Vous êtes sur la page 1sur 16

Coastal Engineering 56 (2009) 297–312

Contents lists available at ScienceDirect

Coastal Engineering
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / c o a s t a l e n g

Evolution of waves and currents over a submerged laboratory shoal


Junwoo Choi a, Chae Ho Lim b, Jong In Lee c, Sung Bum Yoon d,⁎
a
Department of Civil Eng., Hanyang University, 133-791, South Korea
b
National Oceanographic Research Institute, 400-800, South Korea
c
Water Resources Research Dept., Korea Institute of Construction Tech., 411-712, South Korea
d
Department of Civil & Environ. Eng., Hanyang University, Ansan, 426-791, South Korea

a r t i c l e i n f o a b s t r a c t

Article history: The vertically-integrated effect of interaction between waves and wave-induced currents on wave
Received 5 October 2007 transformation over a submerged elliptic shoal was investigated based on numerical simulations of the
Received in revised form 28 August 2008 Vincent and Briggs experiment [Vincent, C.L., Briggs, M.J., 1989. Refraction- diffraction of irregular waves over a
Accepted 10 September 2008
mound. Journal of Waterway, Port, Coastal and Ocean Engineering, 115(2), pp. 269–284.]. The numerical
Available online 11 October 2008
simulations were performed using two numerical wave-current model systems: one, a combination of the
Keywords:
wave model SWAN and the current model SHORECIRC, and the other, a combination of the wave model REF/DIF
Wave transformation and the same current model. A time-dependent, phase-resolving wave and current model, FUNWAVE, was also
Numerical wave model utilized to simulate the experiment. In the simulations, the developed wave-induced currents defocused waves
Wave-induced current behind the shoal and brought on a wave shadow zone that showed relatively low wave height distributions. For
Wave-current interaction the breaking case of monochromatic waves, the wave heights computed using FUNWAVE showed good
Submerged elliptic shoal agreement with the measurements and the resulting wave-induced currents showed a jet-like velocity
distribution in transverse direction. And the computed results of the two model combinations agreed better
with the measurements than the computed results obtained by neglecting wave-current interaction. However,
it was found that for the case in which transverse interference pattern caused by refracted waves was strong,
REF/DIF-SHORECIRC did not correctly evaluate radiation stresses, the gradients of which generate wave-
induced currents. SWAN-SHORECIRC, which cannot deal with the interference patterns, predicted a jet-like
wave-induced current. For breaking random wave cases, the computed results of the two model combinations
and FUNWAVE agreed well with the measurements. The agreements indicate that it is necessary to take into
account the effect of wave-induced current on wave refraction when wave breaking occurs over a submerged
shoal.
© 2008 Elsevier B.V. All rights reserved.

1. Introduction (1989) experiment, they neglected the effect of the wave-induced


current on wave transformation. The REF/DIF (REF/DIF 1 for mono-
Waves propagating over a submerged shoal are transformed by chromatic waves and REF/DIF S for random waves) model, which is a
refraction, diffraction, shoaling, and wave breaking. In addition, waves parabolic wave model including weak nonlinear wave effects, cannot
breaking over the shoal induce a jet-like strong shear current behind take into account the current field without an extra current module.
the shoal. Thus, characteristics of wave transformation are altered due Yoon et al. (2004) simulated wave transformation due to wave-in-
to the breaking wave-induced current. Kirby et al. (1998) and Chen duced currents by using a model system combining REF/DIF S and a
et al. (2000) simulated currents induced by breaking waves behind a current model based on the shallow water equations. Indeed, they
submerged circular shoal by utilizing a fully nonlinear Boussinesq showed that waves breaking over a submerged elliptic shoal induce a
wave model called FUNWAVE. However, it does not appear that the jet-like current behind the shoal, and the breaking wave-induced
wave-induced current in their simulation appreciably influences the current defocuses waves. This is called current-induced refraction, and
wave field because the wave breaking region is not wide enough, and is a counterpart of depth-induced refraction. If the effect of this wave-
therefore the wave-induced current is relatively narrow and weak. induced current is neglected, wave height distributions behind the
While Kirby and Őzkan (1994) developed REF/DIF S, an extended shoal are overestimated (Yoon et al., 2004).
version of REF/DIF 1(Kirby and Dalrymple, 1994) using a spectral ap- To investigate the effect of wave-induced current on random wave
proach for random waves, and simulated the Vincent and Briggs transformation over a submerged shoal, Choi et al. (2007) constructed
two numerical model systems. The first is a combination of REF/DIF S
and SHORECIRC, the latter of which is a nearshore current model
⁎ Corresponding author. Tel.: +82 31 400 5144; fax: +82 31 408 5140. developed by the University of Delaware. It has been verified that
E-mail address: sbyoon@hanyang.ac.kr (S.B. Yoon). SHORECIRC can simulate nearshore circulations induced by breaking

0378-3839/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.coastaleng.2008.09.002
298 J. Choi et al. / Coastal Engineering 56 (2009) 297–312

waves, although this current model needs a wave-driver model to acceleration, and h is local water depth. The directional correction
calculate the radiation stresses, the gradients of which generate wave- coefficients were chosen as a0 = 1,a1 = −0.75, and b1 = −0.25, the values
induced currents (Svendsen et al., 2002). SHORECIRC is a more so- of which were followed by the Booij (1981) approximation. The energy
phisticated current model than the current model developed by Yoon dissipation coefficient aBr for wave breaking is given by Thornton and
et al. (2004). The second model system is a combination of SWAN and Guza (1983) as
SHORECIRC. SWAN, developed by Delft University, is a practical wave pffiffiffi
model in many respects. However, it does not include a module to π f B3 5
α Br ¼ 3 H ð2Þ
compute a current field as REF/DIF does not (Booij et al, 2004). In ad- 4 γ4 h5 rms
dition, FUNWAVE, a fully nonlinear Boussinesq wave model developed _
(Wei et al, 1995) and modified (Chen et al, 2003) by the University of where f is the representative frequency and Hrms the root mean
Delaware, can be utilized to simulate instantaneous wave surfaces and squared wave height. B and γ are the empirical coefficients. For a
wave-induced currents without a separate run of wave and current monochromatic wave condition, REF/DIF 1 (Kirby and Dalrymple,
models. 1994) is employed. The energy dissipation due to wave breaking in
The monochromatic and random wave cases of the Vincent and REF/DIF 1 is given as
Briggs (1989) experiment were simulated by using the two combina- H
tions REF/DIF-SHORECIRC and SWAN-SHORECIRC, as well as the α Br ¼ 0 for bκ Br
h
FUNWAVE model. These computed results were compared with each "  ð3Þ
2 #
other and the measurements. Prior to considering the effect of the K γs h H
wave-induced current, simulations of non-breaking cases neglecting α Br ¼ 1− for zκ Br
2h H h
the current effect were conducted. These simulations helped to
understand the distinction between the different wave models due to where K and γs are empirical constants. Measurements by Horikawa
the effect of wave-induced currents. and Kuo (1966) showed that a stable value of the wave height H = γsh
The numerical models used in this study were developed was approached for waves over a uniform water depth after breaking
independently and have been reliably modified and verified. There- on a plane slope. They suggested that K = 0.15 and γs = 0.4. The
fore, the governing equations and the model closures for each breaking index relation is κBr = 0.78 for plane slopes of 0.015–0.15.
numerical model are described very briefly in Section 2. For numerical SWAN (Booij et al, 1999, 2004) is used for simulating random wave
simulations, the two wave and current model combinations and the transformation due to refraction by topography and current, shoaling,
FUNWAVE model were set up corresponding to the experimental and energy dissipation due to wave breaking and bottom friction
conditions for bathymetry, frequency and directional spectra for (among other reasons). The unknown variable in the governing
random wave cases. Descriptions for the model setup are shown in equation of SWAN is the action density N (i.e., energy density E
Section 3. The simulations of monochromatic and random waves were divided by its frequency σ in directional (θ) and frequency (σ)
conducted using REF/DIF, SWAN and FUNWAVE and the computed spectrum), N(σ,θ) = E(σ,θ)/σ. This wave action density balance equa-
results are compared with the measurements in Section 4 and 5 for tion includes the terms which represent propagation of action in
non-breaking and breaking cases, respectively. The results under geographical space, shifting of the relative frequency due to variations
breaking wave conditions showed the effects of breaking wave- of depths and current in frequency space, and depth-induced and
induced currents on wave transformation. A summary and conclu- current-induced refraction in directional space. And the source term
sions are presented in Section 6. in the equation represents the effects of generation, dissipation and
nonlinear wave-wave interactions. The SWAN model does not include
2. Numerical models wave phases so that interference patterns cannot be accounted for,
even though this model is capable of simulating wave reflection by
The REF/DIF (Kirby and Dalrymple, 1994; Kirby and Özkan, 1994) obstacles. For the diffraction effect in this model, a phase-decoupled
model is utilized to simulate depth-induced and current-induced refraction-diffraction approximation, which is based on the mild slope
refraction, diffraction, shoaling, and energy dissipation of monochro- equation omitting phase information, was suggested by Holthuijsen
matic or random waves. Since this model was developed in para- et al. (2003). In order to consider the effect of mean currents on the
bolic form under the assumption that waves are mainly traveling in the wave transformation, current information should be provided as input
x-direction, it does not predict the waves reflected directly back to the to SWAN using an extra current model. The mean rate of energy
direction they came. However, this model can deal with the y-directional dissipation per unit horizontal area due to wave breaking, Dtot (Battjes
interference patterns (transverse standing waves) due to reflected and Janssen, 1978), is represented as
waves from a lateral wall boundary or refracted waves crossing each
  
