has long used a strategy of controlling model dispersion through manipulation of model truncation
errors.
Boussinesq models were initially brought into the arena of tsunami propagation from their previous use as short wave models for
wind-generated surface waves. The models were thus typically in
Cartesian coordinate formulations, and could be applied directly to
problems over moderate distances, corresponding to the type of distances in which Universal Transverse Mercator (UTM) projections
of position would not become dramatically in error compared with
the correct geographical coordinates. Examples of applications of
the original version of FUNWAVE (Kennedy et al. 2000; Chen
et al. 2000) include studies of source mechanisms and initial propagation (Grilli et al. 2007) as well as subsequent inundation of the
coast of Thailand (Ioualalen et al. 2007) during the December 2004
Indian Ocean tsunami. Tsunami events were simulated using a
quasi-time-dependent source approach, in which the ground motion
is divided into segments, with each segment being introduced as a
static displacement using the formulation of Okada (1985) but
lagged in time to represent the progression of fault slip along the
trench axis. Zhou and Teng (2010) further extended the Cartesian
Boussinesq approach by developing a 2D version of the original O
( m 4) model of Gobbi et al. (2000). Alternately, Yamazaki et al.
(2009) developed a so-called depth-integrated nonhydrostatic

03116005-17

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 14. (Color) (a) Low-pass eld drifter trajectories, displayed according to their release locations (red dots) at 1, 2, and 3, respectively; (b) absolute
dispersion A2(t) for particles seeded inside the surf zone at the same locations in the two models, with eld observations (thick black line), Boussinesq
u (black line), Boussinesq u corrected for wave drift (blue line), and Delft3D u corrected for wave drift (red line) (reprinted from Geiman et al. 2011,
with permission)

following the previous discussion, with p2 evaluated directly in


terms of a simple, linear-overdepth model for w.
For larger propagation distances, accurate simulations depended
on the development of models in geographical coordinates tted to
the Earths surface. These models can be derived from the full 3D
models in spherical coordinates (see, for example, Section 6.2 in
Pedlosky 1979) using the same nondimensionalization in terms of
depth and wavelength. Resulting formulations in classical
Boussinesq form are described in Pedersen and Lovholt (2008),
Lovholt et al. (2008), Kirby et al. (2013), and Zhou et al. (2011). An
ASCE

interesting twist in the scaling analysis arises in the development of


the Rossby number characterizing the importance of Coriolis
effects. Kirby et al. (2013) found that Coriolis effects scale with the
dimensionless parameter e = m , where e h0 =r0 or ocean depth/
earth radius. For cases in which dispersion is strong enough to be
apparent in the scaling, Coriolis effects should inversely become
vanishingly small, indicating that including them in the model formulation represents retention of an inconsistently small effect. The
choice of retaining or deleting the effect is relatively inconsequential, however, for waves with periods in the tsunami range. Finally,

03116005-18

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

Fig. 15. (Color) Simulations of the 2011 Tohoku-oki tsunami: comparison between measured surface elevations at DART buoys (black lines) and full model
simulations; the buoy numbers and lead-in model arrival times are (a) 21418, 0 min, (b) 51407, 5 min, (c) 46404, 6 min, and (d) 32411, 10 min; (e)(h)
modeled surface elevations at DART buoys [times are as indicated in panels (a)(d), respectively] [Note: Model results are offset by the indicated shift
to facilitate waveform comparisons; modeled surface elevations at DART buoys; full model (blue lines), no dispersion (red lines), no Coriolis (blue
dashed lines), and no dispersion/Coriolis (green dashed lines) (reprinted from Ocean Modelling, Vol. 62, Kirby et al., Dispersive tsunami waves in the
ocean: Model equations and sensitivity to dispersion and Coriolis effects, pp. 3955, Copyright 2013, with permission from Elsevier)]

ASCE

03116005-19

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

propagate across the ocean at speeds that are slightly slower than predicted by shallow water theory or Boussinesq theory. This effect was
illustrated in Fig. 15, which reports a progressive increase in observed
lag between Boussinesq model results and measured data with distance from the tsunami source. This indication of an additional phase
speed decrease relative to the Boussinesq results has been traced by
several authors (Allgeyer and Cummins 2014; Tsai et al. 2013) to the
combined effects of water compressibility and earth elasticity. Wang
(2015) has developed a model correction that adjusts still-water depth
slightly to account for these effects, but the effects can and should be
accounted for directly in the model physics.
Run-Up and Inundation

Fig. 16. (Color) Percentage change in maximum wave height envelope


for the Tohoku-oki tsunami for (a) simulations with and without dispersion and (b) simulations with and without Coriolis (reprinted from Ocean
Modelling, Vol. 62, Kirby et al., Dispersive tsunami waves in the ocean:
Model equations and sensitivity to dispersion and Coriolis effects, pp.
3955, Copyright 2013, with permission from Elsevier)

following on the corresponding Cartesian coordinate development,


Yamazaki et al. (2011) and Horrillo et al. (2012) developed a geographical coordinate version of the depth-integrated NHM of
Yamazaki et al. (2009).
Investigations of tsunami propagation conducted using each of
these models has indicated that frequency dispersion effects can
accumulate to a sufcient degree to shift spatial wave patterns and
alter conclusions on the spatial distribution of maximum wave
heights and resulting inundation effects. Kirby et al. (2013) investigated the effects of leaving dispersion and Coriolis force out of simulations of the 2011 Tohoku-oki tsunami. Fig. 15 shows a comparison of full model results to measured response at four DART buoy
locations, and a comparison of full model results to nondispersive
results and results with no Coriolis force, indicating model accuracy
as well as the relative unimportance of Coriolis effects. This conclusion is further borne out in Fig. 16, which shows a synoptic plot of
relative error for the two cases over the entire Pacic basin.
Signicant differences in wave height patterns between BTMs or
NHMs and nondispersive shallow water models have also been
demonstrated by Horrillo et al. (2012) and Zhou et al. (2014) for the
2011 Tohoku-oki event and Zhou et al. (2012) for the 2009
American Samoa event. Glimsdal et al. (2013) derived a parameter
t referred to as the dispersion time, which provides an indication of
whether dispersion is expected to be important in an event.
Recent comparisons between highly resolved tsunami propagation
models and measured data have further shown that tsunamis
ASCE

Accurate description of run-up and inundation during tsunami


events depends on correct treatment of moving shorelines and conservation of volume during run-up and rundown. This has long been
a difcult topic for depth-integrated models, with most shoreline
treatments in nite-difference codes having evolved to serve practical goals rather than as a consequence of mathematical structure of
the model itself. An early method that was widely used was the socalled slot method, in which the bed within a grid cell is replaced
by a bed with a narrow, deep slot or channel excavated in the grid
cell and oriented in a cross-shore direction (Madsen et al. 1997;
Kennedy et al. 2000; Chen et al. 2000). The channel bottom elevation is chosen such that it is deep enough to always be below the
water surface elevation during maximum rundown. With slots
placed in all initially dry grid cells, the entire computational domain
remains wet, at least at the subgrid level of the slots; thus, the problem of determining the shoreline location as a wet/dry boundary is
irrelevant. Spatially integrating over the slots, which are effectively
resolved bathymetry at subgrid resolution, provides the needed
feedback to the governing equations at larger scale. The slot geometry and accompanying modication to bed elevation outside the slots
could be carefully chosen to provide the same cross-sectional area to
an approaching wave and thus conserve mass without a change in
water surface elevation. However, the approach, as implemented in
the centered nite-difference scheme of FUNWAVE (Kennedy
et al. 2000; Chen et al. 2000), was noisy, and end-user versions of
the code often had slots widened to a degree that interfered with the
processes of shoaling and breaking.
An alternate approach, also based on an ad hoc treatment, was
proposed by Lynett et al. (2002). In this approach, the shoreline is
identied by extrapolating landward from water elevations at the
shallowest wet grid points using a linear t. The apparent intersection of the extrapolated surface with a reconstruction of the bed
within the dry grid cell closest to shore would determine whether
the cell in question would become wet or stay dry, with the process
reversed during rundown. The approach as developed worked in a
stable fashion but led to mass conservation problems.
The shift to the use of nite-volume schemes, which came to
dominate the Boussinesq world in the rst decade of this century,
largely solved the shoreline wetting-drying and inundation problem. The Godunov-style schemes used now in most codes use an
approximate solution of the Riemann problem to reconstruct velocities at cell boundaries. The information carried by the Riemann
variables in these solutions is sufcient to describe the motion of a
run-up tip completely in the context of the solution of the NLSWEs
by the method of characteristics. The strength of this approach has
led to extremely robust shallow water models for the description of
events such as ood waves in rivers or ooding events resulting
from dam breaks (George 2011). This modeling technology has basically been carried over into Boussinesq models as a happy byproduct of selection of the nite-volume schemes. The robustness

