Coastal Engineering 142 (2018) 95–105

Contents lists available at ScienceDirect

Coastal Engineering
journal homepage: www.elsevier.com/locate/coastaleng

Laboratory validation of SWASH longshore current modelling T

∗
Rong Zhang , Marcel Zijlema, Marcel J.F. Stive
Department of Hydraulic Engineering, Delft University of Technology, Delft, the Netherlands

A R T I C LE I N FO A B S T R A C T

Keywords: In this paper, the ability of the numerical phase resolving model SWASH (Simulating WAves till SHore) to
Longshore current hindcast wave-induced longshore currents is evaluated. Using default settings for all processes modelled, highly
SWASH accurate results are found for wave heights, mean water levels and longshore currents. While wave current
Numerical modelling interaction is intrinsically modelled, insights into the spatial variation of wave driven longshore currents are
found. Additionally, vertical variations of modelled longshore currents have been compared. Depth uniform
profiles of longshore current within surf zone are noted on plane beaches under regular waves, except for minor
deviations near the shoreline. The apparent validity of a depth-uniform longshore current encourages the use of
a depth-averaged moment balance equation to compute the longshore current. A simpler model is shown to also
be able to predict a proper magnitude of longshore current, although the cross-shore distribution – in contrast
with SWASH - needs tuning for the eddy viscosity and the bottom friction coefficient, since the distribution of the
wave-induced longshore current heavily depends on lateral mixing.

1. Introduction possible non-uniformity of the longshore current along the beach

should be carefully considered (Visser, 1991). To avoid non-uniformity
Longshore currents, generated by obliquely incident breaking of the longshore current, well controlled, highly representative la-
waves, play an important role in driving longshore sediment transport boratory experiments were conducted by Visser
and in changing coastal morphology. The theory of longshore currents (1980,1982,1984a,1984b,1991). Visser found that a virtually uniform
has progressed importantly (Bowen, 1969; Longuet-Higgins, 1970a,b; longshore current is realized when the return flow in the offshore re-
Thornton and Guza, 1986), since the concept of radiation stress was gion of the wave basin is minimal. Therefore, Visser introduced an
introduced (Longuet-Higgins and Stewart, 1964; Longuet-Higgins, active external recirculation system driven by pumps to minimize the
1970a,b). Among the proposed mathematical models, the simplest circulation flow in the wave basin. Because of the well-controlled uni-
model is a 1D model based on momentum flux balance in alongshore formity of the longshore current, this data set has been used as a
direction on a planar beach under monochromatic wave incidence. In benchmark set for developing longshore current theories and validating
steady state and water depth averaged mode, the longshore current numerical models (e.g. Smith et al., 1993; Svendsen and Putrevu, 1994;
driving force is balanced by bottom friction and lateral friction. Al- Chen, 2003; Rijnsdorp et al., 2017; Hsu et al., 2017). This experimental
though the average longshore current velocity across the entire surf example setup, guaranteeing the production of an alongshore uniform
zone could be properly predicted, the exact cross-shore distribution of longshore current in the laboratory, was also adopted by Hamilton and
the longshore current is more difficult to derive, since it highly depends Ebersole (2001). In addition, a unique, unexplored and unpublished
on less-known lateral mixing mechanisms. data set produced by large-scale, outdoor laboratory experiments
To investigate longshore current mechanisms, field investigations (Hulsbergen and ter Horst, 1973) is presented and shown to be a
and laboratory experiments have been carried out. Earlier measure- worthy, additional dataset for validation. Other than the previously
ments are often limited to longshore maximum velocity or average mentioned experiments investigating wave-induced longshore currents
velocity across the entire width of the surf zone, which does not reveal on planar beaches, Reniers and Battjes (1997) carried out experiments
the cross-shore distribution of longshore current. The known first cross- which focused on longshore currents over more geographically com-
shore distribution of longshore current was measured by Galvin and plicated, barred beaches.
Eagleson (1964) along several shore normal profiles. These pioneering The measurements of longshore currents obtained from the above-
data are used widely to test longshore current theories. However, mentioned experiments provide an excellent database to validate and/

∗
Corresponding author.
E-mail address: [email protected] (R. Zhang).

https://doi.org/10.1016/j.coastaleng.2018.10.005
Received 22 April 2018; Received in revised form 28 September 2018; Accepted 13 October 2018
Available online 16 October 2018
0378-3839/ © 2018 Elsevier B.V. All rights reserved.
R. Zhang et al. Coastal Engineering 142 (2018) 95–105