other. In the governing equation of the REF/DIF model, the complex ρg σ
Dtot ¼ α BJ Q b γBJ h ð4Þ
wave amplitude An for the nth wave component in an input frequency- 4 2π
directional spectrum is the unknown variable. Note that each of the
computational components An is characterized by a unidirectional where ρ is sea water density, αBJ is the proportionality coefficient of
P
monochromatic wave at an initial stage, but can be transformed into the rate of dissipation, Q b is the fraction of breaking waves, σ is the
multi-directional wave components due to wave reflection or refraction. mean frequency, and γBJ is the breaker parameter. The fraction of
And in order to take into account the effect of mean currents, mean depth-induced breakers Q b is determined using
current velocity U and V should be computed with an extra current
1−Q b Etot
model. The properties of each computational component, such as wave ¼ −8  2 ð5Þ
lnQ b
number kn, the reference wave number given along an incidence γ BJ h
boundary k̄¯n, phase velocity (C)n, and group velocity (Cg)n, are de-

termined by the spectrum and the following dispersion relationship: where Etot = ∫2π
0 ∫0 E(σ,θ)dσdθ.
A current model, SHORECIRC (Svendsen et al., 2002) is utilized to
2
σ 2n ¼ ðωn −kn U Þ ¼ gkn tanhkn h ð1Þ solve wave-averaged motion such as wave-induced nearshore cir-
culations and infragravity waves. The wave-induced motions in the
where σn is the intrinsic frequency that takes into account the Doppler simulation using SHORECIRC are generated due to excess momentum
effect due to currents U, ωn is absolute frequency, g is gravitational fluxes which can be evaluated using the wave models. The theoretical
J. Choi et al. / Coastal Engineering 56 (2009) 297–312 299

background is well described in Putrevu and Svendsen (1999) and many using the amplitude distributions An of the wave component n com-
references show the applicability of this model towards the simulation puted by using REF/DIF, Sm and Sp are written as
of nearshore currents (Svendsen, 1984, Svendsen et al., 2002). Although
this model was recently extended into a quasi 3-dimensional version
!
ρg Cg n
by using an analytical vertical velocity profile, the depth-integrated Sm ðσ ; θÞ ¼ jAn j2 ð13Þ
2 ðC Þn
2-dimensional version was used in this study. The basic idea of this
model starts from a decomposition of the total fluid velocity into
!
three kinds of velocity components: ρg 2 Cg n
Sp ðσ ; θÞ ¼ jAn j2 −1 ð14Þ
0
4 ðC Þn
uα ¼ uα þ uwα þ Vα ð6Þ

where uαV denotes the turbulent component velocity, uwα the wave while using the action density spectrum at each grid point in SWAN
component velocity defined so that u P = 0 below trough level, and V
w α
P
is the current velocity. The over-bar ( ) denotes a short wave time Cg ðσ ; θÞ
average, and the subscripts α and β denote the directions in a Sm ðσ ; θÞ ¼ σN ðσ; θÞ ð15Þ
C ðσ; θÞ
horizontal Cartesian coordinate system (see below). From the
Reynolds averaged Navier-Stokes equation, the governing equation
 
for SHORECIRC is derived by using depth-integration and averaging σ 2Cg ðσ ; θÞ
Sp ðσ ; θÞ ¼ Nðσ; θÞ −1 ð16Þ
the equation over a short wind wave period. For the equations, the 2 C ðσ ; θÞ
viscous term is modeled based on the turbulent eddy viscosity con-
cept and written as where a wave direction is defined with respect to the direction of
  the main propagating wave, the x-direction, and θ and σ denote
AVβ AVα
ταβ ¼ ρmt þ ð7Þ the direction and frequency of a wave component in the spectral
Axα Axβ domain.
where FUNWAVE, a time-dependent phase-resolving wave and current
model, employs the fully-nonlinear Boussinesq equations derived
rffiffiffiffiffiffiffi  1=3
fcw D by Wei et al. (1995). This model can simulate wave reflections as
mt ¼ C1 κ u0 h þ Mh þm0 þ ms : ð8Þ well as interference patterns in any direction differently from the
2 ρ
above wave models. And the FUNWAVE model is not only used for
In this equation, C1 and M are empirical coefficients, κ is the simulating monochromatic and random waves but also capable of
Karman constant, fcw is the bottom friction coefficient related to both simulating nearshore hydrodynamics including interaction between
wave and current, ν0 is the empirical ambient eddy viscosity, νs is the waves and currents. Surface elevation ζ and horizontal velocity u⁎
Smagorinsky eddy viscosity with an empirical coefficient Cs (Smagor- at the water depth z = z⁎ = − 0.531h0 are the unknown variables in the
insky, 1963), D is the energy dissipation rate due to wave breaking governing equations, where h0 is the still water depth. An
(equivalent to Dtot in SWAN), and u0 is the magnitude of wave velocity additional term is added in the momentum equation for partial-
evaluated at the bottom. To evaluate the bottom friction due to both rotational motions for a new version (Chen et al., 2003), and is
wave and current, an energy-weighted mean wave direction is chosen written as
as a representative direction.
The excess momentum fluxes Sαβ due to wave fluctuations (Mei, V ¼ ð−vT ω1 ; uT ω1 Þ ð17Þ
1989) are expressed as
where
P 1
ζ
Sαβ ¼ ∫−h pδαβ þ ρ uwα uwβ dz−δαβ ρgh2 ð9Þ
0 2 AzT A A AzT
ω1 ¼ ½j  ðh0 uT Þ þ zT ðj  uT Þ −
where the total water depth h is defined as the sum of mean free Ax Ay Ay Ay
P ð18Þ
surface ζ and water depth h0 under still water level. The surface A A
P  ½j  ðh0 uT Þ þ zT ðj  uT Þ
elevation ζ = ζ + η, η is the wave component surface displacement, p is Ax Ax
the pressure, and δαβ is the Kronecker delta. For small ratios of wave
amplitude to wave length, δαβ can be approximated as which allows for second-order effects of vertical vorticity, where j is
 the 2-dimensional gradient tensor. For practical application, the
ζ 
Sαβ ¼ ∫−h ρ uwα uwβ dz governing equations require that the model schemes take into account
0 ( )
ρgη2 ζ @  ζ physical effects such as bottom friction, turbulent viscous damping,
þ δαβ þ ∫−h ∫zζ ρ uwα ww dz0 dz−∫−h ρw2w dz ð10Þ
2 0 @ xα 0 and wave breaking as well as some numerical needs such as wave
generation and boundary absorption. For wave generation, FUNWAVE
which is called the radiation stress tensor by Longuet-Higgins and employs the internal source approach (Wei et al., 1999), which is
Stewart (1962, 1964). In REF/DIF and SWAN, the radiation stresses at a based on an ad-hoc source mechanism by Larsen and Dancy (1983). To
spatial grid location (x,y)can be obtained by generate random waves corresponding to a given frequency-direc-
tional spectrum, the internal source term added to the continuity
 