03116005-20

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 17. Sequence of snapshots of the solitary wave run-up event: (a) t = 6.4 s; (b) t = 8.4 s; (c) t = 14.4 s [Note: Experimental conguration of Lynett et al.
(2010b); model results were computed using FUNWAVE-TVD (reprinted from Ocean Modelling, Vol. 43, Shi et al., A high-order adaptive time-stepping
TVD solver for Boussinesq modeling of breaking waves and coastal inundation, pp. 3651, Copyright 2012, with permission from Elsevier)]

of this approach motivated the redevelopment of the FUNWAVE


code as FUNWAVE-TVD (Shi et al. 2012), when it had been
shown that the original FUNWAVE code had failed in a test of runup and rundown in comparison to data provided by the experiment
of Lynett et al. (2010b), in which a solitary wave runs up over a
complex topography consisting of an elevated mound located at the
promontory of a V-shaped coastal bathymetry. Run-up over this bathymetry is illustrated in Fig. 17.
Inundation modeling and mapping for tsunami hazard assessment is often performed using bare earth digital elevation models
(DEMs) at 10 to 30-m horizontal resolution (Grilli et al. 2015).
The capabilities of either BTM or NLSWE models extend far
beyond this level of capability, however, it is, in principle, possible now to compute inundation events at resolutions that would
allow the study of loading, scour, and potential collapse of individual structures. Park et al. (2013) described a detailed laboratory and numerical study of tsunami inundation through a model
of a built environment, based on the coastal city of Seaside,
Oregon. Fig. 18 shows the instrument layout in the model and the
built infrastructures location in the laboratory facility. Fig. 19
shows a comparison of measured ow depth, velocity, and derived
momentum ux Hu2 with numerical results obtained using
FUNWAVE-TVD. Existing models are capable of providing highquality predictions of ow patterns in complex environments and
should greatly aid in developing more nuanced hazard assessments
for use by emergency management agencies.
ASCE

NHM Paradigm: Beyond Boussinesq Modeling?


The development of higher-order Boussinesq models and their
extension to cover a more complete set of physics leads to a rapid
increase in model complexity, as evidenced by the rst simple
extension from O ( m 2) to O ( m 4) models for ideal uids (Gobbi
et al. 2000). In addition, the vertical structure of physical effects
such as boundary-layer turbulence or the prole of transported scalars may not be simple to parameterize in regions of more rapidly
varied ow, which poses a possible limit to the potential accuracy
of depth-integrated equations. As a result, there has been a great
deal of recent interest in the development of a new category of freesurface models that retain the relative efciency of depth-integrated
solvers as well as providing access to a fully 3D determination of
ow characteristics when needed. The 3D modeling can take on a
wide range of meanings, with the high-resolution extreme involving
detailed modeling of turbulent ows and complex water surfaces
using volume of uid (VOF) or level set models on xed grids,
meshless Lagrangian particle methods such as smooth particle
hydrodynamics (SPH), or a range of other modern candidates. At
the opposite extreme, there is an increasing desire to allow for arbitrary vertical ow structure and large vertical acceleration effects
but without the need to obtain or use horizontally varying spatial information at scales that are any ner than already provided by the
Boussinesq model being replaced. This latter class of model has
come to be referred to as the NHM.

03116005-21

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 18. (Color) Modeled coastal community in Seaside, Oregon, with instrument locations: (a) layout of wave basin; (b) close-up of instrument layout in modeled area (reprinted from Park et al. 2013, with permission)

The development of the nonhydrostatic modeling approach is


proceeding rapidly and is deserving of its own review, and only a
few main points are touched on here. Work in this direction was initiated mainly by Casulli and Stelling (1998) and Casulli (1999), with
subsequent contributions from a number of research groups (Stelling
and Zijlema 2003; Yuan and Wu 2004; Bradford 2005; Young et al.
ASCE

2007; Zijlema and Stelling 2008; Bradford 2011). At least two readily accessible open source models have been developed: SWASH
(Zijlema et al. 2011) and NHWAVE (Ma et al. 2012).
An alternate approach to the general Navier-Stokes-like solution
strategy used in most NHMs is proposed in Antuono and Brocchini
(2013), in which the continuity and horizontal momentum equations

03116005-22

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Flow Depth
0.35

Flow Depth (m)

0.3
0.25
0.2
0.15
0.1
0.05
0
20

22

24

26

28

30

32

34

36

38

40

34

36

38

40

Cross-Shore Velocity
Cross-Shore Velocity (m/s)

3
2.5
2
1.5
1
0.5
0
20

22

24

26

28

30

32

Time (sec)

Specific Momentum Flux (m 3 /s 2 )

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

Time (sec)

Cross-Shore Momentum Flux


1

Simulated
Measured

0.8
0.6
0.4
0.2
0
20

22

24

26

28

30

32

34

36

38

40

Time (sec)

Fig. 19. (Color) FUNWAVE-TVD results for ow depth, velocity, and momentum ux at measurement location A1 (data from Park et al. 2013)

are integrated over depth, whereas information about the nonhydrostatic component of the pressure eld leads to a Poissons equation
for vertical velocity w, which is then solved to obtain the 3D ow
eld for use in the integrated 2D continuity and momentum equations. The proposed method has not been implemented numerically
to date but is potentially more efcient than the direct approach to
the 3D problem used to date.
Nonhydrostatic modeling provides a robust means for computing a wide range of phenomena in the coastal ocean. The models
gain their efciency from a combination of several factors. First, the
acceptance of the idea that the surface will not be reproduced at
higher resolution than used in Boussinesq models leads to the adoption of a locally smooth, single-valued representation. This allows
the mass conservation equation to be integrated over depth to specify the surface location in terms of the divergence of volume ux, as
in Eq. (22). The approach differs from Boussinesq theory, however,
in retaining the 3D momentum equations. These are typically
solved using a split-step algorithm, as in other treatments of NavierStokes equations; in the present case, it is convenient to take the rst
step to represent solution of the hydrostatic problem, after which a
Poisson problem for the nonhydrostatic pressure correction is
solved, followed by a second step, which updates the velocity eld
in response to the nonhydrostatic pressure correction. Several additional details that can help (and are not universally adopted) include
mapping the vertical coordinate onto a xed strip using a s coordinate transformation (xing the location of both bottom and surface
in the computational grid), and using a Keller-type variable stencil
with pressure specied at the upper face of grid cells, placing the
specication of the surface pressure boundary condition exactly at
the surface position. The resulting equations are usually solved
robustly using the same approaches that work for 2D NLSWE or
Boussinesq models. For the case of surface wave propagation in
ASCE