or calibrate numerical simulation models. The primary aim of this ζ ζ

∂ζ ∂
paper is to validate the ability to simulate the spatial variation of wave- ∂t
+
∂x
∫ udz + ∂∂y ∫ vdz = 0
−d −d (2.7)
driven longshore currents by the phase resolving, non-hydrostatic nu-
merical model SWASH (Simulating Wave till SHore). Validation is Where t is the time, ζ is the free surface elevation from still water level,
preferred over calibration, since validation relies on default parameter z = -d is the bottom, d is the still water depth.
settings while calibration relies on adjusted parameter settings for a The dynamic boundary condition at the bottom is confined to
specific dataset. A verified (validated or calibrated) model is useful to bottom friction. The bottom friction stress is based on a quadratic
design hard and soft coastal interventions and/or instrument deploy- U U
friction law τb = Cf h , where h = ζ + d is the total water depth, U is
ment in experiments. Additionally, a very simplified 1D model is in- the depth averaged velocity, and Cf is a dimensionless friction coeffi-
troduced to explore the most basic physics behind wave-induced cient. At the free surface, zero atmospheric pressure ( ph = pnh = 0 ) is
longshore currents. assumed and no surface stresses are considered.
In the present work, we set up the wave-flow model SWASH with The turbulent stresses are given based on eddy viscosity closure
default parameters to validate its ability to simulate wave-induced equations. The horizontal viscosity and vertical viscosity are estimated
longshore currents on a laboratory scale. The governing equations of by the Smagorinsky type model (Smagorinsky, 1963) and the k- ε model
the SWASH model are listed in subsection 2.1. Then, a separate sim- (Launder and Spalding, 1974), respectively.
plified depth averaged 1D model is demonstrated in subsection 2.2. The For a comprehensive description of the SWASH model and its nu-
configurations of the chosen laboratory experiments are described in merical schemes, reference is made to Zijlema et al. (2011), Smit et al.
the following subsection 2.3. The results of SWASH modelling long- (2013) and Rijnsdorp et al. (2014). Extensive validations and applica-
shore currents are presented and compared to the 1D model results in tions of SWASH model can be found in the literature. For instance,
Section 3. Finally, in Section 4 a discussion and in Section 5 conclusions Rijnsdorp et al. (2014) evaluated the ability of SWASH model to si-
are given. mulate infragravity wave dynamics and found a good agreement with
flume observations. de Bakker et al. (2016) designed a numerical study
2. Methods using the SWASH model to investigate nonlinear energy transfers be-
tween waves, especially focusing on beach steepness effects on non-
2.1. Numerical model SWASH linear infragravity wave interactions. Suzuki et al. (2017) validated the
capability of SWASH model to estimate wave overtopping for im-
SWASH (Zijlema et al., 2011) is a flow-wave model which can si- permeable coastal structures against physical model data.
mulate unsteady, non-hydrostatic free-surface flow in the time-domain,
having a wide application range in coastal waters. The governing
equations are non-linear shallow water equations including non-hy- 2.2. 1D model
drostatic effects. The local continuity equation and momentum equa-
tions in 3D configuration are given as The depth averaged 1D model comprises a wave model and a cur-
rent model. The wave model is based on the wave energy balance
∂u ∂v ∂w equation in the cross-shore direction including the wave roller effect,
+ + =0
∂x ∂y ∂z (2.1) and the current model is based on the momentum balance equation in
∂τxy the alongshore direction. In contrast to the phase resolving SWASH
∂u ∂uu ∂uv ∂uw 1 ∂ph + pnh ∂τ ∂τxz
+ + + + = xx + + model in multi-layered mode, the 1D model does not resolve the free
∂t ∂x ∂y ∂z ρ ∂x ∂x ∂y ∂z (2.2)
flow surface, and the vertical flow structures. The 1D model thus is a
∂v ∂vu ∂vv ∂vw 1 ∂ph + pnh ∂τyx ∂τyy ∂τyz very simplified model, expected however to include the basic physics
+ + + + = + + behind the process of wave-induced currents. A comparison with the
∂t ∂x ∂y ∂z ρ ∂y ∂x ∂y ∂z (2.3)
SWASH model will reveal whether including a physically more so-
∂w ∂wu ∂wv ∂ww 1 ∂ph + pnh ∂τzx ∂τzy ∂τzz phisticated approach is more robust.
+ + + + +g= + +
∂t ∂x ∂y ∂z ρ ∂z ∂x ∂y ∂z
(2.4) 2.2.1. Wave modelling
Where u and v represent the horizontal velocity in x and y direction The governing equation determining the wave dynamics is the en-
respectively, x is the direction normal to the shore, and y the direction ergy balance equation of wave energy:
parallel along the shore. The moving free surface ζ (x , y, t ) and the
d
stationary bottom z = -d(x,y) restrict the water column. The hydro- (Ew cg cos θ) = εw
dx (2.8)
static pressure is explicitly expressed as p h = ρg (ζ − z ) , so ∂x ph = ρg ∂x ζ
(where ∂x stands for ∂/ ∂x ), ∂y ph = ρg ∂y ζ and ∂z ph = −ρg (where g is the where x is the shore normal direction, E w is wave energy, ε w is mean
gravitational acceleration). pnh is the non-hydrostatic pressure part. τij organised wave energy dissipation rate due to wave breaking and
are the turbulent stresses (where i or j denotes the coordinates). bottom friction. cg is wave group velocity. Compared to dominant wave
Equation (2.1) is the local continuity equation, and Equations (2.2)- breaking dissipation, bottom frictional dissipation is negligible, except
(2.4) are momentum equations including the effects of mixing and ex- the very shallow run-up region (Thornton and Guza, 1983). For a
ternal forces. monochromatic wave, using linear wave theory, the wave energy is
The kinematic boundary conditions at the free surface and bottom expressed by the wave height:
are
1
∂ζ ∂ζ ∂ζ Ew = ρgH 2
w|z = ζ = +u +v 8 (2.9)
∂t ∂x ∂y (2.5)
and the group velocity cg is described by
∂d ∂d
w|z =−d = −u −v 1 kh
∂x ∂y (2.6) cg = c ⎛ + ⎞
⎝2 sinh 2 kh ⎠ (2.10)
Integrating the local continuity Equation (2.1) with substituting
kinematic boundary conditions at surface and bottom Equations. The wave energy dissipation ε w serves as a source term in the roller
(2.5)–(2.6), a global continuity equation is induced model (Stive and De Vriend, 1994)

96
R. Zhang et al. Coastal Engineering 142 (2018) 95–105

d The bottom friction stress is determined by the quadratic law:

− εw − (2Er c cos θ) = cτr
dx (2.11) ⇀
τb = ρCf < V U > (2.20)
where τr is the shear stress between roller and water interface, for a
⇀
steady roller it is given by Duncan (1981). where V is depth-averaged longshore current velocity, U is the in-
stantaneous total horizontal velocity vector and Cf is a friction coeffi-
A
τr = ρr g sin β cient:
L (2.12)
g
where A is cross-sectional area of the roller, β the slope of wave front, L Cf =
C2 (2.21)
wave length. ρr is density of the roller, generally less than undisturbed
sea water for the entrainment of air at the aerated wave surface. Here, 12h ⎞
C = 18log ⎛ ⎜ ⎟

having ρr ≈ ρ , to keep roller mass constant at a decrease in roller area ⎝ ks ⎠ (2.22)

A.
When the current velocity is much smaller compared to the wave
The second term of equation (2.9) is the gradient of roller energy
orbital velocity, the bottom friction stress τb can be linearized by the
flux, followed the definition by Svendsen (1984).
following equation (Liu and Dalrymple, 1978)
1 2A
Er = ρc 2
2 r L (2.13) τb = ρCf um V (1 + sin2 θ)
π (2.23)
The wave height transformation proposed by Thornton and Guza
where um is total velocity variance.
(1983) is utilised. The wave dissipation is modelled by the classical ⇀
The parametrisation of < V U > for random waves adopts the
periodic bore dissipation function as
formulation of Wright and Thompson (1983) in an empirical form is
f (BH )3 1
εw = ρg 22
4 h (2.14) ⇀ V
< V U > = σt V ⎜⎛α 2 + ⎛ ⎞ ⎞⎟
⎜ ⎟

in which B expresses the deviation of wave breaking dissipation from ⎝ ⎝ σt ⎠ ⎠ (2.24)

bore dissipation. Where α is a coefficient, the best fit for α is 1.16 by (Feddersen et al.,
For the ensemble, the average wave energy dissipation rate is cal- 2000).
culated by The wave-averaged and depth-integrated longshore momentum
∞ equation governing the longshore current is
f¯ B3
εw = ρg
4 h
∫ H 3pb (H ) dH dhτxy
0 (2.15) τb = Fy + ρ
dx (2.25)
where p b is the probability density of breaking wave distribution.
The Thornton and Guza (1983) model modified the distribution of where τxy is the horizontal turbulent stress.
random wave heights from a Rayleigh distribution by a weighting dV
function, the breaking wave dissipation is related to Hrms via: τxy = νt
dx (2.26)

⎡ ⎤ where νt is the eddy viscosity.