Sαβ ¼ ∫∫ eαβ ðθÞSm ðσ ; θÞ þ δαβ Sp ðσ; θÞ dθdσ ð11Þ equation can be written as

where M J

f ðx; y; t Þ ¼ qðxÞ ∑ ∑ Ds Amj ; fm ; θj sin km ysinθj −2πfm t þ /mj ð19Þ


  m¼1 j¼1
cos θw
2
sinθw cosθw
eαβ ðθÞ ¼ ð12Þ
sinθw cosθw sin2 θw where q(x) = exp[− βs(x − xs)2] is a Gaussian shape function in which
βs = 5/(λ/4)2 is a shape coefficient for the source function, which is
in which θw(x,y) is the wave direction of each wave component in a in turn determined from the peak wave component. The peak
spatial domain, θ is the wave direction in a directional spectrum, and wave length is λ and xs is the central location of the source in the
300 J. Choi et al. / Coastal Engineering 56 (2009) 297–312

x-direction. Ds(Amj,km,θj) is the magnitude of the source function


written as
h i
2

2Amj cosθj ð2πfm Þ −ðα s þ 1=3Þgk4m h30
Ds Amj ; km ; θj ¼ h iqffiffiffiffi k cosθ 2  ð20Þ
2 π ð m jÞ
2πfm km 1−α s ðkm h0 Þ β exp − 4β
s s

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where Amj ¼ 2Eð f ; θÞdfdθ is the deterministic amplitude of a
wave traveling in the θj direction with a frequency fm and a wave
number km corresponding to fm. The random phase independent
of frequency and direction is ϕmj, which is uniformly distributed
over the interval (0,2π), and α s ¼ −0:390. For energy dissipation
due to wave breaking, FUNWAVE employs the momentum mixing
Fig. 1. Location of the submerged elliptic shoal, the centerline at y = 12.5 m and the
scheme introduced by Kennedy et al. (2000) and Chen et al. (2000). In transect 4 at x = 12.2 m (Vincent and Briggs, 1989).
addition, an extra diffusion term is added for the diffusivity that is
strongly localized on the front face of a broken wave. In this scheme
the empirical coefficient pffiffiffiffiffiffiffi η(I)
tffi that controls
pffiffiffiffiffiffiffithe
ffi onset of wave breaking
structed as shown in Fig. 1. The computational grids were uniform
varies between 0:35 gh0 and 0:65 gh0 . FUNWAVE also includes with dimensions of 0.1 × 0.1 m in the spatial domain. The center of the
energy dissipation due to turbulent mixing, based on the Smagorinsky shoal used in their experiment was located at x = 6.1 m and y = 12.5 m
(1963) subgrid approach. This subgrid mixing term defines the eddy in the computational domain with 19 m in the x-direction, which is
viscosity as the main wave propagating direction, and 25 m in the y-direction. In
order to avoid erroneous lateral boundary situations due to no
"      #1=2 incoming wave along the lateral wall boundaries, we appended an
AU 2 AV 2 1 AU AV 2
ms ¼ cm ΔxΔy þ þ þ ð21Þ extra 10 m in the y-direction of the actual computations using the
Ax Ay 2 Ax Ay
three wave models. In utilizing SHORECIRC, a no-flux wall boundary
condition was specified along all lateral boundaries. In addition, for
in which U and V are the velocity components of the time-averaged the simulation performed using FUNWAVE, an additional domain for
current field, cm is the mixing coefficient, and Δ x and Δy are the grid internal wave generation and sponge absorption layer were required.
spacings in the x and y directions, respectively. In addition, the bottom Thus, we appended an extra 15 m in the x-direction.
shear stress is modeled by The random wave spectra for computations were formed using the
TMA and the wrapped normal distribution function. For the boundary
fc input wave spectra of REF/DIF, 61 directional components and 41
Rf ¼ uT juT j ð22Þ
h0 þζ frequency components were used. For the energy spectral domain in
the simulations using SWAN, we used 61 directions over the 180°
where fc is the bed shear stress coefficient. (from −90 to 90°), which indicates that the directional grid Δθ = 3°, and
41 frequencies in the range from 0.5 to 1.5 Hz. The frequency cutoffs
3. Model setup followed those of the Vincent and Briggs experiment. The spectral
space consists of 2501 bins. The 3-degree resolution is within the
Vincent and Briggs (1989) performed experiments on monochro- range recommended by Booij et al. (2004) for narrow spectrum
matic waves and frequency-direction spreading random waves which condition (with the 10-degree directional spreading). For FUNWAVE,
pass over a submerged elliptic shoal by using a directional spectral which is the wave-phase-resolving model that requires higher
wave generator (DSWG) in Coastal and Hydraulics Laboratory's wave resolution than the other wave models, 121 directional components
basin of the U.S. Army Engineer Research and Development Center. and 81 frequency components were used to generate incident random
The spectra for random wave generation were formed using TMA waves. The parameters for narrow and broad-banded frequency and
(Bouws et al., 1985) for the frequency spectrum and a wrapped normal directional spectra in the experiment of Vincent and Briggs are shown
distribution function for the directional spreading function (Borgman, in Table 1.
1984). The experiments included many cases with combinations of For SWAN, the diffraction option was switched on with the filter
unidirectional, narrow and broad spectra in direction, and monochro- width around 0.7 m. The 0.7 m filter width was chosen for numerical
matic, narrow, and broad spectra in frequency. The wave conditions stability. This is larger than the filter width (0.4 m) used by Holthuijsen
used for the simulations in this study are shown in Table 1. et al. (2003) to achieve the diffraction effect in the simulations of the
For numerical simulations of the Vincent and Briggs (1989) ex- Vincent and Briggs experiment using SWAN. Their spatial grid was
periment, the bathymetry with a submerged elliptic shoal was con- 0.2 × 0.2 m. Holthuijsen et al. (2003) recommended that the directional

Table 1
Input wave conditions (Vincent and Briggs, 1989)