homogeneous inviscid uids, model results usually show good


agreement with experimental results using as few as three or four
discretization levels in the vertical (Ma et al. 2012). The most timeconsuming aspect of the numerics is the solution of Poissons
equation, which makes the model more expensive to run than
Boussinesq-type models, even in the ideal uid case. However, available Poisson solvers (such as HYPRE , which is used in NHWAVE)
are highly developed and scale well in multiprocessor, distributed
memory environments. As a result, the numerical development of
this class of model, which lacks the higher derivative terms common
to all Boussinesq models, is actually simpler and more transparent
than the Boussinesq approach. The models are easily extended to
incorporate additional physical effects by the addition of scalar transport equations and adjustment of surface and body force expressions.
NHWAVE has been extended to account for suspended sediment
load (Ma et al. 2014), interaction with a submerged plant canopy
(Ma et al. 2013b), and landslide tsunami generation by dispersed
gravity ows (Ma et al. 2013a) and layer-averaged granular debris
ows (Ma et al. 2015), among others. For the case of motions with
baroclinic effects in the form of density stratication, or signicant
vertical variations in obstacles such as plant canopies, it is usually
more desirable to retain a greater degree of vertical resolution to
obtain an accurate description of the velocity eld. However, a comparable increase in resolution in the solution of the pressure eld can
impose an excessive computational burden. Strategies have been
developed in which the nonhydrostatic component of the pressure
eld is solved for on a grid with much lower vertical resolution and
then interpolated back to ner resolution for use in the momentum
equations (van Reeuwijk 2002; Shi et al. 2015). Fig. 20 shows an
example of a baroclinic lock exchange problem in which interfacial
shear, developed as heavier water ows under lighter water, leads to
large Kelvin-Helmholtz billows. Four cases are shown, with Fig.

03116005-23

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 20. (Color) Nonhydrostatic simulations of the lock exchange problem with (a) velocity and pressure solved at 20 vertical levels, (b) velocity and
pressure solved at 200 and 20 levels, respectively, (c) velocity and pressure each solved at 200 vertical levels, and (d) the hydrostatic case (reprinted
from Ocean Modelling, Vol. 96, Shi et al., Pressure Decimation and Interpolation (PDI) method for a baroclinic non-hydrostatic model, pp. 265
279, Copyright 2015, with permission from Elsevier)

20(c) representing a best solution with 200 vertical levels used in


both the velocity and pressure elds. In contrast, Fig. 20(a) shows
the failure of the calculation for the same initial conditions using 20
vertical levels for both velocity and pressure. The intermediate panel
[Fig. 20(b)] shows the effect of the decimation and interpolation
strategy in NHWAVE, with the pressure eld being solved on 20
vertical levels and the velocity eld being solved on 200, yielding
accurate results. Given that most of the computational burden resides
in the pressure solution, the additional overhead in going from 20 to
200 levels in the velocity calculation is actually fairly minimal (Shi
et al. 2015). Strategies involving decimation of the nonhydrostatic
pressure solution are thus very promising, but there needs to be further effort to develop an understanding of when the strategy is
expected to work and what level of decimation is allowable.
Wave breaking in NHMs may occur naturally as a consequence
of the combined physics and numerics, with different conclusions
being reached by different groups. Smit et al. (2013) described the
need to introduce the hybrid effect of turning off nonhydrostatic
corrections to achieve accurate breaking in SWASH, unless vertical
grid resolution is increased signicantly, whereas Ma et al. (2012)
and, more recently, Derakhti et al. (M. Derakhti, J. T. Kirby, F. Shi,
and G. Ma, Wave breaking in the surf zone and deep water in a
non-hydrostatic RANS model. Part 1: Organized wave motions
and Wave breaking in the surf zone and deep water in a nonhydrostatic RANS model. Part 2: Turbulence and mean circulation, submitted, Ocean Modelling, Amsterdam, Netherlands)
show that NHWAVEs handling of breaking does not require either an increase of resolution or adoption of a hybrid model strategy. The fact that different models show variations in behavior in
the breaking process indicates that there is still an incomplete
understanding of the interaction between physics and numerics in
the neighborhood of rapid ow changes in the different models.
For models that use Godunov-type nite-volume schemes, for
ASCE

example, the role of numerical dissipation introduced by limiters


near rapid ow changes as a component of the total dissipation is
signicant and can even be dominant. A potential approach to this
problem is identied next as future work.

Future Avenues
The growing importance of the class of NHMs is a given, and it is clear
that this approach will take precedence in a number of coastal modeling
areas as access to larger multiprocessor environments becomes more
prevalent. However, Boussinesq modeling, taken to represent the general class of 2D depth-integrated modeling strategies, should remain a
preferred method in practice for any type of ow for which a simplied
model of vertical ow structure is useful and adequate. The following
are several areas in which developments are likely in the near future.
(This list is by no means exclusive or complete.)
Treatment of the Nonhydrostatic Pressure Correction
As pointed out by Castro-Orgaz et al. (2015) and discussed earlier,
there is a great deal of correspondence between typical Boussinesq
formulations in which nonhydrostatic pressure corrections are
expressed in terms of derivatives of the horizontal velocity eld and
models in which the localization of the correction is maintained in
the horizontal momentum equations and then resolved as a separate
step, as in the work of Yamazaki et al. (2009) on water waves or
treatments of landslide motions over steeper slopes, such as
Denlinger and Iverson (2004). The second choice has computational advantages because it does not stray outside the scope of
well-developed treatments for the hydrostatic NLSWEs, whereas
the rst choice introduces higher-order derivatives in the formulation and requires the use of additional diagonal matrix solvers. The
schemes that have been developed for the rst choice have also

03116005-24

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

evolved to their own degree of robustness; however, and it is not


clear whether one choice or the other is generally better for conventional numerical approaches. This topic should see some rapid evolution over in the future.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

Development of Graphical Processing


UnitBased Schemes
The potential for signicant reduction in model execution time
provided by graphical processing unit (GPU) hardware automatically leads to what will certainly be an avenue for rapid development in numerical approaches. Recent generations of general purpose GPU (GPGPU) processors now provide architectures with
enough memory and word length to handle complex scientic
computations accurately. Successful implementations of solvers
for the NLSWEs on GPUs are now numerous and range from the
application of conventional nite-volume schemes (starting with
Kuo et al. 2011) to lattice Boltzmann (LBM) schemes, which
have also become well developed in this application (Zhou 2003).
Janen et al. (2012) showed a recent example of a LBM solution
of the NLSWEs for wave run-up on complex bathymetry. The
methods developed include the provision for implementing bottom stress terms and other effects as applied body forces. This
provision could accommodate a second-order pressure correction
as discussed previously, giving a pathway to performing
Boussinesq/LBM calculations in the highly efcient GPGPU
computational framework.
Mixtures and Transport
The generalization of the Boussinesq model to incorporate turbulent
mixing and transport (Kim and Lynett 2011, 2013; Son and Lynett
2014b) will certainly be extended to incorporate a suspended sediment load as part of the mixture formulation in the near future. A
recent example is the work of Kim (2015). Given the long timescale
for coastal storms relative to individual wave periods, the problem
of estimating regional morphology change in response to a hurricane or similar event will remain numerically challenging for some
time. However, the shorter duration of tsunami events relative to
individual wave periods within the event makes the computation of
coastal response to a tsunami a realistic prospect (Tehranirad et al.
2015). This approach should be further extended to incorporate stochastic representations of larger debris elements. Resulting models
could be quite useful in the application of modeling tsunami inundation in which entrained debris from already destroyed infrastructure
would, in turn, lead to a potentially more accurate calculation of
subsequent loading on unfailed structures resulting from the mixture of uid and suspended solid elements.
Wave Breaking
Treatment of wave breaking is possibly still one of the more ad hoc
aspects of Boussinesq or NHM development. A variety of problems
remain. First, there is still very little concrete information available
on the interplay between physical and numerical dissipation in the
determination of energy ux decay rates. In particular, for the case
of 3D NHMs, the overall formulation does not provide a constraint
on the amount of energy lost across a breaking front, as it does in
mass and momentum-conserving, depth-integrated shallow ows.
The role of truncation errors in differencing schemes, as well as the
choice of limiters used in reconstructing uxes, in determining a
numerically imposed dissipation rate, is poorly understood, and the
problem needs considerably more study.
ASCE