3 ⎢ ⎥
3 π 3 ¯ Hrms 1
εw = B f ρg M ⎢1 − ⎥
16 h 2 5/2 2.3. Laboratory experiments
⎢
⎢
⎣
⎛1 +
⎝
( )
Hrms
γh
⎞ ⎥
⎠ ⎥ ⎦ (2.16)
2.3.1. Visser (1991) experiments
H Visser (1991) conducted a series of experiments to measure the
⎤
M = 1 + tanh ⎡8 ⎜⎛ rms − 1⎟⎞ ⎥ spatial distribution of wave-induced longshore currents. To achieve a
⎢ γh
⎣ ⎝ ⎠⎦ (2.17) uniformity of the longshore current in the wave basin, Visser (1991)
Where Hrms is the root mean square wave height, and γ is the wave used a pumped recirculation in the experiments and fine-tuned the
breaker index, B denotes the intensity of wave breaking, is usually the optimal recirculation discharge to minimize the return flow. The high
order of 1, but should be larger cause some researchers thought the quality control makes these experimental data one of the most popular
classical hydraulic jump function underestimates the wave breaking calibration and/or validation sets for numerical models. Two cases of
dissipation (Thornton and Guza, 1983; Stive, 1984). According to the these experiments confined to regular waves and a 1:20 slope with a
fitting of the model results to laboratory and field data by Thornton and smooth concrete bottom were simulated, viz. Case 4 and Case 5 of
Guza (1983), the range of calibrated value of B is between 0.8 and 1.7. Visser (1991) experiments (in short V91_C4 and V91_C5, respectively)
The wave radiation stress in the longshore direction is given by The wave parameters are summarised in Table 1. The subscript “1”
refers to values at the wave generators. The cross-shore profile is shown
cg
Sxy, w = Ew cos θ sin θ in Fig. 1, (a).
c (2.18)
Assuming a longshore uniform beach, the driving force of the wave- 2.3.2. Hamilton et al. (2001) experiments
induced current is due to the cross-shore gradient of the longshore ra- Another suitable data set of measured wave-induced longshore
diation shear stress, including the roller contribution: current for validation was produced by Hamilton and Ebersole (2001)
in the wave basin of the Large-scale Sediment Transport Facility (LSTF).
sin θ d
Fy = − (Ew cg cos θ + 2Er c cos θ) The smooth concrete beach slope was 1/30 (Fig. 1, b). Case 6N with
c dx (2.19)
regular waves was simulated. Similar to Visser (1991), a multiple
pumping system was utilised to optimise the uniformity of longshore
2.2.2. Current modelling current.
The current in alongshore direction is determined by the alongshore
momentum equation. The above current driving force (Eq (2.19)) is 2.3.3. Hulsbergen et al. (1973) experiments
balanced by bottom friction and lateral friction. In the present model, a Although the purpose of the H73 experiments was to investigate the
depth-averaged mode is considered. interaction between pile screen groins and a wave-driven longshore

97
R. Zhang et al. Coastal Engineering 142 (2018) 95–105

Table 1
The parameters of the experiments.
Exp. wave H1 (m) T1 (s) Θ1 (⁰) m d1 (m) L1 (m) H1/L1 H1/d1 k1 d1 Beach type

V91_C4 Regular 0.078 1.02 15.4 1:20 0.350 1.46 0.053 0.22 1.51 plane
V91_C5 Regular 0.071 1.85 15.4 1:20 0.350 3.19 0.022 0.20 0.70 plane
H01_6N Regular 0.182 2.50 10.0 1:30 0.667 5.94 0.031 0.27 0.71 plane
H73_W Regular 0.030 1.04 15.0 1:35 0.300 1.45 0.021 0.10 1.30 plane
R97_SA243 Regular 0.08 1.00 30.0 1:20 0.55 1.53 0.052 0.15 2.26 barred
R97_SO014 Irregular 0.07 1.20 30.0 1:20 0.55 2.09 0.034 0.13 1.65 barred

resolved on the fine grid. Such a two-grid system reduces computational

efforts substantially by an order of magnitude. The validation of the
subgrid approach showed that the predictions by subgrid simulations
are comparable to that of fully resolved simulations (Rijnsdorp et al.,
2017).

3.1.1. Case 4
For Case 4 of V91 experiments, the time step was set at Δt = 0.005 s,
the grid resolution was set Δx = 0.03 m in cross-shore direction and
Δy = 0.044 m in alongshore direction (400 × 128 grid cells in total).
The obliquely incident regular waves with a period of 1.02 s, an am-
plitude of 0.039 m and an incident angle of 15.4⁰, were generated at the
offshore boundary. A periodic boundary was used at the lateral
boundaries to limit the length of an unbounded beach. In the laboratory
experiment, the bottom was made of smooth concrete. The roughness
height value of a smooth concrete bottom is suggested as 0.001 m by
Visser (1984a,b), 0.0005 m by Reniers and Battjes (1997), and
0.0004 m by Rijnsdorp et al. (2017). Hereinafter, the roughness height
was chosen at ks = 0.0005 m for this following cases as a default setting
which is qualified as validation and not calibration, since it is well
Fig. 1. The bottom profile of each experiment: (a), Visser (1991); (b), Hamilton within the default range.
and Ebersole (2001); (c), Hulsbergen and ter Horst, 1973; (d) Reniers and After a computational duration of 150 s, the current field reaches a
Battjes (1997). stationary state as demonstrated in Fig. 2, indicating that the differ-
ences between the longshore current profiles at t = 120 s and t = 150 s
current, the base case without groins provides an additional valuable, are relatively small. In Fig. 3, the computed instantaneous water level
validation/calibration data set. The wave only experiment was con- and the phase- and depth-averaged current field are shown. As waves
ducted in a wave basin of Delft Hydraulics. The bottom contours are approach the shore, waves are transforming due to shoaling effects and
straight and parallel to the shoreline. Regular waves with H = 0.03 m changing direction because of depth refraction. Then wave heights start
and T = 1.04 s were generated approaching the shore at an incidence to decrease once depth limited breaking occurs. It is clearly shown that
angle of 15 ⁰. there is a spatial lag between the location of maximal horizontal current
velocity and that of the maximal wave height.
The cross-shore variations of wave height, mean water level and
2.3.4. Reniers and Battjes (1997) experiments
longshore current are shown in Fig. 4. The blue lines denote the com-
Reniers and Battjes (1997) conducted laboratory experiments for
puted results, while the red circles are the measured V91 data. The
wave driven longshore currents on barred and non-barred beaches. As
simulation successfully captures the right location of wave breaking
the above mentioned four cases are all about plane beaches, two tests
point. The waves start to break at x = 5 m, 2 m offshore to the shoreline
with a barred beach (Fig. 1, d), test SA243 under regular waves and test
at x = 7 m. The wave heights are slightly overpredicted within the surf
SO014 under irregular waves, were chosen to be reproduced. The
zone, which suggests that the model may slightly underestimate the
height of the bar is about 10 cm, of which the inner slope is 1:8. The
plane slope of the beach seaward is 1:10 while the shoreward slope is
1:20. The water depth at the wave maker is 0.55 m. The wave para-
meters are listed in Table 1.