Input Case ID Peak period (s) Significant wave height (cm) γTMA σTMA (degree)
Non-breaking case Monochromatic M2 1.3 2.54 – –
Narrow direction Broad frequency N3 1.3 2.54 2 10
Narrow frequency N4 1.3 2.54 20 10
Broad direction Broad frequency B3 1.3 2.54 2 30
Narrow frequency B4 1.3 2.54 20 30
Breaking case Monochromatic M3 1.3 13.5 – –
Narrow direction-frequency N5 1.3 19.0 20 10
Broad direction-frequency B5 1.3 19.0 2 30
J. Choi et al. / Coastal Engineering 56 (2009) 297–312 301

resolutions Δθ of wave spectra should be small compared to their short numerical ramping interval equivalent to approximately 2 peak
directional spreading to predict reasonable diffraction effects using wave periods was added. During the ramping interval at the initial
SWAN. It should be addressed that the 3-degree directional resolution stage the gradients of radiation stress in the current model increases
in this study is low for quasi-monochromatic waves with an extremely gradually from zero to the actual value computed by the wave models
narrow directional spectrum. However, the resolution higher than the in order to prevent an abrupt transition error. In the simulation using
present was beyond the limitation of our computer storage system. FUNWAVE, we allowed the waiting time to include a ramping interval
Moreover, the effect of wind, wave-wave interactions, and the white- and the elapsed time needed for the high frequency waves to arrive at
capping implemented in the SWAN model were neglected in these x = 15.25 m. In the three wave models the computed results of wave
simulations. The chosen resolution in frequency domain was not high height information were the values of Hmo for the random waves and
enough to consider the effect of wave-wave interactions in these the values of Hrms for the monochromatic wave. The wave height data
simulations. However, from some trial and error tests, the effect of shown in this study are the values non-dimensionalized by the wave
wave-wave interactions was found to be less significant than the effect height generated at the wavemaker (Yoon et al., 2004).
of wave-current interaction in the simulations for wave-induced The radiation stresses generating the wave-induced current in
currents. SHORECIRC were obtained from the directional frequency action
The breaking wave cases of monochromatic and random waves in density spectra (SWAN) or the amplitude distributions of each com-
the experiment of Vincent and Briggs were simulated mainly to study ponent (REF/DIF) as written in Eqs. (10)–(16). In addition, the depth-
the influence of wave-induced currents on wave transformation. Prior integrated mean current velocities computed from SHORECIRC were
to conducting simulations of the breaking waves passing over the applied as input data to simulate wave fields using REF/DIF and SWAN.
submerged shoal, simulations of the non-breaking cases, in which the In other words, a main program controlled the feedback of wave- and
wave-induced current effect can be neglected, were performed also to current-subprograms, REF/DIF and SHORECIRC or SWAN and SHOR-
determine differences between the REF/DIF, SWAN, and FUNWAVE ECIRC, for wave and current interactions. Under the assumption that
models. the rate of wave field change would be slower than the rate of current
Vincent and Briggs (1989) stated the followings: 36.4 s wave field change, the wave field was computed every 10 s during which the
elevation data for the monochromatic waves and 260 s data for the current field was computed with a 0.02 s time-step size for
spectral waves were sampled at 50 Hz. A sufficient waiting time was SHORECIRC. This recalculation was repeated for the entire simulation
allowed to elapse in order to permit the slower-traveling high time, which was around 260 s. Unlike the aforementioned two wave
frequency component waves to travel to the remote transect at models, FUNWAVE is able to solve time-dependent wave surface
x = 15.25 m prior to the collection of data after the start of the DSWG. motion and horizontal flow motion at the phase-resolving time level.
Wave height information was obtained using zero-crossing analysis. In other words, an additional current model is not required to solve
For the spectral waves a comparison of the significant wave height H1/3 the current field in FUNWAVE.
(obtained from zero-crossing analyses) and the Hmo (obtained from For detailed descriptions of numerical schemes on the models, see
spectral analyses) value indicated very good agreement, with minor Booij et al. (2004) for SWAN, Kirby and Özkan (1994) for REF/DIF,
differences occurring only at the top of the shoal. In addition, it is worth Svendsen et al. (2002) for SHORECIRC, and Kirby et al. (1998) for
mentioning that the wave height data presented in their study were FUNWAVE. Most physical or empirical coefficients required for the
normalized by the wave height at “gage 10,” which was located at models followed default values of each model in these simulations.
x = 3.05 m and y = 20.12 m as shown in Fig. 1. In this paper, the For the empirical coefficients of the energy dissipation due to wave
experimental data were normalized by target wave height at the breaking, we chose the values as B = 1 and γ = 0.6 in REF/DIF S, and
wavemaker, since the wave height at “gage 10” varies according to K = 0.14 and γs = 0.4 in REF/DIF 1. A breaking index relation κBr = 0.68,
the chosen phase lags between directional components of incident which is smaller than 0.78 (the suggested value for a plane slope) was
random waves along the wavemaker. This will be discussed in more used in REF/DIF 1 for the bathymetry with a submerged shoal. The
detail later. directional correction coefficients a0 = 1, a1 = −0.75, and b1 = −0.25,
Accordingly, numerical simulations in this study were performed following the Booij (1981) approximation, were used in REF/DIF. Kirby
for longer than 260 s with a time step Δt = 0.02 s for both SHORECIRC (1986) showed that the errors in this approximation are small when
and FUNWAVE. In this study the stationary mode was used for the θ ≤ 40°. Although the largest angle of the broad band amplitude
SWAN model. For the wave and current model combination systems, a spectra used for REF/DIF S was around 60°, the wave energy for

Fig. 2. Wave height computed with (a) REF/DIF, (b) SWAN and (c) FUNWAVE for non-breaking monochromatic wave case M2.
302 J. Choi et al. / Coastal Engineering 56 (2009) 297–312

aging the instantaneous fluid velocity over approximately 3 peak wave


periods.

4. Non-breaking waves

Prior to considering the effect of wave-induced current, to check


the performance of the different wave models, numerical simulations
neglecting wave-current interaction were conducted for the non-
breaking wave conditions.

4.1. Monochromatic waves

Figs. 2 and 3 show the results for the non-breaking monochromatic


wave case M2 of the Vincent and Briggs experiments simulated using
REF/DIF, SWAN, and FUNWAVE. Since the SWAN model deals with
only random waves, it is basically inadequate to simulate monochro-
matic waves. To achieve a similar effect from quasi-monochromatic
wave simulation using SWAN, the incident waves were approximated
with a very narrow frequency and directional spectrum (Holthuijsen
et al., 2003). In other words, along wave input boundary the energy
density corresponding to the peak wave was set in 3 bins with one
frequency and a very narrow directional distribution among 2501
spectral bins. As already addressed, for this quasi-monochromatic
wave simulation, the 3-degree directional resolution used in this
study is too low to compute wave diffraction effect using SWAN. The
results of FUNWAVE were the Hrms values obtained by spectral
analyses of the instantaneous free surface time history for 36.4 s after
the waiting time of 16.9 s.
Fig. 3. Wave height distribution computed with REF/DIF, SWAN and FUNWAVE for non- The horizontal distributions of wave height in Fig. 2 show the
breaking monochromatic wave case M2 along (a) centerline and (b) transect 4.
focusing of waves due to depth-induced refraction behind the shoal. In
addition, Fig. 2 shows some of the characteristics of the wave models.
40° ≤ θ ≤ 60° is less than 10% of the total wave energy in the spectra. For REF/DIF does not deal with wave reflection due to underwater
SWAN, the proportionality coefficient of the dissipation rate was geometrical changes (Kirby and Özkan, 1994). However, it does deal with
αBJ = 1 and the fraction of breaking waves was γBJ = 0.8. This value 0.8 is transverse standing waves due to wave refraction in the y-direction. In
not the default. However, it is within the range proposed by Battjes other words, the interference patterns due to superposition of waves
and Stive (1985) and close to the values 0.78 of Wood et al. (2001). In moving in opposite y-directions can be taken into account in REF/DIF.
fact, this value 0.8 was chosen for closer results to the measurements SWAN cannot deal with the interference patterns because the wave
after some trial and error tests. In utilizing SHORECIRC, we set the phases are neglected. Moreover, SWAN has a limited diffraction effect,
parameters related to turbulent eddy viscosity as C1 = 0.2, M = 0.2, which is modeled depending on the grid size in the spatial and directional
κ = 0.4, fcw = 0.02, ν0 = 0.002, and Cs = 0.2, which are within a range space. FUNWAVE manages the interference patterns due to either
recommended in the SHORECIRC manual. In theffi simulation conducted
pffiffiffiffiffi wave refraction or wave reflection. Fig. 3 shows a comparison between
ðI Þ
using FUNWAVE, the value ηt ¼ 0:35 gh was chosen for the the computed results and the measurements along the centerline and
breaking energy dissipation coefficient, the mixing coefficient for the transect 4. It indicates that the wave heights along the centerline
eddy viscosity cm = 0.15, and the bed shear stress coefficient for energy at y=12.5 m (top) and the transect at x=12.2 m (bottom) computed by
dissipation due to bottom friction fc = 0.0015. In order to obtain the RFF/DIF and FUNWAVE agree with the measurements, and confirms that
eddy viscosity for energy dissipation due to wave breaking and SWAN is inadequate to simulate monochromatic waves with the
turbulent mixing, the underlying flow field was calculated by aver- interference patterns.