Second, the models considered here do not provide any intrinsic


information on the details of stress at the water surface or structure
of the turbulent roller region, or established front face, of a breaking
wave. These errors lead to difculties in obtaining accurate descriptions of mean ows and turbulence levels, which, again, are often
handled by ad hoc treatments. The assumption that the water surface
captured in a Boussinesq or NHM is essentially a locally smooth,
single-valued function of horizontal position implies the a priori use
of an averaging lter to remove unresolved detail in both surface geometry and turbulence in the intermittent air-water region. This
averaging process is basically never addressed in the historical development of either model framework. Brocchini and Peregrine
(2001) and Brocchini (2002) have addressed this question and have
shown how the averaging process can lead to the imposition of a
surface stress at the averaged (or smoothed) surface location as well
as providing a boundary condition on the level of turbulent kinetic
energy. The use of these turbulent boundary conditions, resulting
from averaging over unresolved scales, could greatly improve the
understanding of physical versus numerical dissipation in this class
of models, and, indeed, could move a great deal of the apparent dissipation back into the physical realm, but the conditions are difcult
to apply in practice and have not often been used. It is important
that this aspect of the problem be tackled for both the 2D and 3D
models considered here.

Conclusions
Boussinesq models represent a highly evolved means for computing
free-surface ows in cases in which motion is dominated by the
external, or barotropic, gravity forcing, and in which the vertical
distribution of internal ow properties can be readily parameterized
for the purposes of integrating over depth. In such cases,
Boussinesq or Boussinesq-type models represent an indispensable
tool for the calculation of waves and wave-driven processes ranging
in scale from short wind waves in the nearshore to ocean basin
scale propagation of tsunamis. The framework is presently being
extended to cover a range of problems that have previously been
generally treated by models in the long wave limit, such as granular
debris ows (Denlinger and Iverson 2004; Castro-Orgaz et al.
2015) and other versions of complex natural ow phenomena.
In cases in which ow elds depend on baroclinic response, or
in which vertical distributions of either transported material or
applied body forces (such as vertically varying plant canopy density) are strongly spatially variable and thus not easily parameterized for vertical integration, it is not clear that there is an avenue forward for the Boussinesq approach. Given the availability of general
3D NHM formulations such as NHWAVE (Ma et al. 2012) and
SWASH (Zijlema et al. 2011), it seems more logical to use a modeling framework that allows one to replicate the spatial complexity of
the domain from the outset. This choice will certainly not be dominant in the near future, because the relative efciency of Boussinesq
models in comparison to fully 3D codes will still be a deciding factor for model choice for applications that are adequately described
by the depth-integrated model framework.

Acknowledgments
The author thanks the Ofce of Naval Research, the Army
Research Ofce, the National Science Foundation, and the
Delaware Sea Grant Program for their support of his efforts in this
area over the years. The work discussed here includes the efforts
of a number of present and former students, postdocs, and

03116005-25

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

colleagues at the University of Delaware, and the author extends


thanks to all for providing an exciting and productive environment
over the years. Present work on FUNWAVE-TVD and NHWAVE
is funded by National Science Foundation grants OCE-1334325,
OCE-1435147, and CMMI-1537232 and the National Tsunami
Hazard Mitigation Program.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

References
Abbott, M. B., Petersen, H. M., and Skovgaard, O. (1978). On the numerical modelling of short waves in shallow water. J. Hydraul. Res., 16(3),
173204.
Agnon, Y., Madsen, P. A., and Schffer, H. A. (1999). A new approach to
high-order Boussinesq models. J. Fluid Mech., 399, 319333.
Allgeyer, S., and Cummins, P. (2014). Numerical tsunami simulation
including elastic loading and seawater density stratication. Geophys.
Res. Lett., 41(7), 23682375.
Antuono, M., and Brocchini, M. (2013). Beyond Boussinesq-type equations: Semi-integrated models for coastal dynamics. Phys. Fluids, 25,
016603.
Bai, Y., and Cheung, K. F. (2013). Dispersion and nonlinearity of multilayer non-hydrostatic free-surface ow. J. Fluid Mech., 726, 226260.
Benjamin, T. B. (1966). Internal waves of nite amplitude and permanent
form. J. Fluid Mech., 25(02), 241270.
Boussinesq, J. (1872). Theorie des ondes et des remous qui se propagent le
long dun canal rectangulaire horizontal, en communiquant au liquide
contenu dans ce canal des vitesses sensiblement pareilles de la surface
au fond. J. Mathematiques Pures et Appliquees. Deuxime Serie, 17,
55108 (in French).
Bradford, S. (2005). Godunov-based model for nonhydrostatic wave dynamics. J. Waterway, Port, Coastal, Ocean Eng., 10.1061/(ASCE)0733
-950X(2005)131:5(226), 226238.
Bradford, S. (2011). Nonhydrostatic model for surf zone simulation. J.
Waterway, Port, Coastal, Ocean Eng., 10.1061/(ASCE)WW.1943
-5460.0000079, 163174.
Brocchini, M. (2002). Free surface boundary conditions at a bubbly/
weakly splashing air-water interface. Phys. Fluids, 14(6), 18341840.
Brocchini, M. (2013). A reasoned overview on Boussinesq-type models:
The interplay between physics, mathematics and numerics. Proc. R.
Soc. London, Ser. A, 469(2160), 20130496.
Brocchini, M., and Peregrine, D. H. (2001). The dynamics of strong turbulence at free surfaces. Part 2. Free-surface boundary conditions. J.
Fluid Mech., 449, 255290.
Castro-Orgaz, O., Hutter, K., Giraldez, J. V., and Hager, W. H. (2015).
Nonhydrostatic granular ow over 3-D terrain: New Boussinesq-type
gravity waves? J. Geophys. Res. Earth Surf., 120(1), 128.
Casulli, V. (1999). A semi-implicit nite difference method for nonhydrostatic, free-surface ows. Int. J. Numer. Methods Fluids, 30(4),
425440.
Casulli, V., and Stelling, G. (1998). Numerical simulation of 3D quasihydrostatic, free surface ows. J. Hydraul. Eng., 10.1061/(ASCE)0733
-9429(1998)124:7(678), 678686.
Chazel, F., Benoit, M., Ern, A., and Piperno, S. (2009). A double-layer
Boussinesq-type model for highly nonlinear and dispersive waves.
Proc. R. Soc. London, Ser. A, 465(2108), 23192346.
Chen, Q. (2006). Fully nonlinear Boussinesq-type equations for waves and
currents over porous beds. J. Eng. Mech., 10.1061/(ASCE)0733
-9399(2006)132:2(220), 220230.
Chen, Q., Dalrymple, R. A., Kirby, J. T., Kennedy, A. B., and Haller, M. C.
(1999). Boussinesq modeling of a rip current system. J. Geophys. Res.,
104(C9), 2061720637.
Chen, Q., Kaihatu, J., and Hwang, P. (2004). Incorporation of wind effects
into Boussinesq wave models. J. Waterway, Port, Coastal, Ocean
Eng., 10.1061/(ASCE)0733-950X(2004)130:6(312), 312321.
Chen, Q., Kirby, J., Dalrymple, R., Kennedy, A., and Chawla, A. (2000).
Boussinesq modeling of wave transformation, breaking and runup. II:
2D. J. Waterway, Port, Coastal, Ocean Eng., 10.1061/(ASCE)0733
-950X(2000)126:1(48), 4856.
ASCE