3. Results

3.1. V91 experiments

To verify the ability of the SWASH model in simulating wave-in-

duced longshore currents, a set of numerical simulations was conducted
to compare to Case 4 and Case 5 of the Visser (1991) experiments. The
number of vertical layers was set at 20, which is fine enough to resolve
the flow vertical structure. To achieve a more efficient computation Fig. 2. The cross-shore profile of computed longshore current at t = 60 s (blue
time, the so-named subgrid method (Rijnsdorp et al., 2017) was applied line), t = 90 s (red line), t = 120 s (yellow line), t = 150 s (purple line). (For
to derive the vertical accelerations and pressure gradients on a coarser interpretation of the references to colour in this figure legend, the reader is
grid of 4 layers. The remaining wave and mean flow dynamics are referred to the Web version of this article.)

98
R. Zhang et al. Coastal Engineering 142 (2018) 95–105

Fig. 5. Case 5 (H = 0.071 m, T = 1.85 s, θ = 15.4⁰) of V91 experiments; the

cross-shore distribution of wave height (a), wave set up (b) and longshore
current velocity (c), (blue lines: simulations, red circles: measurements). (For
interpretation of the references to colour in this figure legend, the reader is
referred to the Web version of this article.)

Fig. 3. Case 4 (H = 0.078 m, T = 1.02 s, θ = 15.4⁰) of V91 experiments; (a)

snapshot of the computed wave field (the unit of the colour bar is m) and (b) the 3.1.2. Case 5
phase-averaged and depth-averaged current field. (For interpretation of the For this case, with a longer wave period of 1.85 s, the grid resolution
references to colour in this figure legend, the reader is referred to the Web could be increased to Δx = 0.05 m and Δy = 0.087 m (resulting in 240
version of this article.) × 140 grid cells in total). Other parameters are kept the same as in Case
4. The computed wave height is nearly the same as the measured data
(Fig. 5). The waves started to break at x = 4.5 m, 2.5 m offshore to the
shoreline at x = 7 m. Although the longshore current velocity is slightly
overestimated, its cross-shore profile is well reproduced.

3.2. H01_6N experiment

In this simulation the time step was set at Δt = 0.005 s, the grid
resolution was set Δx = 0.08 m and Δy = 0.16 m (275 × 225 grid cells
in total). The roughness height was set at ks = 0.0005 m, the same as
the default value. The number of vertical layers was set at 20, which is
found fine enough to resolve the vertical variation of mean flow
structure from previous simulation experience, while the coarse subgrid
with 4 pressure layers was turned on.
The comparable simulation results with the measurement data are
shown in Fig. 6. The simulation accurately reproduces the wave height
transformation. The finer resolution of the numerical model predicts
Fig. 4. Case 4 of V91 experiments; the cross-shore distribution of wave height
the wave breaking point at x = 10.5 m, while the measured maximal
(a), wave set up (b) and longshore current velocity (c), (blue lines: simulations, wave height is at x = 9.48 m and the next measurement point is beyond
red circles: measurements). (For interpretation of the references to colour in 10.5 m. The wave height variation cross-shore shows a good agreement
this figure legend, the reader is referred to the Web version of this article.) with the observed values (Fig. 6, (a)). Both the locations of the wave
set-up and the maximal longshore current are accurately computed.
Similarly to the V91_C4 simulation, within the surfzone, the wave set-
energy dissipation rate. This observed deviation is similar to the results
up is slightly overestimated while the longshore current is slightly un-
of the same case simulated by Chen, 2003, who found the same over-
derestimated. This pattern of deviations between simulations and
prediction using a phase-resolving Boussinesq type model. Never-
measurements of H01_6N and V91_C4 experiments may be due to their
theless, the computed wave set up and longshore current velocity agree
more similar wave steepness.
very well with the measured data. The slight overestimation of the
wave set up is consistent with the trend of the computed wave height,
and may be partly due to a minor underestimation of the longshore 3.3. H73_W experiment
current velocity.
As shown in Fig. 4 c, the model quite accurately predicts the cross- In addition to the above validations against the well-known V91 and
shore variation of the longshore current as well as the location and H01 experiments, a unique, not well-known, but valuable data set is
magnitude of the maximum velocity. Overall, the well-matched results obtained from (Hulsbergen and ter Horst, 1973). As part of their
reveal the validity of the wave-breaking scheme and of the transfor- comprehensive measurements of the impact of groins on the longshore
mation of momentum flux in both longshore and cross-shore directions. current, here their wave only case is used for additional validation.
To reproduce the experiments numerically, the time step was set at
Δt = 0.005 s, the grid resolution was set Δx = 0.03 m and

99
R. Zhang et al. Coastal Engineering 142 (2018) 95–105

Fig. 6. Case H01_6N (H = 0.182 m, T = 2.5 s, θ = 10⁰); the cross-shore dis-

tribution of wave height (a), wave set up (b) and longshore current (c), (blue
lines: simulations, red circles: measurements). (For interpretation of the refer-
ences to colour in this figure legend, the reader is referred to the Web version of
this article.)