Fig. 4. Wave height computed with (a) REF/DIF, (b) SWAN and (c) FUNWAVE for non-breaking random wave case N4.
J. Choi et al. / Coastal Engineering 56 (2009) 297–312 303

Fig. 5. Wave height distribution computed with REF/DIF, SWAN and FUNWAVE for non-breaking random wave case N3 and N4 along (a) centerline and (b) transect 4.

4.2. Random waves number of separate wave components (An), each of which has a single
frequency and a single direction. Each of those computational com-
For numerical simulation of random waves using the REF/DIF ponents (An) is characterized by a unidirectional monochromatic wave
model the input waves can be prepared by two options. The first at an initial stage, but can be transformed into multi-directional wave
option is to discretize the whole spectrum of the input waves into a components due to refraction over a variable water depth region.

Fig. 6. Wave height distribution computed with REF/DIF, SWAN and FUNWAVE for non-breaking random wave case B3 and B4 along (a) centerline and (b) transect 4.
304 J. Choi et al. / Coastal Engineering 56 (2009) 297–312

Fig. 7. Schematic representation of wave breaking and refraction pattern over a shoal: (a) focusing zone without wave-induced current and (b) shadow zone with wave-induced
current (Yoon et al., 2004).

Thus, phase interference between the multi-directional wave compo- However, the interference patterns in the transverse direction for the
nents in an individual computational component (An) is still allowed random wave cases are not apparent because a number of spectral
as in the case of monochromatic version of REF/DIF. However, during wave components with different directions were superposed. The
the computation each of the computational components (An) is in- wave heights calculated using REF/DIF and SWAN are symmetric with
dependently calculated (except for the energy dissipation due to respect to the centerline of y = 12.5 m, while this symmetry is not clear
bottom friction and wave breaking) and the phase interference with for the case of FUNWAVE. As discussed earlier the phase interference
the other computational components is neglected. Thus, the wave between wave components with different directions along the
height (Hmo) remains constant along the generation line and in the internal generation line is responsible for this asymmetry of wave
computational domain if the water depth is uniform. The second height distribution in the result of FUNWAVE.
option, which needs a slight modification of the original code, is to Figs. 5 and 6 show comparisons of the computed and measured
discretize only the frequency spectrum (Kirby and Dalrymple, 1994). wave heights for the cases of narrow (N3 and N4) and broad (B3 and
Each of the discretized components includes implicitly the informa- B4) directional spectra, respectively. The wave heights along the
tion of directional distribution for a given frequency. In this case the center line (y = 12.5 m) and the transect 4 (x = 12.2 m) calculated using
phase interference between the waves propagating in different
directions (with random phases) is possible within each frequency
component. Thus, the wave height (Hmo) for the multi-directional
waves would be non-uniform even along the generation line with
constant water depth (Suh and Dalrymple, 1992). In practice the first
option is more frequently employed. In the present study the first
option was selected. In other words, the phase interference between
different computational components (An) was neglected as in the case
of SWAN. On the other hand, the FUNWAVE is a phase-resolving
primitive-variable model which can deal with random waves without
dividing the variables into separate computational components, thus,
full-interactions between wave components are assured. However, in
order to prescribe the incident random waves along the internal
wavemaker of FUNWAVE, discretizing the wave spectrum into a
number of components is necessary as in the case of REF/DIF or SWAN.
The instantaneous free surfaces for each spectral wave component are
calculated and superposed with phase lags randomly distributed over
the interval of 0 to 2π. As a result, the phase interferences between
directional components appear in the wave height (Hmo) along the
generation line. Thus, it is hard to achieve a uniform distribution of
wave height along the generation line and, in turn, in the computa-
tional domain. For the input waves with directional spreading the use
of measured wave height at a particular location, e.g. “gage 10”, as a
reference wave height for the normalization of measured data is not
appropriate for this reason.
The numerical results for non-breaking random wave cases are
shown in Figs. 4–6, which show wave height distributions computed
by REF/DIF, SWAN, and FUNWAVE. The wave height distributions of
FUNWAVE were the Hmo values obtained by spectral analyses of the
instantaneous free surface time history for 260 s after the waiting time
of 27.3 s.
Fig. 4 presents the results of the N4 case with a narrow frequency- Fig. 8. Wave height distribution with and without the current effect and mean current profile
directional spectrum. The horizontal distributions of wave height (bottom panel) for monochromatic wave case M3 along (a) centerline and (b) transect 4.
show wave focusing due to depth-induced refraction behind the shoal. RD=REF/DIF, RD+SC=REF/DIF and SHORECIRC, SW=SWAN, SW+SC=SWAN and SHORECIRC.
J. Choi et al. / Coastal Engineering 56 (2009) 297–312 305

Fig. 9. Wave height, mean current velocity and mean surface elevation of the 36.4-second simulation computed with REF/DIF and SHORECIRC (top panels) and SWAN and SHORECIRC
(bottom panels) for monochromatic wave case M3.

Fig. 10. Wave height distribution and mean current profile computed with FUNWAVE and the combined models for monochromatic wave case M3 along (a) centerline and (b)
transect 4. RD = REF/DIF, RD + SC = REF/DIF and SHORECIRC, SW = SWAN, SW + SC = SWAN and SHORECIRC.
306 J. Choi et al. / Coastal Engineering 56 (2009) 297–312