Chen, Q., Kirby, J. T., Dalrymple, R. A., Shi, F., and Thornton, E. B.
(2003). Boussinesq modeling of longshore currents. J. Geophys. Res.,
108(C11), 3362.
Choi, J., Kirby, J. T., and Yoon, S. B. (2015a). Boussinesq modeling of
longshore currents in the SandyDuck experiment under directional random wave conditions. Coastal Eng., 101, 1734.
Choi, J., Kirby, J. T., and Yoon, S. B. (2015b). Reply to Discussion to
Boussinesq modeling of longshore currents in the Sandy Duck experiment under directional random wave conditions by J. Choi, J. T. Kirby
and S. B. Yoon. Coastal Eng., 106, 46.
Choi, W., and Camassa, R. (1999). Fully nonlinear internal waves in a
two-uid system. J. Fluid Mech., 396, 136.
Cienfuegos, R., Barthelemy, E., and Bonneton, P. (2006). A fourth-order
compact nite volume scheme for fully nonlinear and weakly dispersive
Boussinesq-type equations. Part I: Model development and analysis.
Int. J. Numer. Methods Fluids, 51(11), 12171253.
Cienfuegos, R., Barthelemy, E., and Bonneton, P. (2010). Wave-breaking
model for Boussinesq-type equations including roller effects in the mass
conservation equation. J. Waterway, Port, Coastal, Ocean Eng., 10
.1061/(ASCE)WW.1943-5460.0000022, 1026.
Clark, D. B., Elgar, S., and Raubenheimer, B. (2012). Vorticity generation
by short-crested wave breaking. Geophys. Res. Lett., 39(24), L24604.
Clark, D. B., Feddersen, F., and Guza, R. T. (2011). Modeling surf zone
tracer plumes: 2. Transport and dispersion. J. Geophys. Res.,
116(C11) C11028.
Cox, D. T., Kobayashi, N., and Okayasu, A. (1995). Experimental and numerical modeling of surf zone hydrodynamics. Rep. No. CACR-95-07,
Center for Applied Coastal Research, Dept. of Civil and Environmental
Engineering, Univ. of Delaware, Newark, DE.
DAlessandro, F., and Tomasicchio, G. R. (2008). The BCI criterion for
the initiation of breaking process in Boussinesq-type equations wave
models. Coastal Eng., 55, 11741184.
Debsarma, S., Das, K. P., and Kirby, J. T. (2010). Fully nonlinear higherorder model equations for long internal waves in a two-uid system. J.
Fluid Mech., 654, 281303.
Denlinger, R. P., and Iverson, R. M. (2004). Granular avalanches across
irregular three-dimensional terrain: 1. Theory and computation. J.
Geophys. Res, 109(F1), F01014.
Donahue, A. S., Zhang, Y., Kennedy, A. B., Westerink, J. J., Panda, N., and
Dawson, C. (2015). A Boussinesq-scaled, pressure-Poisson water
wave model. Ocean Modell., 86, 3657.
Douglass, S. (1990). Inuence of wind on breaking waves. J. Waterway,
Port, Coastal, Ocean Eng., 10.1061/(ASCE)0733-950X(1990)116:
6(651), 651663.
Erduran, K. S., Ilic, S., and Kutija, V. (2005). Hybrid nite-volume nitedifference scheme for the solution of Boussinesq equations. Int. J.
Numer. Methods Fluids, 49(11), 12131232.
Feddersen, F. (2014). The generation of surfzone eddies in a strong alongshore current. J. Phys. Oceanogr., 44(2), 600617.
Feddersen, F., Clark, D. B., and Guza, R. T. (2011). Modeling surf zone
tracer plumes: 1. Waves, mean currents, and low-frequency eddies. J.
Geophys. Res., 116(C11), C11027.
Feddersen, F., and Veron, F. (2005). Wind effects on shoaling wave
shape. J. Phys. Oceanogr., 35, 12231228.
Fuhrman, D. R., and Madsen, P. A. (2009). Tsunami generation, propagation, and run-up with a high-order Boussinesq model. Coastal Eng.,
56(7), 747758.
Fuhrman, D. R., Madsen, P. A., and Bingham, H. B. (2004). A numerical
study of crescent waves. J. Fluid Mech., 513, 309341.
Fuhrman, D. R., Madsen, P. A., and Bingham, H. B. (2006). Numerical
simulation of lowest-order short-crested wave instabilities. J. Fluid
Mech., 563, 415441.
Galan, A., Simarro, G., Orla, A., Simarro, J., and Liu, P. (2012). Fully
nonlinear model for water wave propagation from deep to shallow
waters. J. Waterway, Port, Coastal, Ocean Eng., 10.1061/(ASCE)WW
.1943-5460.0000143, 362371.
Geiman, J. D., Kirby, J. T., Reniers, A. J. H. M., and MacMahan, J. H.
(2011). Effects of wave averaging on estimates of uid mixing in the
surf zone. J. Geophys. Res., 116(C4), C04006.

03116005-26

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

Geist, E. L., Lynett, P. J., and Chaytor, J. D. (2009). Hydrodynamic modeling


of tsunamis from the Currituck landslide. Mar. Geol., 264(12), 4152.
George, D. L. (2011). Adaptive nite volume methods with well-balanced
Riemann solvers for modeling oods in rugged terrain: Application to
the Malpasset dam-break ood (France, 1959). Int. J. Numer. Methods
Fluids, 66(8), 10001018.
Glimsdal, S., Pedersen, G., Harbitz, C. B. L., and Lovholt, F. (2013).
Dispersion of tsunamis: does it really matter? Nat. Hazards Earth Syst.
Sci., 13, 15071526.
Gobbi, M. F., and Kirby, J. T. (1999). Wave evolution over submerged
sills: tests of a high-order Boussinesq model. Coastal Eng., 37(1),
5796.
Gobbi, M. F., Kirby, J. T., and Wei, G. (2000). A fully nonlinear
Boussinesq model for surface waves. Part 2. Extension to O(kh)4. J.
Fluid Mech., 405, 181210.
Green, A. E., and Naghdi, P. M. (1976). A derivation of equations for
wave propagation in water of variable depth. J. Fluid Mech., 78(02),
237246.
Grilli, S., Ioualalen, M., Asavanant, J., Shi, F., Kirby, J., and Watts, P.
(2007). Source constraints and model simulation of the December 26,
2004, Indian Ocean tsunami. J. Waterway, Port, Coastal, Ocean Eng.,
10.1061/(ASCE)0733-950X(2007)133:6(414), 414428.
Grilli, S. T., et al. (2015). Modeling of SMF tsunami hazard along the
upper U. S. East Coast: Detailed impact around Ocean City, MD. Nat.
Hazards, 76(2), 705746.
Horrillo, J., Knight, W., and Kowalik, Z. (2012). Tsunami propagation
over the North Pacic: Dispersive and nondispersive models. Sci.
Tsunami Hazards, 31(3), 154177.
HYPRE [Computer software]. Lawrence Livermore National Laboratory,
Livermore, CA.
Ioualalen, M., Asavanant, J., Kaewbanjak, N., Grilli, S. T., Kirby, J. T., and
Watts, P. (2007). Modeling the 26 December 2004 Indian Ocean tsunami:
Case study of impact in Thailand. J. Geophys. Res., 112(C7), C07024.
Janen, C. F., Grilli, S. T., and Krafczyk, M. (2012). Efcient simulations
of long wave propagation and runup using a LBM approach on GPGPU
hardware. Proc., 22nd Int. Offshore and Polar Engineering Conf.,
Rhodes, Greece, International Society of Offshore and Polar Engineers,
Cupertino, CA, 145152.
Jeffreys, H. (1925). On the formation of water waves by wind. Proc. R.
Soc. London, Ser. A, 107(742), 189206.
Johnson, D., and Pattiaratchi, C. (2006). Boussinesq modelling of transient
rip currents. Coastal Eng., 53(56), 419439.
Karambas, T., and Koutitas, C. (2002). Surf and swash zone morphology
evolution induced by nonlinear waves. J. Waterway, Port, Coastal,
Ocean Eng., 10.1061/(ASCE)0733-950X(2002)128:3(102), 102113.
Karambas, T., and Memos, C. (2009). Boussinesq model for weakly nonlinear fully dispersive water waves. J. Waterway, Port, Coastal, Ocean
Eng., 10.1061/(ASCE)0733-950X(2009)135:5(187), 187199.
Kazolea, M., Delis, A. I., Nikolos, I. K., and Synolakis, C. E. (2012). An
unstructured nite volume numerical scheme for extended 2D
Boussinesq-type equations. Coastal Eng., 69, 4266.
Kazolea, M., Delis, A. I., and Synolakis, C. E. (2014). Numerical treatment
of wave breaking on unstructured nite volume approximations for
extended Boussinesq-type equations. J. Comput. Phys., 271, 281305.
Kemp, P. H., and Simons, R. R. (1982). The interaction between waves
and a turbulent current: waves propagating with the current. J. Fluid
Mech., 116, 227250.
Kemp, P. H., and Simons, R. R. (1983). The interaction of waves and a turbulent current: waves propagating against the current. J. Fluid Mech.,
130, 7389.
Kennedy, A., Chen, Q., Kirby, J., and Dalrymple, R. (2000).
Boussinesq modeling of wave transformation, breaking, and runup.
I: 1D. J. Waterway, Port, Coastal, Ocean Eng., 10.1061/(ASCE)0733
-950X(2000)126:1(39), 3947.
Kennedy, A. B., Kirby, J. T., Chen, Q., and Dalrymple, R. A. (2001).
Boussinesq-type equations with improved nonlinear performance.
Wave Motion, 33(3), 225243.
Kennedy, A. B., Kirby, J. T., and Gobbi, M. F. (2002). Simplied higherorder Boussinesq equations I. Linear simplications. Coastal Eng.,
44(3), 205229.
ASCE