Δy = 0.044 m (resulting in 380 × 128 grid cells). The roughness height

was set at ks = 0.0005 m, again in the default range. The vertical ve-
locity layers were chosen 20 while the coarse pressure layers were 4. Fig. 8. Case R97_SA243 (H = 0.08 m, T = 1 s, θ = 30⁰); the cross-shore dis-
For this condition, the computed maximal alongshore velocity is tribution of wave height (a), wave set up (b), longshore current (c) and the
0.14 m/s at x = 6.5 m (Fig. 7c). The mean longshore current velocities bottom profile (d), (blue lines: simulations, red circles: measurements). (For
show satisfactory agreement with the measured data, but the velocities interpretation of the references to colour in this figure legend, the reader is
are slightly underestimated. Because of the crude assumption that the referred to the Web version of this article.)
vertical profiles of the longshore currents within the surf zone are
uniform, the measured velocities by float points in the upper water level are accurate. The computed maximum wave set-up reaches a
column might need to be interpreted as depth-averaged velocities, value of 0.0092 m, which is approximately equal to 0.0075 m computed
which may partially explain the underestimation. The calculated from linear wave theory.
breaker point is at about x = 5.5 m (Figs. 7a), 2 m seaward to the
shoreline at x = 7.5 m, consistent with the observation in the experi-
ment (Hulsbergen and ter Horst, 1973). Wave heights and set-up were 3.4. R97 experiments
not measured, but as the longshore current velocities are much more
sensitive to the parameters than the wave heights and mean water level, 3.4.1. Case SA243
we have confidence that the simulated wave heights and mean water As the above all cases are about plane beaches, a more complex
barred slope, of the experiment SA243 conducted by Reniers and
Battjes (1997), was considered. The case Re97_SA243 was simulated
and compared with experimental data. The numerical domain consists
of 440 grids (Δx =0.03 m) in cross-shore direction and 100 grids
(Δy =0.031 m) in alongshore direction. Other parameters and settings
are the same as the default.
It is shown (Fig. 8) that the locations of the wave breaking, wave set
down and maximal longshore current are all accurately calculated. The
waves were starting breaking on the outer bar slope when the water
depth is rapidly reduced. Then the spatially lagged maximal longshore
current appeared on the bar crest. The observed bimodal shape of
longshore current was also reproduced, where one peak location is at
the bar crest and the other peak is near the shoreline. The second peak
was not measured but visually observed by dye movement in the ex-
periment due to that the limited water depth (< 3 cm) is beyond the
usage range of measurement apparatuses (Reniers and Battjes, 1997).

3.4.2. Case SO014

Unlike the previous cases all tested under regular waves, case
Fig. 7. Case H73_W (H = 0.03 m, T = 1.04 s, θ = 15⁰); the cross-shore dis- SO014 was tested with irregular waves. The numerical grid resolution is
tribution of wave height (a), wave set up (b) and longshore current (c), (blue Δx = 0.041 m in cross-shore direction and Δy = 0.084 in alongshore
lines: simulations, red circles: measurements). (For interpretation of the refer- direction. Other parameters and settings are kept the same as previous
ences to colour in this figure legend, the reader is referred to the Web version of simulations. Again, a good agreement between modelled and measured
this article.) longshore current velocities was obtained (Fig. 9).

100
R. Zhang et al. Coastal Engineering 142 (2018) 95–105

settings.

3.6. Validated and predicted vertical variations of longshore currents

Given the availability of measurement data, only the vertical var-

iations of longshore currents of experiment H01_6N could be validated
(Fig. 11). The calculated longshore current velocities V(z) matched well
with observed values at five different locations, except for an under-
estimation at the nearest location (X = 1.16 m offshore). From the
furthest offshore location (X = 8.52 m offshore where h = 0.274 m) to
the third nearest to shoreline location (X = 4.12 m offshore where
h = 0.149 m), little vertical variations of the velocity profiles were
observed (Fig. 11). Therefore, the vertical profiles can be characterised
as depth uniform. At a further inshore location (X = 2.76 m), the tur-
bulent bottom boundary layer becomes dominant, and the vertical
profile changes from a rather depth uniform to a logarithmic profile.
To investigate the vertical variation of longshore current for the
other five experiments, the depth variation of the predicted normalised
longshore current v̄/c is illustrated in Fig. 12, where c is the local
shallow wave celerity. The vertical profile locations in the figure are
ranging from 0.4 to 1.4 times the width of the surf zone, increasing at a
0.2 times width interval. At 1.4 times xb, the longshore current is almost
zero for all cases. Similarly, the normalised longshore currents reach
their largest value halfway across the breaker zone on planar beaches of
the first four experiments. Because of the stronger bottom friction near
the shoreline, a slight deviation from a depth uniform profile arises,
Fig. 9. Case R97_SO014 (Hs = 0.07 m, Tp = 1.2 s, θ = 30⁰); the cross-shore
where the normalised longshore current at the upper water column is
distribution of wave height (a), wave set up (b), longshore current (c) and the
bottom profile (d), (blue lines: simulations, red circles: measurements). (For
slightly larger than that near bottom. Further offshore, the profiles of
interpretation of the references to colour in this figure legend, the reader is the longshore current are nearly vertical, verifying that the longshore
referred to the Web version of this article.) current is rather depth uniform within most breaker zones under reg-
ular waves on planar beaches.
The generally depth uniform trend is supported by similar ob-
3.5. Analysis of modelled longshore currents
servations in the Hamilton and Ebersole (2001) and Visser (1991) la-
boratory experiments, and in the Sandy Duck field experiments (Reniers
To evaluate the performance of the SWASH model, three statistical
et al., 2004). One possible reason would be the breaking wave induced
parameters, i.e. root mean square error (RMSE), relative bias (the sum
turbulence injected downward, smoothing the vertical current profile
of errors over the sum of observed values) and correlation coefficient
(for instance, Fig. 1 in Svendsen and Lorenz, 1989, and Fig. 1 in Church
(R2) are computed for all six numerical simulations. The results are
and Thornton, 1993). Another possible reason would be the dispersive
listed in Table 2. For both wave height and longshore current compu-
mixing by the interaction of mean longshore currents and cross-shore
tations, the root mean square errors are confined to a relatively small
currents (Putrevu and Svendsen, 1993). Due to the strong influences of
value. All correlation coefficients are larger than 0.97, revealing a
dispersion and turbulence on the vertical variations of longshore cur-
strong correlation between the simulated results and measurements.
rent profiles, the boundary layer would not be fully developed with the
For the specific interest in modelling longshore currents of this study,
existence of breaking waves. It is shown that the generally depth uni-
comparisons between modelled and observed values are depicted in
form longshore current profile is confined to regular waves on planar
Fig. 10. The relative small RMSE and strong correlation coefficient
beaches. When the other two experiments with a barred slope were
validate the capacity of modelling longshore currents of SWASH with
considered (Fig. 12), it is found that the vertical profiles tend to be more
just default set parameters on both barred and non-barred beaches.
uniform under random waves (Re97_SO014) than under regular waves
However, the much larger relative bias of R97_SO014 shows that the
(Re97_SA243). However, the vertical profiles of longshore currents are
swash model results are much worse for irregular waves than the results
much more logarithmic over the bar.
of other experiments under regular waves with default parameter
3.7. SWASH model vs 1D model
Table 2
Statistics of the computed bulk parameters against the measurements. If the very shallow zone is not considered and the vertical variation
2
of the cross-shore current is not of interest, a simple 1D longshore
H RMSE (m) R Relative bias (%)
momentum balance model is shown sufficient to compute the longshore
V91_C4 0.0086 0.9798 9,08 current. This phase-averaged 1D model complemented, is used to test
V91_C5 0.0056 0.9931 1,16 the H73_W experiment, Re97_SA243 and Re97_SO014 experiments,
H01_6N 0.0188 0.9728 5,25 compared with the validated simulation by the SWASH model. To ac-
R97_SA243 0.0039 0.9968 −6,34
curately predict the peaks in the cross-shore distribution of longshore
R97_SO014 0.0085 0.9948 20,30
V̄ RMSE (m/s) R2 Relative bias (%) current and wave setup and set down, the roller model developed by
V91_C4 0.0415 0.9700 −9,76 Dally and Brown (1995) is included. The additional roller model has
V91_C5 0.0348 0.9815 −11,33 been affirmed necessary to shift the computed maximum longshore
H01_6N 0.0396 0.9871 −11,13 current landward of the breaker point (Osiecki and Dally, 1996;
H73 0.0195 0.9707 −25,07
Reniers, 1999), to bridge the deviation of the numerical models from
R97_SA243 0.0582 0.9593 5,55
R97_SO014 0.0703 0.9523 −38,69 the measurements (Visser, 1984a,b; Reniers and Battjes, 1997).
The calculated results of the H73_W experiment are shown in