various wave models show reasonable agreements with the measured strong gradients induce a jet-like current behind the shoal. The jet-like
ones. Some irregularity in wave heights in the transverse direction is wave-induced current defocuses waves behind the shoal. Yoon et al.
observed in the calculated results using FUNWAVE, while this ir- (2004) explained this mechanism using the refraction diagram depicted
regularity disappears in the cases of simulated results using REF/DIF in Fig. 7 with neglecting wave diffraction as : The breaking of waves over
and SWAN as discussed earlier. The measurements also show the the shoal induces strong currents in the direction of wave propagation,
irregularity. Even though the details of the irregularity are slightly and the breaking-induced currents, in turn, defocus the converging
different from each other, the magnitude of irregularity is in the same wave rays outwards from the central part behind the shoal. If there were
level. This irregularity in the wave height calculated using FUNWAVE no breaking-induced current, the converging wave rays would cause the
appears from the generation line due to phase interference between wave height to grow again in the focusing zone. However, instead of the
wave components with different directions. The irregularity is focusing zone, a shadow zone appears there due to defocused wave rays,
persistent even though the free surface data of a longer time span and two caustics are formed next to the shadow zone.
are analyzed to achieve a stationary wave height.
From the comparisons between numerical results and the 5.1. Monochromatic waves
experimental measurements it can be concluded that all the wave
models gave a reasonable prediction for non-breaking random waves In this section, monochromatic waves transformed due to breaking
with frequency and directional spectra. In addition it can be inferred wave-induced current were investigated using simulations computed
that the non-breaking waves of the present experimental cases were by REF/DIF-SHORECIRC and SWAN-SHORECIRC, as well as FUNWAVE.
little influenced by wave-induced current because the results using The wave and current fields are transformed by their interactions. For
REF/DIF and SWAN without coupling the current module agree with the breaking monochromatic wave conditions, REF/DIF 1 was em-
the experimental measurements and with those using FUNWAVE. ployed instead of REF/DIF S, and the energy density corresponding to
the peak wave was set in one frequency and three directional bins
5. Breaking waves among 2501 spectral bins for an input boundary spectrum of SWAN.
The computational results of a breaking monochromatic wave case
Numerical simulations including wave-current interaction were (M3) are presented in Figs. 8–11. These are the results at t = 36.4 s in
conducted for the breaking wave conditions. In the computations the the SHORECIRC time level, and correspond to the data obtained by
breaking of waves over a submerged elliptic shoal rapidly increases the analyzing the instantaneous free surface for 36.4 s after the waiting
gradients of radiation stresses (or excess momentum fluxes) and the time of 16.9 s as mentioned above. Figs. 8 and 9 show results

Fig. 11. Wave height, mean current velocity and mean surface elevation of the first 36.4-second (top panels) and the second 36.4-second (bottom panels) FUNWAVE simulation after
16.9-second waiting time for monochromatic wave case M3.
J. Choi et al. / Coastal Engineering 56 (2009) 297–312 307

simulated using the combinations REF/DIF-SHORECIRC and SWAN- directional information of monochromatic waves is included in the
SHORECIRC. In Fig. 8 the wave heights computed using REF/DIF and phase of the complex amplitude. The phase interference between
SWAN without coupling with SHORECIRC are also presented. The top waves propagating in different directions is assured in the computa-
panel of Fig. 8 shows values along the centerline and the two bottom tion, and the wave amplitude can be easily obtained by taking an
panels along the transect 4. When the effect of wave-induced currents absolute value of the complex amplitude. However, when the angle of
is excluded, waves are focusing behind the shoal, and thus one local propagation direction is calculated for the radiation stresses at a grid
peak in wave height distribution appears near x = 10 m of the point, a single representative angle is determined without decompos-
centerline y = 12.5 m as conceptually predicted in Fig. 7(a). However, ing the waves into directional components. Thus, for the transformed
when the current effect is included, the local peak disappears along waves with multi-directional wave components, the contributions
the centerline. Instead two local peaks appear next to the centerline as from each directional component cannot be appropriately reflected on
shown in the middle panel of Fig. 8. The computed results with the the calculation of radiation stresses. On the other hand, in the SWAN
effect of wave-induced currents agree better with experimental model each spectral bin has its own wave height and angle of direction
measurements for both wave models of REF/DIF and SWAN. However, at any grid location. Since the phase interference between multi-
the accuracy of the models is not satisfactory. Moreover, the current directional wave components in the SWAN model is neglected, the
velocities calculated using different models do not agree with each simulated wave heights are less accurate than those of REF/DIF for
other. The velocity distribution computed with SWAN-SHORECIRC is the monochromatic case. However, the SWAN model is free from the
analogous to the Gaussian distribution of a jet-like current, while that uncertainties involved in the determination of wave directions for
computed using REF/DIF-SHORECIRC is not. the radiation stresses. Judging from the overall pattern of wave-
The horizontal views of these simulations obtained using the two induced currents shown in Fig. 9, the result produced by SWAN
model combinations are shown in Fig. 9 presenting wave height appears to be more realistic. This topic is further discussed later in this
transformed by wave-current interaction, wave-induced current and section by comparing the present results with those simulated using
mean surface elevation. The wave-induced current simulated using FUNWAVE.
REF/DIF-SHORECIRC does not show any organized pattern and appears The computational results of a breaking monochromatic wave case
to be unrealistic, while that calculated using SWAN-SHORECIRC is (M3) simulated using FUNWAVE are presented in Figs. 10 and 11. As
analogous to a jet-like current. The discrepancy between two current already mentioned, FUNWAVE can solve the instantaneous wave field
patterns is caused by the difference between the methods to evaluate and current field at one phase-resolving time level. Thus, it handles
the radiation stresses in each wave model. In the REF/DIF 1 model the interference patterns of standing waves in either the x- or y-direction.

Fig. 12. Wave height, mean current velocity and mean surface elevation of the 260-second simulation computed with REF/DIF and SHORECIRC (top panels) and SWAN and SHORECIRC
(bottom panels) for random wave case N5.
308 J. Choi et al. / Coastal Engineering 56 (2009) 297–312

Fig. 13. Wave height, mean current velocity and mean surface elevation of the 260-second simulation computed with REF/DIF and SHORECIRC (top panels) and SWAN and SHORECIRC
(bottom panels) for random wave case B5.

Fig. 14. Wave height distribution with and without the current effect and mean current profile (bottom panel) for random wave case N5 and B5 along (a) centerline and (b) transect 4.
RD = REF/DIF, RD + SC = REF/DIF and SHORECIRC, SW = SWAN, SW + SC = SWAN and SHORECIRC (Choi et al., 2007).
J. Choi et al. / Coastal Engineering 56 (2009) 297–312 309

Note that results simulated using this model always include the wave- first 36.4-second simulation (top panels). The wave height distribu-
induced current effect. Fig. 10 shows the horizontal profile of wave tions in Fig. 11 show the interference patterns of transverse standing
height and mean current computed by FUNWAVE along the centerline waves behind the shoal. Additionally, the figure shows that a jet-like
and the transect 4. The wave height Hrms was obtained by spectral wave-induced current is developing, and the waves become defo-
analyses in the same way that the measurements were obtained and cused due to the wave-induced current behind the shoal.
the mean motions were obtained by averaging their instantaneous
quantities over 36.4 s after the waiting time. The results are compared 5.2. Random waves
with those of the two model combinations, with and without in-
cluding the effect of the wave-induced current. Along the transect 4 Numerical results from the model systems, REF/DIF-SHORECIRC and
behind the shoal, the wave height distribution computed using SWAN-SHORECIRC, with breaking random wave N5 and B5 conditions
FUNWAVE shows a perfect agreement with the measurements. Since are shown in Figs. 12–14. Figs. 12 and 13 show the horizontal
the calculated wave heights using FUNWAVE agrees with the distributions of wave height, developed wave-induced currents, and
measurements, it can be inferred that the current field calculated mean surface elevation for the cases of N5 and B5, respectively. The
using FUNWAVE is reliable. The wave-induced current simulated figures show the results after a 260-second simulation that includes the
using FUNWAVE has a Gaussian profile of a jet-like current which is effect of wave-induced currents. For the 260-second simulation, each
similar to that computed with SWAN-SHORECIRC. However, the field of the wave and current is transformed by their interactions. As
current profile calculated using REF/DIF-SHORECIRC is totally different described in Choi et al. (2007) the results show that, when wave
from that of FUNWAVE. This confirms that the SWAN model has an breaking occurs, the waves refracted due to the submerged shoal suffer
advantage over the REF/DIF model in the evaluation of radiation additional refraction due to breaking wave-induced currents. The
stresses when the interference between the waves with different current-induced refraction defocuses the waves behind the shoal. A
directions occurs. shadow zone occurs and two weak focuses appear due to the
Fig. 11 shows the horizontal views of wave height, mean currents, converging waves refracted by the current. For these random wave
and mean surface calculated using FUNWAVE for breaking mono- cases, the influence of the interference patterns of transverse standing
chromatic wave case M3. The wave height and mean motions are waves is weak.
obtained from numerical results of the first 36.4 s (top panels) and the In Fig. 14 the numerical results using the two model systems are
second 36.4 s (bottom panels) after a waiting time of 16.9 s. The wave compared with the measurements for N5 and B5 cases. The figure
height compared to the measurement data was the results from the shows wave heights and current velocity computed by including and