Kim, D. H. (2015). H2D morphodynamic model considering wave, current


and sediment interaction. Coastal Eng., 95, 2034.
Kim, D. H., and Lynett, P. J. (2011). Turbulent mixing and passive scalar
transport in shallow ows. Phys. Fluids, 23, 016603.
Kim, D. H., and Lynett, P. J. (2013). A s -coordinate transport model
coupled with rotational Boussinesq-type equations. Environ. Fluid
Mech., 13(1), 5172.
Kim, D. H., Lynett, P. J., and Socolofsky, S. A. (2009). A depth-integrated
model for weakly dispersive, turbulent, and rotational uid ows.
Ocean Modell., 27(34), 198214.
Kirby, J. T. (1988). Weakly nonlinear long waves in varying channels.
Developments in Theoretical and Applied Mechanics, S. Y. Wang,
R. M. Hackett, S. L. Deleeuw, and A. M. Smith, eds., Vol. 14, Proc.,
Southeastern Conf. on Theoretical and Applied Mechanics, Biloxi, MS,
111118.
Kirby, J. T. (1997). Nonlinear, dispersive long waves in water of variable
depth. Gravity waves in water of nite depth, J. N. Hunt, Ed.,
Computational Mechanics Publications, Boston, 55125.
Kirby, J. T. (2003). Boussinesq models and applications to nearshore
wave propagation, surf zone processes and wave-induced currents.
Advances in coastal modeling, V. C. Lakhan, Ed., Elsevier,
Amsterdam, 141.
Kirby, J. T., Briganti, R., Brocchini, M., and Chen, Q. (2006). Lagrangian
particle statistics of numerically simulated shear waves. EOS. Trans.
AGU, Vol. 87, Abstract OS41C0633.
Kirby, J. T., Shi, F., Nicolsky, D., and Misra, S. (2016). The 27 April 1975
Kitimat, British Columbia submarine landslide tsunami: A comparison
of modeling approaches. Landslides, 114.
Kirby, J. T., Shi, F., Tehranirad, B., Harris, J. C., and Grilli, S. T. (2013).
Dispersive tsunami waves in the ocean: Model equations and sensitivity to dispersion and coriolis effects. Ocean Modell., 62, 3955.
Korteweg, D. J., and de Vries, G. (1895). On the change of form of long
waves advancing in a rectangular canal, and on a new type of long stationary waves. Philos. Mag., 39(240), 422443.
Kuo, F. A., Smith, M., Hsieh, C. W., Chou, C. Y., and Wu, G. S. (2011).
GPU acceleration for general conservation equations and its application to several engineering problems. Comput. Fluids, 45(1),
147154.
Lepelletier, T., and Raichlen, F. (1987). Harbor oscillations induced by
nonlinear transient long waves. J. Waterway, Port, Coastal, Ocean
Eng., 10.1061/(ASCE)0733-950X(1987)113:4(381), 381400.
Lesser, G. R., Roelvink, J. A., van Kester, T. M., and Stelling, G. S.
(2004). Development and validation of a three-dimensional morphological model. Coastal Eng., 51: 883915.
Leveque, R. J. (2002). Finite volume methods for hyperbolic problems,
Cambridge University Press, Cambridge, U.K.
Li, N., Roeber, V., Yamazaki, Y., Heitmann, T. W., Bai, Y, and Cheung,
K. F. (2014). Integration of coastal inundation modeling from storm
tides to individual waves. Ocean Modell., 83, 2642.
Liu, K., Chen, Q., and Kaihatu, J. (2016). Modeling wind effects on shallow water waves. J. Waterway, Port, Coastal, Ocean Eng., 10.1061
/(ASCE)WW.1943-5460.0000314, 04015012.
Liu, Z., and Fang, K. (2015). Two-layer Boussinesq models for coastal
water waves. Wave Motion, 57, 88111.
Long, W. (2006). Boussinesq modeling of waves, currents and sediment
transport. Ph.D. thesis, Univ. of Delaware, Newark, DE.
Lovholt, F., Pedersen, G. K., and Gisler, G. (2008). Oceanic propagation
of a potential tsunami from the La Palma Island. J. Geophys. Res., 113,
C09026.
Lu, X., Dong, B., Mao, B., and Zhang, X. (2015). A two-dimensional
depth-integrated non-hydrostatic numerical model for nearshore wave
propagation. Ocean Modell., 96(Part 2), 187202.
Lynett, P. J. (2006). Wave breaking velocity effects in depth-integrated
models. Coastal Eng., 53(4), 325333.
Lynett, P. J. and Liu, P. L. F. (2002a). A numerical study of submarinelandslide-generated waves and run-up. Proc. R. Soc. London, Ser. A,
458, 28852910.
Lynett, P. J., and Liu, P. L. F. (2002b). A two-dimensional, depthintegrated model for internal wave propagation over variable bathymetry. Wave Motion, 36, 221240.