101
R. Zhang et al. Coastal Engineering 142 (2018) 95–105

Fig. 10. Comparison of modelled and observed longshore current velocities for each experiment showing RMSE, Bias, and R2 values. Black solid lines represent
perfect agreement.

Fig. 13. For the 1D model, the wave set up is the most insensitive to the 4. Discussion
free parameters. In contrast, wave height is sensitive to the wave
breaker index γ and the wave breaking fraction parameter B. The well The ability of the SWASH model in simulating longshore currents
calibrated γ is found to be 0.78. The value 1.75 of B is used which is the has been investigated and validated on barred and non-barred beaches
optimal value obtained by fitting to a laboratory data set (Power et al., under obliquely incident regular and irregular waves. Since the purpose
2013). Besides, the variation of the wave front angle β of the roller of this study is to validate the SWASH model with uncalibrated default
model and the viscosity dominate the cross shore profile of the long- free parameters, the parameters (e.g. bottom roughness height, hor-
shore current. The value of β is chosen 0.1 while the eddy viscosity is izontal eddy viscosity) were set the same for all the simulation cases.
taken as 0.01 m2/s. The bottom roughness height, which significantly Without any variation of the free parameters, the SWASH model pro-
affects the magnitudes of longshore current velocities, was calibrated to vided promising results. It can be expected that after calibration, the
be 0.0015 m. The calibrated bottom roughness value used in the 1D accuracy of the SWASH model could even be improved. For instance,
model is three times the recommended empirical value 0.0005 m used calibrating bottom friction coefficients would significantly affect mod-
in the SWASH model. elled longshore current magnitudes, for which there is a large varia-
In Fig. 14, the 1D model provides good results for the Re97_SA243 bility found in the literature.
experiment on a barred slope under regular waves, when the bottom The simulation results of the longshore currents on a barred slope
roughness height increased to 0.01 m. For the experiment Re97_SO014 (Re97_SA243 and Re97_SO014) are consistent with that measured in
on the same slope but under irregular waves, a good agreement was the laboratory. Confined to a well-controlled laboratory environment,
obtained (shown in Fig. 15) when the roller slope parameter was both the modelled and observed longshore currents have the same
changed from 0.1 to 0.05, and the bottom roughness height 0.0005 m cross-shore variations. The pattern is that the maximal longshore cur-
was used. rent occurs where the wave breaking induced turbulence is the

102
R. Zhang et al. Coastal Engineering 142 (2018) 95–105

Fig. 11. Case H01_6N, the comparison of modelled and observed vertical variations of V(z), (blue line: simulation, red circle: measurement). (For interpretation of
the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 12. The vertical profiles of normalised longshore shore current v¯/ c , c(= gd ) is local wave celerity. The vertical black dash line denotes the vertical profile
location and indicates the zero value of each vertical profile of v̄ / c (blue line). The dash dot and solid black lines represent the mean water level and slope bottom,
respectively. The scale is given in Re97_SO014 figure. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of
this article.)

strongest. The second peak velocity appears near the shoreline. lateral mixing significantly affects the maximum longshore current
However, the field experiment DELILAH shows that the maximum velocity position. It was concluded that the maximal longshore current
longshore current velocity is on the trough but not over the bar crest. velocities observed in the trough are not attributed to the pure local
Reniers and Battjes (1997) attempted to test the effects of alongshore wave forcing (Reniers and Battjes, 1997). The deviation of observed
variation and lateral mixing of their numerical model on the position of and predicted maximum velocity position may due to the failure to
the maximum. However, neither an alongshore pressure gradient nor identify the alteration of the local forcing mechanism (Church and

103
R. Zhang et al. Coastal Engineering 142 (2018) 95–105

Fig. 13. Comparison between the simulated results of the H73_W experiment
by the 1D model (blue line) and the results obtained by the SWASH model
(black dash line), (red circles: measurements). (For interpretation of the re-
ferences to colour in this figure legend, the reader is referred to the Web version
of this article.)