Fig. 15. Wave height, mean current velocity and mean surface elevation of the 260-second FUNWAVE simulation after 27.3 s waiting time for random wave case N5 (top panels) and B5
(bottom panels).
310 J. Choi et al. / Coastal Engineering 56 (2009) 297–312

excluding the effect of the wave-induced currents. In the computation condition B5 (bottom panels). Note that the wave shadow zone in each
without coupling with SHORECIRC, the waves refracted over the case has relatively low wave heights behind the shoal due to wave
shoulder of the shoal are focusing behind the shoal. Thus, as shown in defocusing affected by wave-induced currents. The overall distribu-
the left middle panels of Fig. 14, the wave height distribution for N5 tions of wave height, mean current and mean free surface are close to
case along the transect 4, x = 12.2 m, has a pattern with a local peak at the results of the two model systems shown in Figs. 12 and 13.
the center, y = 12.5 m. The same result can be found in Özkan and Kirby However, the results simulated by FUNWAVE are not symmetric with
(1993). However, when the wave-induced current is included, the respect to the centerline, as discussed for the cases of non-breaking
pattern of wave height distribution is reversed, and the computational random waves.
results agree with both the measured data of Vincent and Briggs The profiles of wave height and mean current computed by
(1989) and the calculated results of Yoon et al. (2004). Goda (2004) FUNWAVE along the centerline and the transect 4 are presented in
conducted the simulations for the same cases of N5 and B5 using the Figs. 16 and 17 for N5 and B5 cases, respectively. These results are
parabolic equation wave model developed by Hirakuchi and Mar- compared with the measurements as well as the results of REF/DIF-
uyama (1986) with some improvements on breaking criteria with the SHORECIRC (left panels) and the results of SWAN-SHORECIRC (right
gradational breaker index. From simulations neglecting the effect of panels). The figures also show the results excluding the effect of wave-
the wave-induced current, Goda (2004) concluded that his results induced current computed by the combined models. The results of
were more accurate than those of Özkan and Kirby (1993) or Yoon FUNWAVE agree with those of the model systems including the effect
et al. (2001) because of the improved breaking criteria in his com- of wave-induced current as well as the measurements, and the results
putation. However, the discrepancy between his result and the mea- of SWAN-SHORECIRC are nearly identical to those of FUNWAVE.
sured data at the center of transect 4 was not fully explained. In the The results reconfirm that the wave-induced currents prevent
present study it is clearly shown that the wave-induced current is the the wave focusing behind the shoal. The distribution of the jet-like
key mechanism to resolve this controversial issue. wave-induced current had been slightly tilted to the left side for the
The results of REF/DIF-SHORECIRC are closer to the measurements 260-second simulation time due to the asymmetry of input wave
than those of SWAN-SHORECIRC as mentioned by Choi et al. (2007) height along the generation line as discussed earlier.
but the differences are too small to judge the superiority. Both of the
results show that waves are not focused behind the shoal but rather 6. Conclusions and summary
refracted outside of the jet-like wave-induced current.
Numerical simulations of the breaking random wave cases N5 and A jet-like current induced by waves breaking over a submerged
B5 were also conducted using FUNWAVE. Fig. 15 shows the horizontal shoal changes refraction of waves and the resultant wave transforma-
distributions of wave height, mean current, and mean free surface tion influences the wave-induced current. This transformation due to
obtained by spectral analyzing or averaging the computed instanta- wave-current interactions was investigated using numerical simula-
neous quantities for 260 s after a waiting time of 27.3 s for the narrow- tions for breaking monochromatic and random wave cases. The
spectral wave condition N5 (top panels) and the broad-spectral wave numerical simulations were conducted using two combinations of

Fig. 16. Wave height distribution and mean current profile computed with FUNWAVE and the combined models for random wave case N5 along (a) centerline and (b) transect 4.
RD = REF/DIF, RD + SC = REF/DIF and SHORECIRC, SW = SWAN, SW + SC = SWAN and SHORECIRC.
J. Choi et al. / Coastal Engineering 56 (2009) 297–312 311

Fig. 17. Wave height distribution and mean current profile computed with FUNWAVE and the combined models for random wave case B5 along (a) centerline and (b) transect 4.
RD = REF/DIF, RD + SC = REF/DIF and SHORECIRC, SW = SWAN, SW + SC = SWAN and SHORECIRC.

wave and current models, REF/DIF-SHORECIRC and SWAN-SHORE- the wave-current interaction was not accurately reflected in the
CIRC, as well as the FUNWAVE model. simulation of wave field. The poor performance of the REF/DIF-
To check the performance of various wave models, numerical SHORECIRC model is attributed to the uncertainties involved in the
simulations neglecting the wave-induced currents were conducted evaluation of radiation stresses for the interference pattern of multi-
first for non-breaking wave conditions, and the calculated wave directional waves.
heights were compared with the experimental measurements For the breaking random wave cases the resulting performances of
reported earlier. The results showed that two wave models, REF/DIF the wave-current interaction models are summarized in the follow-
and FUNWAVE, gave good agreements with the experimental ings: Since the incident random waves have directional spectrum, the
measurements for both monochromatic waves and random waves. effect of the phase interference between directional components on
However, the SWAN model is less accurate for the monochromatic the wave climate behind the shoal is negligible. Thus, the simulated
case, while it gave reasonable agreements for the random cases. This is results using REF/DIF-SHORECIRC were similar to those using SWAN-
natural because this model is originally developed to simulate random SHORECIRC and FUNWAVE. The wave heights simulated using three
waves. models gave good agreements with the measured data, and the jet-
For the breaking monochromatic wave case the resulting perfor- like currents induced by the gradient of radiation stresses of breaking
mances of these three wave-current interaction models are summar- waves were generated behind the shoal. The performance of the two
ized in the followings: When waves were breaking on the top of a combined wave and current models is comparable to that of
submerged shoal, a strong jet-like current was developed behind the FUNWAVE except for the irregularity and asymmetry of wave heights
shoal. As a result, the waves refracted towards the center behind the and currents in the transverse direction. Thus, it can be concluded that
shoal were defocused and two local peaks appeared apart laterally the wave-current interaction models based on the conventional
from the center. All three wave-current interaction models gave the radiation stress concept can be successfully applied to simulate the
same pattern of wave transformation, and the results agreed qual- random waves with directional spectrum.
itatively well with the measurements. However, when the wave- Based on the numerical results simulated using various wave-
induced current was intentionally neglected, the wave height current interaction models, it is confirmed again that wave-induced
distribution patterns along the transect 4 were completely reversed currents play a significant roll in the transformation of waves breaking
from the measured ones. Thus, the effect of wave-induced current on locally in the area of interest. For the particular situation when
the transformation of waves was confirmed again. The details on the transverse interference patterns are present due to refraction of
performance of the three wave-current interaction models, however, monochromatic waves, the interference patterns cannot be simulated
were different. The FUNWAVE model gave a perfect agreement with in the SWAN-SHORECIRC model and their radiation stresses are not
the measured wave height behind the shoal, and a jet-like current was correctly evaluated in the REF/DIF-SHORECIRC model. For this
generated. The SWAN-SHORECIRC model is less accurate in the simu- situation models based on primitive variables approach such as
lation of wave height, but a jet-like current was generated behind the FUNWAVE should be a better choice at the expense of computational
shoal as in the case of FUNWAVE. On the other hand, the REF/DIF- effort. However, for random waves of directional spectra, the two
SHORECIRC model does not give an organized jet-like current. Thus, combined wave and current models worked reasonably well.
312 J. Choi et al. / Coastal Engineering 56 (2009) 297–312