03116005-27

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

Lynett, P. J. and Liu, P. L. F. (2004). A two-layer-approach to wave modelling. Proc. R. Soc. London, Ser. A, 460(2049), 26372669.
Lynett, P. J., and Liu, P. L. F. (2005). A numerical study of the run-up generated by three-dimensional landslides. J. Geophys. Res., 110(C3),
C03006.
Lynett, P. J., Melby, J. A., and Kim, D. H. (2010a). An application of
Boussinesq modeling to hurricane wave overtopping and inundation.
Ocean Eng., 37, 135153.
Lynett, P. J., Swigler, D. T., Son, S., Bryant, D., and Socolofsky, S. (2010b).
Experimental study of solitary wave evolution over a 3d shallow shelf.
Proc., 32nd Int. Conf. on Coastal Engineering, J. M. Smith, and P.
Lynett, eds., Coastal Engineering Research Council, ASCE, Reston, VA.
Lynett, P. J., Wu, T. R., and Liu, P. L. F. (2002). Modeling wave runup
with depth-integrated equations. Coastal Eng., 46(2), 89107.
Ma, G., Chou, Y. J., and Shi, F. (2014). A wave-resolving model for nearshore suspended sediment transport. Ocean Modell., 77, 3349.
Ma, G., Kirby, J. T., Hsu, T. J., and Shi, F. (2015). A two-layer granular
landslide model for tsunami wave generation: Theory and computation. Ocean Modell., 93, 4055.
Ma, G., Kirby, J. T., and Shi, F. (2013a). Numerical simulation of tsunami
waves generated by deformable submarine landslides. Ocean Modell.,
69, 146165.
Ma, G., Kirby, J. T., Su, S. F., Figlus, J., and Shi, F. (2013b). Numerical
study of turbulence and wave damping induced by vegetation canopies.
Coastal Eng., 80, 6878.
Ma, G., Shi, F., and Kirby, J. T. (2012). Shock-capturing non-hydrostatic
model for fully dispersive surface wave processes. Ocean Modell.,
4344, 2235.
Madsen, P. A., and Agnon, Y. (2003). Accuracy and convergence of velocity formulations for water waves in the framework of Boussinesq
theory. J. Fluid Mech., 477, 285319.
Madsen, P. A., Bingham, H. B., and Lu, H. (2002). A new Boussinesq
method for fully nonlinear waves from shallow to deep water. J. Fluid
Mech., 462, 130.
Madsen, P. A., Bingham, H. B., and Schffer, H. A. (2003). Boussinesqtype formulations for fully nonlinear and extremely dispersive water
waves: Derivation and analysis. Proc. R. Soc. London, Ser. A,
459(2033), 10751104.
Madsen, P. A., and Fuhrman, D. R. (2010). Higher-order Boussinesq-type
modelling of nonlinear wave phenomena in deep and shallow water.
Advances in numerical simulation of nonlinear water waves, Q. Ma,
Ed., World Scientic, Hackensack, NJ.
Madsen, P. A., Fuhrman, D. R., and Wang, B. (2006). A Boussinesq-type
method for fully nonlinear waves interacting with a rapidly varying bathymetry. Coastal Eng., 53(56), 487504.
Madsen, P. A., Murray, R., and Sorensen, O. R., (1991). A new form of the
Boussinesq equations with improved linear dispersion characteristics.
Coastal Eng., 15(4), 371388.
Madsen, P. A., and Schffer, H. A. (1999). A review of Boussinesq-type
equations for surface gravity waves. Advances in coastal and ocean engineering, Phillip L. F. Liu, ed., Vol.5, World Scientic, Singapore, 194.
Madsen, P. A., Sorensen, O. R., and Schffer, H. A., (1997). Surf zone dynamics simulated by a Boussinesq type model. Part I. Model description
and cross-shore motion of regular waves. Coastal Eng., 32(4),
255287.
Mase, H., and Kirby, J. T. (1992). Modied frequency domain KdV equation for random wave shoaling. Proc., 23d Int. Conf. Coastal
Engineering, ASCE, Reston, VA, 474487.
McWilliams, J. C., Restrepo, J. M., and Lane, E. M. (2004). An asymptotic
theory for the interaction of waves and currents in coastal waters. J.
Fluid Mech., 511, 135178.
Memos, C., Klonaris, G., and Chondros, M. (2016). On higher-order
Boussinesq-type wave models. J. Waterway, Port, Coastal, Ocean
Eng., 10.1061/(ASCE)WW.1943-5460.0000317, 04015011.
Nwogu, O. (1993). Alternative form of Boussinesq equations for nearshore
wave propagation. J. Waterway, Port, Coastal, Ocean Eng., 10.1061
/(ASCE)0733-950X(1993)119:6(618), 618638.
Okada, Y. (1985). Surface deformation due to shear and tensile faults in a
half-space. Bull. Seismol. Soc. Am., 75(4), 11351154.
ASCE

Okamoto, T., and Basco, D. R. (2006). The relative trough Froude number
for initiation of wave breaking: Theory, experiments and numerical
model conrmation. Coastal Eng., 53(8), 675690.
Panda, N., Dawson, C., Zhang, Y., Kennedy, A. B., Westerink, J. J., and
Donahue, A. S. (2014). Discontinuous Galerkin methods for solving
Boussinesq-Green-Naghdi equations in resolving non-linear and dispersive surface water waves. J. Comput. Phys., 273, 572588.
Park, H., Cox, D. T., Lynett, P. J., Wiebe, D. M., and Shin, S. (2013).
Tsunami inundation modeling in constructed environments: A physical
and numerical comparison of free-surface elevation, velocity and momentum ux. Coastal Eng., 79, 921.
Parsons, T., et al. (2014). Source and progression of a submarine landslide
and tsunami: The 1964 Great Alaska earthquake at Valdez. J. Geophys.
Res. Solid Earth, 119(11), 85028516.
Pedersen, G., and Lovholt, F. (2008). Documentation of a global
Boussinesq solver. Preprint series. Mechanics and applied mathematics, Univ. of Oslo. hhttps://www.duo.uio.no/handle/10852/10184i.
Pedlosky, J. (1979). Geophysical uid dynamics, Springer, New York.
Peregrine, D. H. (1966). Calculations of the development of an undular
bore. J. Fluid Mech., 25(02), 321330.
Peregrine, D. H. (1967). Long waves on a beach. J. Fluid Mech., 27(04),
815827.
Peregrine, D. H. (1998). Surf zone currents. Theor. Comput. Fluid Dyn.,
10(1), 295309.
Pierini, S. (1989). A model for the Alboran Sea internal solitary waves. J.
Phys. Oceanogr., 19(6), 755772.
Proudman, J. (1929). The effects on the sea of changes in atmospheric
pressure. Geophys. J. Int., 2, 197209.
Rego, V. S., Kirby, J. T., and Thompson, D. (2001). Boussinesq waves on
vertically sheared currents. Proc., 4th Int. Symp. on Ocean Wave
Measurement and Analysis, Waves01, ASCE, Reston, VA, 904913.
Roeber, V., and Cheung, K. F. (2012). Boussinesq-type model for energetic
breaking waves in fringing reef environments. Coastal Eng., 70, 120.
Schffer, H. A., and Madsen, P. A. (1995). Further enhancements of
Boussinesq-type equations. Coastal Eng., 26(12), 114.
Schffer, H. A., Madsen, P. A., and Deigaard, R. (1993). A Boussinesq
model for waves breaking in shallow water. Coastal Eng., 20(34),
185202.
Serre, F. (1953). Contribution letude des ecoulements permanents et
variables dans les canaux. Houille Blanche, 8, 374388.
Shen, C. (2001). Constituent Boussinesq equations for waves and currents. J. Phys. Oceanogr., 31, 850859.
Shi, F., Kirby, J. T., Harris, J. C., Geiman, J. D., and Grilli, S. T. (2012). A
high-order adaptive time-stepping TVD solver for Boussinesq modeling
of breaking waves and coastal inundation. Ocean Modell., 4344,
3651.
Shi, J., et al. (2015). Pressure decimation and interpolation (PDI)
method for baroclinic non-hydrostatic model. Ocean Modell., 96,
265279.
Simarro, G. (2013). Energy balance, wave shoaling and group celerity
in Boussinesq-type wave propagation models. Ocean Modell., 72,
7479.
Simarro, G. (2015). Discussion to Boussinesq modeling of longshore currents in the SandyDuck experiment under directional random wave conditions by J. Choi, J. T. Kirby and S. B. Yoon. Coastal Eng., 106,
3031.
Simarro, G., Orla, A., and Galan, A. (2013). Linear shoaling in
Boussinesq-type wave propagation models. Coastal Eng., 80,
100106.
Smit, P., Zijlema, M., and Stelling, G. (2013). Depth-induced wave breaking
in a non-hydrostatic, near-shore wave model. Coastal Eng., 76, 116.
Son, S., and Lynett, P. J. (2014a). Interaction of dispersive water waves
with weakly sheared currents of arbitrary prole. Coastal Eng., 90,
6484.
Son, S., and Lynett, P. J. (2014b). Nonlinear and dispersive free surface
waves propagating over uids with weak vertical and horizontal density
variation. J. Fluid Mech., 748, 221240.
Spydell, M., and Feddersen, F. (2009). Lagrangian drifter dispersion in the
surf zone: Directionally spread, normally incident waves. J. Phys.
Oceanogr., 39, 809830.