Fig. 15. Comparison between the simulated results of the R97_SO014 experi-
ment by the 1D model (blue line) and the results obtained by the SWASH model
(black dash line), (red circles: measurements). (For interpretation of the re-
ferences to colour in this figure legend, the reader is referred to the Web version
of this article.)

all the time and then needs a calibration. For instance, the calibrated
bottom roughness height varies a factor of twenty for the simulations on
a barred slope. In addition, the cross-shore distribution of the longshore
current depends on the eddy viscosity value in order to mimic the
horizontal momentum flux diffusion, which is generally unknown.
Different correlations between the eddy viscosity, characteristic length
scale (i.e. wave height or water depth), and characteristic velocity (i.e.
wave celerity or turbulent velocity) are available in the literature (i.e.
Longuet-Higgins, 1970b; Battjes. J.A., 1975, Putrevu and Svendsen,
1993). The simplest constant eddy viscosity was used in the simulations
with a calibration, instead of choosing an available eddy viscosity
model. While this finding illustrates the dominant mechanism for the
generation of the longshore current distribution, the uncertainty of the
free parameter value limits the prediction ability of the 1D model.
Compared to the 1D model, the SWASH model is more accurate without
having to pay attention to tuning the free parameters. When the vertical
resolution is fine enough to resolve the vertical structure of mean flow,
the order of O(20) in this study, and the computational domain is
properly long with minor lateral boundary effects, the SWASH model
Fig. 14. Comparison between the simulated results of the R97_SA243 experi- reproduces the longshore current profiles in laboratory experiments
ment by the 1D model (blue line) and the results obtained by the SWASH model
fairly well, while only proper magnitude of longshore current velocity
(black dash line), (red circles: measurements). (For interpretation of the re-
could be predicted by the 1D model. Therefore, the SWASH model with
ferences to colour in this figure legend, the reader is referred to the Web version
of this article.)
default parameter settings is suggested to predict the distribution of
longshore currents accurately.

Thornton, 1993). For complex forces (e.g. wind and tidal effects) in the
field environment was not considered in the laboratory experiments 5. Conclusions
and the numerical simulations by the SWASH model except for pure
wave forces, the SWASH model could only reproduce the laboratory The SWASH model has been set up to simulate wave induced cur-
experiment results well, but could not contribute to illuminating the rent against six laboratory experiments, four of which are confined to
possible reason why the observed positions of maximal longshore cur- obliquely incident regular waves on non-barred beaches, and two ex-
rents appear in the trough. periments are on a barred beach under regular waves and irregular
It is shown that a simple 1D model is able to compute a longshore waves respectively. Nearly perfect agreements are found between the
current distribution comparable to the SWASH model. However, the computed and measured wave heights, mean water levels and long-
bottom roughness coefficient dominates the magnitude of computed shore currents. The longshore current generally spreads within 2 times
longshore current velocity, of which the realistic value is not optimum the surf zone width (Longuet-Higgins,1970b). For simpler well-con-
trolled experiments with regular wave incidence to alongshore uniform