Acknowledgements Kirby, J.T., 1986. Rational approximations in the parabolic equation method for water
waves. Coastal Engineering 10, 335–378.
Kirby, J.T., Dalrymple, R.A., 1994. Combined refraction/diffraction model REF/DIF 1,
We sincerely thank M. J. Briggs of Coastal and Hydraulics Labora- version 2.5, User Manual. Technical Report CACR-94-22, University of Delaware.
tory of the U.S. Army Engineer Research and Development Center for Kirby, J.T., Özkan, H.T., 1994. Combined refraction/diffraction model for spectral wave
providing us with the experimental measurements that were used in conditions, REF/DIF S, version 1.1, User Manual. Technical Report CACR-94-04, University
of Delaware.
our study. We greatly appreciate the constructive comments from the Kirby, J.T., Wei, G., Chen, Q., Kennedy, A.B., Dalrymple, R.A., 1998. Fully nonlinear
two anonymous reviewers on an early version of the manuscript. Boussinesq wave model, User Manual. Technical Report CACR-98-06, University of
Delaware.
Larsen, J., Dancy, H., 1983. Open boundaries in short wave simulations—a new approach.
References
Coastal Engineering 7, 285–297.
Longuet-Higgins, M.S., Stewart, R.W.,1962. Radiation stresses and mass transport in gravity
Battjes, J.A., Janssen, J.P.F.M., 1978. Energy loss and set-up due to breaking of random waves with applications to surf-beats. Journal of Fluid Mechanics 13, 481–504.
waves. Proceedings of the 16th International Conference Coastal Engineering. ASCE, Longuet-Higgins, M.S., Stewart, R.W., 1964. Radiation stresses in water waves; a physical
pp. 569–587. discussion with applications. Deep-Sea Research 11, 529–562.
Battjes, J.A., Stive, M.J.F., 1985. Calibration and verification of a dissipation model for Mei, C.C., 1989. The applied dynamics of ocean surface waves. World Scientific 451-485.
random breaking waves. Journal of Geophysical Research 90 (No. C5), 9159–9167. Özkan, H.T., Kirby, J.T., 1993. Evolution of breaking directional spectral waves in the near-
Booij, N., 1981. Gravity Waves on Water with Non-uniform Depth and Current, Doctoral shore. Ocean Wave Measurement and Analysis (Proc. WAVES’93). ASCE, pp. 849–863.
dissertation, Technical University of Delft, The Netherlands, pp. 131. Putrevu, U., Svendsen, I.A., 1999. Three-dimensional dispersion of momentum in wave-
Booij, N., Ris, R.C., Holthuijsen, L.H., 1999. A third-generation wave model for coastal induced nearshore currents. European Journal of Mechanics-B/Fluids 409~427.
regions 1. Model description and validation. Journal of Geophysical Research 104 Smagorinsky, J., 1963. General circulation experiments with the primitive equations I.
(No. C4), 7649–7666. The basic experiment. Monthly Weather Review 91, 99–165.
Booij, N., Haagsma, I.J.G., Holthuijsen, L.H., Kieftenburg, A.T.M.M., Ris, R.C., van der Suh, K.D., Dalrymple, R.A., 1993. Application of an angular spectrum model to simu-
Westhuysen, A.J., Zijlema, M., 2004. SWAN Cycle III version 40.41, User Manual. lation of irregular wave propagation. Journal of Waterway, Port, Coastal and Ocean
Delft University of Technology. Engrg 119 (5), 505–520.
Borgman, L.E., 1984. Directional spectrum estimation for the Sxy gages. Technical Svendsen, I.A.,1984. Mass flux and undertow in a surfzone. Coastal Engineering 8, 347–365.
Report. Coastal Engineering Reseach Center, Vicksburg. Svendsen, I.A., Hass, K., Zhao, Q., 2002. Quasi-3D Nearshore Circulation Model SHORECIRC
Bouws, E., Gunther, H., Rosental, W., Vincent, C.L., 1985. Similarity of the wind wave version 2.4, User Manual. Technical Report, University of Delaware.
spectrum in finite depth water, Part I-spectral form. Journal of Geophysical Research 90 Thornton, E.B., Guza, R.T., 1983. Transformation of wave height distribution. Journal of
(C1), 975–986. Geophysical Research 88 (C10), 5925–5938.
Chen, Q., Kirby, J.T., Dalrymple, R.A., Kennedy, A.B., Chawla, A., 2000. Boussinesq modeling Vincent, C.L., Briggs, M.J., 1989. Refraction- diffraction of irregular waves over a mound.
of wave transformation, breaking and runup. II: two horizontal dimensions. Journal of Journal of Waterway, Port, Coastal and Ocean Engineering 115 (2), 269–284.
Waterway, Port, Coastal and Ocean Engineering 126 (1), 48–56. Wei, G., Kirby, J.T., Grilli, S.T., Subramanya, R., 1995. A fully nonlinear Boussinesq model for
Chen, Q., Kirby, J.T., Dalrymple, R.A., Shi, F., Thornton, E.B., 2003. Boussinesq modeling of surface waves. Part 1: highly nonlinear unsteady wave. Journal of Fluid Mechanics 294,
longshore current. Journal of Geophysical Research 108 (C11), 26–1-26-18. 71–92.
Choi, J., Lim, C.-H., Jeon, Y.-J., Yoon, S.B., 2007. Numerical simulation of irregular wave Wei, G., Kirby, J.T., Sinha, A., 1999. Generation of waves in Boussinesq models using a
transformation due to wave induced current. Journal of Hydro-Environment Research. source function method. Coastal Engineering 36, 271–299.
doi:10.1016/j.jher.2007.08.001. Wood, D.J., Muttray, M., Oumeraci, H., 2001. The SWAN model used to study wave evolution
Goda, Y., 2004. A 2-D random wave transformation model with gradational breaker in a flume. Ocean Engineering 28, 805–823.
index. Coastal Engineering Journal 46 (1), 1–38. Yoon, S.B., Cho, Y.-S., Lee, C., 2004. Effects of breaking-induced currents on refraction-
Hirakuchi, H., Maruyama, K.,1986. An extension of the parabolic equation for application to diffraction of irregular waves over submerged shoal. Ocean Engineering 31, 633–652.
obliquely incident waves. Proc. 33rd Japanese Conf. Coastal En, pp. 114–118. Yoon, S.B., Lee, J.W., Yeon, Y.J., Choi, B.H., 2001. A note on the numerical simulation of
Horikawa, K., Kuo, C.-T., 1966. A study of wave transformation inside surfzone. Proceedings wave deformation over a submerged shoal. Proc. 1st Conf. Asian and Pacific Coastal
of the 10th Conference on Coastal Engineering, Tokyo, pp. 217–233. Eng. Dalian, China, pp. 315–325.
Holthuijsen, L.H., Herman, A., Booij, N., 2003. Phase-decoupled refraction–diffraction for
spectral wave models. Coastal Engineering 49, 291–305.
Kennedy, A.B., Chen, Q., Kirby, J.T., Dalrymple, R.A., 2000. Boussinesq modeling of wave
transformation, breaking, and runup. I: 1D. Journal of Waterway, Port, Coastal and
Ocean Engineering 126, 39–47.

Vous aimerez peut-être aussi