03116005-28

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Downloaded from ascelibrary.org by Pontificia Universidad Catolica de Chile (UC) on 10/04/16. Copyright ASCE. For personal use only; all rights reserved.

Stelling, G. S., and Zijlema, M. (2003). An accurate and efcient nitedifference algorithm for non-hydrostatic free-surface ow with application to wave propagation. Int. J. Numer. Methods Fluids, 43(1), 123.
Sullivan, P. P., McWilliams, J. C., and Melville, W. K. (2007). Surface
gravity wave effects in the oceanic boundary layer: large-eddy simulation with vortex force and stochastic breakers. J. Fluid Mech., 593,
405452.
Tehranirad, B., Kirby, J. T., Shi, F., and Grilli, S. T. (2015). Does morphological adjustment during tsunami inundation increase levels of hazard? Proc., Coastal Structures & Solutions to Coastal Disasters Joint
Conf., COPRI, ASCE, Reston, VA, in press.
Tissier, M., Bonneton, P., Marche, F., Chazel, F., and Lannes, D. (2012).
A new approach to handle wave breaking in fully non-linear
Boussinesq models. Coastal Eng., 67, 5466.
Tomasson, G. G., and Melville, W. K. (1992). Geostrophic adjustment in a
channel: nonlinear and dispersive effects. J. Fluid Mech., 241, 2357.
Tonelli, M., and Petti, M. (2009). Hybrid nite volumenite difference
scheme for 2DH improved Boussinesq equations. Coastal Eng.,
56(56), 609620.
Tonelli, M., and Petti, M. (2010). Finite volume scheme for the solution of
2D extended Boussinesq equations in the surf zone. Ocean Eng., 37(7),
567582.
Tonelli, M., and Petti, M. (2011). Simulation of wave breaking over complex bathymetries by a Boussinesq model. J. Hydraul. Res., 49(4),
473486.
Tonelli, M., and Petti, M. (2012). Shock-capturing Boussinesq model for
irregular wave propagation. Coastal Eng., 61, 819.
Toro, E. F. (2001). Shock-capturing methods for free-surface shallow ows,
John Wiley & Sons, New York.
Tsai, V. C., Ampuero, J. P., Kanamori, H., and Stevenson, D. J. (2013).
Estimating the effect of earth elasticity and variable water density on
tsunami speeds. Geophys. Res. Lett., 40(3), 492496.
van Reeuwijk, M. (2002). Efcient simulation of non-hydrostatic free surface
ows. M.Sc. thesis, Delft Univ. of Technology, Delft, Netherlands.
Veeramony, J., and Svendsen, I. A. (2000). The ow in surf zone waves.
Coastal Eng., 39(24), 93122.
Walkley, M., and Berzins, M. (1999). A nite element method for the onedimensional extended Boussinesq equations. Int. J. Numer. Methods
Fluids, 29(2), 143157.
Walkley, M., and Berzins, M. (2002). A nite element method for the twodimensional extended Boussinesq equations. Int. J. Numer. Methods
Fluids, 39(10), 865885.
Wang, D. (2015). An ocean depth-correction method for reducing model
errors in tsunami travel time: Application to the 2010 Chile and 2011
Tohoku tsunamis. Sci. Tsunami Hazards, 34(1), 122.
Wei, G., and Kirby, J. (1995). Time-dependent numerical code for
extended Boussinesq equations. J. Waterway, Port, Coastal, Ocean
Eng., 10.1061/(ASCE)0733-950X(1995)121:5(251), 251261.

ASCE

Wei, G., Kirby, J. T., Grilli, S. T., and Subramanya, R. (1995). A fully nonlinear Boussinesq model for surface waves. part 1. Highly nonlinear
unsteady waves. J. Fluid Mech., 294, 7192.
Xiao, H., Young, Y., and Prevost, J. (2010). Hydro- and morpho-dynamic
modeling of breaking solitary waves over a ne sand beach. Part II:
Numerical simulation. Mar. Geol., 269(34), 119131.
Yamazaki, Y., Cheung, K. F., and Kowalik, Z. (2011). Depth-integrated,
non-hydrostatic model with grid-nesting for tsunami generation,
propagation and run-up. Int. J. Numer. Methods Fluids, 67(12),
20812107.
Yamazaki, Y., Kowalik, Z., and Cheung, K. F. (2009). Depth-integrated,
non-hydrostatic model for wave breaking and run-up. Int. J. Numer.
Methods Fluids, 63, 14481470.
Yoon, S. B., and Liu, P. L. F. (1989). Interactions of currents and
weakly nonlinear water waves in shallow water. J. Fluid Mech.,
205, 397419.
Young, C. C., Wu, C. H., Kuo, J. T., and Liu, W. C. (2007). A higher-order
s -coordinate non-hydrostatic model for nonlinear surface waves.
Ocean Eng., 34(10), 13571370.
Yuan, H., and Wu, C. H. (2004). An implicit three-dimensional fully nonhydrostatic model for free-surface ows. Int. J. Numer. Methods
Fluids, 46(7), 709733.
Zelt, J. A. (1991). The run-up of nonbreaking and breaking solitary
waves. Coastal Eng., 15(3), 205246.
Zhang, Y., Kennedy, A. B., Panda, N., Dawson, C., and Westerink, J. J.
(2013). Boussinesq-Green-Naghdi rotational water wave theory.
Coastal Eng., 73, 1327.
Zhou, H., Moore, C. W., Wei, Y., and Titov, V. V. (2011). A nested-grid
Boussinesq-type approach to modelling dispersive propagation and
runup of landslide-generated tsunamis. Nat. Hazards Earth Syst. Sci.,
11(10), 26772697.
Zhou, H., and Teng, M. (2010). Extended fourth-order depth-integrated
model for water waves and currents generated by submarine landslides. J. Eng. Mech., 10.1061/(ASCE)EM.1943-7889.0000087,
506516.
Zhou, H., Wei, Y., and Titov, V. V. (2012). Dispersive modeling of the
2009 Samoa tsunami. Geophys. Res. Lett., 39(16), L16603.
Zhou, H., Wei, Y., Wright, L., and Titov, V. V. (2014). Waves and currents
in Hawaiian waters induced by the dispersive 2011 Tohoku tsunami.
Pure Appl. Geophys., 171(12), 33653384.
Zhou, J. (2003). Lattice Boltzmann methods for shallow water ows,
Springer, New York.
Zijlema, M., Stelling, G., and Smit, P. (2011). SWASH: An operational
public domain code for simulating wave elds and rapidly varied ows
in coastal waters. Coastal Eng., 58(10), 9921012.
Zijlema, M., and Stelling, G. S. (2008). Efcient computation of surf zone
waves using the nonlinear shallow water equations with non-hydrostatic
pressure. Coastal Eng., 55(10), 780790.

03116005-29

J. Waterway, Port, Coastal, Ocean Eng., 03116005

J. Waterway, Port, Coastal, Ocean Eng.

Das könnte Ihnen auch gefallen