104
R. Zhang et al. Coastal Engineering 142 (2018) 95–105

beaches considered in this paper, the longshore current is shown to Hsu, C.-E., Hsiao, S.-C., Hsu, J.-T., 2017. Parametric analyses of wave-induced nearshore
develop mostly within 1.5 times the width of the surf zone. Along current system. J. Coast Res. 33, 795–801. https://doi.org/10.2112/JCOASTRES-D-
16-00027.1.
vertical direction, the longshore current is quite depth uniform on plane Hulsbergen, C.H., ter Horst, W., 1973. Effect of Permeable Pile Screens on Coastal
beaches. The exception, existing in very shallow water, is that the Currents, vol. M1148 Delft Hydraulics laboratory report, Delft (in Dutch).
vertical profile of longshore current deviates from a depth uniform Launder, B.E., Spalding, D.B., 1974. The numerical computation of turbulent flows.
Comput. Methods Appl. Mech. Eng. 3, 269–289. https://doi.org/10.1016/0045-
profile. The stronger bottom friction in shallower water enforces the 7825(74)90029-2.
vertical profile to transform into a logarithmic profile. Ignoring the vary Liu, P.L.-R., Dalrymple, R.A., 1978. Bottom frictional stresses and longshore currents due
shallow zone nearshore, approximately stretching over 0.4 times surf to waves with large angles of incidence. J. Mar. Res. 36, 357–375.
Longuet-Higgins, M.S., 1970a. Currents generated by obliquely incident sea waves: 1. J.
zone width offshore, longshore currents maybe assumed depth uniform Geophys. Res. 75, 6778–6789.
under regular waves on plane beaches. Both the numerical simulations Longuet-Higgins, M.S., 1970b. Longshore current generated by obliquely incident sea
and laboratory measurements show rather depth uniform vertical waves: 2. J. Geophys. Res. 75, 6790–6801.
Longuet-Higgins, M.S., Stewart, R.W., 1964. Radiation stresses in water waves; a physical
structure of mean longshore current under regular waves on plane
discussion, with applications. Deep Sea Res. 11, 529–562.
beaches. This explains partly why the simpler 1D depth-averaged mo- Osiecki, D., Dally, W., 1996. The influence of rollers on longshore currents. In: 25th
mentum balance equation can be sufficient to compute the longshore International Conference on Coastal Engineering. Orlando, pp. 3419–3430.
current. However, the assumption of depth uniform longshore current Power, H.E., Baldock, T.E., Callaghan, D.P., Nielsen, P., 2013. Surf zone states and energy
dissipation regimes — a similarity model. Coast Eng. J. 55, 1350003. https://doi.org/
does not apply to a bared beach. 10.1142/S0578563413500034.
In conclusion, the phase-resolving SWASH model is capable to Putrevu, U., Svendsen, I. a., 1992. A mixing mechanism in the nearshore region. In: 23rd
compute wave-induced longshore currents. The subgrid approach ac- International Conference on Coastal Engineering. Venice, pp. 2758–2771. https://
doi.org/10.1061/9780872629332.210.
celerates the computation without sacrificing the accuracy. The less Reniers, a. J.H.M., Battjes, J. a., 1997. A laboratory study of longshore currents over
computational effort shows that this phase-resolved model is a pro- barred and non-barred beaches. Coast. Eng. 30, 1–22. https://doi.org/10.1016/
mising alternative to a phase-averaged model. Without tuning effort on S0378-3839(97)00017-3.
Reniers, a. J.H.M., Thornton, E.B., Stanton, T.P., Roelvink, J. a., 2004. Vertical flow
free parameters, the SWASH model predicts wave generated longshore structure during Sandy Duck: observations and modeling. Coast. Eng. 51, 237–260.
current fairly well on planar and barred beaches. More complex con- https://doi.org/10.1016/j.coastaleng.2004.02.001.
ditions such as combined wave and steady current condition, or rea- Reniers, A., 1999. Longshore Current Dynamics. Delft University of Technology.
Rijnsdorp, D.P., Smit, P.B., Zijlema, M., 2014. Non-hydrostatic modelling of infragravity
listic random waves in combination with tidal currents in fields would waves under laboratory conditions. Coast. Eng. 85, 30–42. https://doi.org/10.1016/
be investigated in further work. J.COASTALENG.2013.11.011.
Rijnsdorp, D.P., Smit, P.B., Zijlema, M., Reniers, A.J.H.M., 2017. Efficient non-hydrostatic
modelling of 3D wave-induced currents using a subgrid approach. Ocean Model. 116,
Acknowledgements
118–133. https://doi.org/10.1016/j.ocemod.2017.06.012.
Smagorinsky, J., 1963. General circulation experiments with the primitive equations.
The first author is sponsored by the Chinese Scholarship Council Mon. Weather Rev. 91, 99–164. https://doi.org/10.1175/1520-0493(1963)
(grant number: 201406060020). We gratefully thank Dr. Hulsbergen 091<0099:GCEWTP>2.3.CO;2.
Smit, P., Zijlema, M., Stelling, G., 2013. Depth-induced wave breaking in a non-hydro-
and Dr. Visser for their explanations about their experiments and Dr. static, near-shore wave model. Coast. Eng. 76, 1–16. https://doi.org/10.1016/J.
Rijnsdorp for sharing his professional expertise in SWASH modelling. COASTALENG.2013.01.008.
Finally, the suggestions and advice from the reviewers, Professor Smith, J.M., Larson, M., Kraus, N.C., 1993. Longshore current on a barred beach: field
measurements and calculation. J. Geophys. Res. Ocean. 98, 22717–22731. https://
Baldock and two anonymous reviewers, are very appreciated which doi.org/10.1029/93JC02116.
helped to improve this paper. Stive, M.J.F., 1984. Energy dissipation in waves breaking on gentle slopes. Coast. Eng. 8,
99–127.
Stive, M.J.F., De Vriend, H.J., 1994. Shear stresses and mean flow in shoaling and
Appendix A. Supplementary data breaking waves. In: 24th International Conference on Coastal Engineering. Kobe, pp.
594–608.
Supplementary data to this article can be found online at https:// Suzuki, T., Altomare, C., Veale, W., Verwaest, T., Trouw, K., Troch, P., Zijlema, M., 2017.
Efficient and robust wave overtopping estimation for impermeable coastal structures
doi.org/10.1016/j.coastaleng.2018.10.005.
in shallow foreshores using SWASH. Coast. Eng. 122, 108–123. https://doi.org/10.
1016/j.coastaleng.2017.01.009.
References Svendsen, I. a., Lorenz, R.S., 1989. Velocities in combined undertow and longshore cur-
rents. Coast. Eng. 13, 55–79. https://doi.org/10.1016/0378-3839(89)90032-X.
Svendsen, I.A., 1984. Wave heights and set-up in a surf zone. Coast. Eng. 8, 303–329.
Battjes, J.A., 1975. Modelling of turbulence in the surfzone. In: Proc. Symp. Model. Svendsen, I.A., Putrevu, U., 1994. Nearshore mixing and dispersion. Proc. Math. Phys. Sci.
Techniques. San Francisco, CA, pp. 1050–1061. 445, 561–576. https://doi.org/10.1080/037454809495909.
Bowen, A., 1969. The generation of longshore currents on a plane beach. J. Mar. Res. 27, Thornton, E.B., Guza, R.T., 1986. Surf zone longshore currents and random waves: field
206–215. data and models. J. Phys. Oceanogr. 16, 1165–1178. https://doi.org/10.1175/1520-
Chen, Q., 2003. Boussinesq modeling of longshore currents. J. Geophys. Res. 108, 3362. 0485(1986)016<1165:SZLCAR>2.0.CO;2.
https://doi.org/10.1029/2002JC001308. Thornton, E.B., Guza, R.T., 1983. Transformation of wave height distribution. J. Geophys.
Church, J.C., Thornton, E.B., 1993. Effects of breaking wave-induced turbulence within a Res. 88, 5925. https://doi.org/10.1029/JC088iC10p05925.
longshore-current model. Coast. Eng. 20, 1–28. Visser, P.J., 1991. Laboratory measurements of uniform longshore currents. Coast. Eng.
Dally, W.R., Brown, C.A., 1995. A modeling investigation of the breaking wave roller with 15, 563–593. https://doi.org/10.1016/0378-3839(91)90028-F.
application to cross-shore currents. J. Geophys. Res. 100, 24873–24883. https://doi. Visser, P.J., 1984a. Uniform longshore current measurement and calculations. In: 19th
org/10.1029/95JC02868. International Conference on Coastal Engineering. Houston, pp. 2192–2207.
de Bakker, A.T.M., Tissier, M.F.S., Ruessink, B.G., 2016. Beach steepness effects on Visser, P.J., 1984b. A mathematical model of uniform longshore currents and the com-
nonlinear infragravity-wave interactions: a numerical study. J. Geophys. Res. Ocean. parison with laboratory data. In: Comm. on Hydraulics, Rep. 84-2, (Dep. Civil Eng.,
121, 554–570. https://doi.org/10.1002/2015JC011268. Delft Univ. of Technology, Delft, The Netherlands).
Duncan, J.H., 1981. An experimental investigation of breaking waves produced by a Visser, P.J., 1982. The Proper Longshore Current in a Wave Basin. pp. 82–91 Delft, the
towed hydrofoil. Proc. R. Soc. A Math. Phys. Eng. Sci. 377, 331–348. https://doi.org/ Netherlands. Rep.
10.1098/rspa.1981.0127. Visser, P.J., 1980. Longshore current flows in wave basin. In: 17th International
Feddersen, F., Guza, R.T., Elgar, S., Department, T.H.C.H., 2000. Velocity moments in Conference on Coastal Engineering. Sydney, pp. 462–479.
alongshore bottom stress parameterizations. J. Geophys. Res. 105, 8673–8686. Wright, D.G., Thompson, K.R., 1983. Time-averaged forms of the nonlinear stress law. J.
Galvin, C.J., Eagleson, J.P.S., 1964. Experimental Study of Longshore Currents on a Plane Phys. Oceanogr. 13, 341–345. https://doi.org/10.1175/1520-0485(1983)
Beach. Hydrodynamics Laboratory Report, vol. 63 Massachusetts Institute of 013<0341:TAFOTN>2.0.CO;2.
Technology, Cambridge. Zijlema, M., Stelling, G., Smit, P., 2011. SWASH: an operational public domain code for
Hamilton, D.G., Ebersole, B.A., 2001. Establishing uniform longshore currents in a large- simulating wave fields and rapidly varied flows in coastal waters. Coast. Eng. 58,
scale sediment transport facility. Coast. Eng. 42, 199–218. https://doi.org/10.1016/ 992–1012. https://doi.org/10.1016/J.COASTALENG.2011.05.015.
S0378-3839(00)00059-4.

105