Hydrodynamics of a headland-bay beachNearshore current circulation

Rodolfo Silva
a,
, Asuncin Baquerizo
b
, Miguel ngel Losada
b
, Edgar Mendoza
a
a
Instituto de IngenieraUNAM, Ciudad Universitaria, DF, Mexico
b
Centro Andaluz de Medio AmbienteUniversidad de Granada, Spain
a b s t r a c t a r t i c l e i n f o
Available online 4 November 2009
Keywords:
Headland-bay beaches
Nearshore circulation
Shoal inuence
Wave currents
Analytic solution
Mild-slope equation
Radiation stress tensor
This paper is devoted to the analysis of the hydrodynamic equilibrium of a headland or semi-elliptic shaped
beach. It is shown that the state of equilibrium depends not only on the in- and outgoing sediment but also
on the accommodation of the sediment within the embayment. The shape and relative depth of shoals, or
settling zones, also directly affect the wave and current patterns inside the bay, within which the resultant
breaker line almost stops wave-induced currents at some locations, whereas the magnitude of current
increases at other locations. Several numerical tests are analytically conducted in a semi-elliptic beach with
two symmetrical shoals of varying relative depth where circulatory current systems are detected and
analyzed. Numerical modelling for wave climate and wave-induced current estimation is also presented in
order to corroborate results and provide a tool for complicated and/or physical domains. The results lead to a
redenition of the concept of equilibrium for headland-bay beaches taking into account not only the net
sediment transport but also the role of the formation and disappearance of settling zones, as well as
sediment interchanges between the beach and shoals.
2009 Elsevier B.V. All rights reserved.
1. Introduction
According to Hsu et al. (2008) the stability of a headland-bay, or
crenulated, beach, is determined by the balance between the
incoming and outgoing sediment in the beach. It is this balance
which allows beach classication, given as \, static equilibrium, and
unstable due to a reduction in sediment supply and reshaping
beaches caused by the installation of a coastal structure or
modication to an existing wave diffraction point.
In crenulate beaches the sediment is partially or totally blocked at
the headlands and they usually have therefore very little sand supply
from neighbouring updrift areas. They show a characteristic curved
coastline which is the result of a continuous adjustment to the
incoming wave fronts. This gradual adaptation is mainly driven by the
longshore sediment transport that takes place at locations where
the waves break obliquely to the bathymetric contours. Under the
prevailing wave climate, the circulatory system in a crenulate or
headland-bay beach is relatively weak and net sediment transport is
not signicant. Therefore, the shoreline planform in such a beach
closely reects equilibrium conditions. However, the beach does not
remain static, but responds to climatic changes, storms and calms;
gradually adapting its planform to the prevailing wave conditions.
Understanding the morphodynamics of crenulate beaches is not
possible without rst examining the hydrodynamic phenomena that
take place within the bay. The complexity and wide range of temporal
and spatial scales of the processes involved make such a description
very challenging.
Over a short period covering a few sea states, during a storm for
example, the physical activity in a headland-bay is mainly dominated by
wind waves and their transformation by bathymetric changes through
shoaling, refraction, reection and, at the headland, mainly diffraction. As
the water depth becomes shallower, the wave prole becomes steeper
and breaks when a threshold, approximately H
L
/h, is reached (H
L
being
local wave height andhthe still water level). This positionis the beginning
of the surf zone, where the height of the breaking wave falls, dissipating a
great deal of energy, mainly through turbulence. As waves propagate into
shallow water, the gradients of the excess momentum ux of the waves
causearisingandloweringof themeanwater level (set-upandset-down)
and may induce a complex circulatory system, particularly intense inside
the surf zone, which broadly consists of undertow, longshore and rip
currents. These wave-induced currents interact with the incoming waves
modifying the wave propagation pattern. Finally, at the shoreline, the
wave begins a landward run-up process in the form of a thin sheet of
water which returns seaward by gravity.
In this general picture of beach dynamics, the alongshore drift
(generated in the direction of the longshore component of the incident
wave when wave breaking takes place oblique to the bathymetric
contours) is the dominant motion averaged over the wind wave period.
Although the current magnitude is often weak, it is however capable of
transporting the sediment mobilised by wave action. This is the principal
cause of morphological changes at the beach over short periods.
It is also easy to observe that inside the bay, outside the surf zone,
shoals or zones of sediment accumulation develop, usually under
moderate wave action and these are swept away under storm
Coastal Engineering 57 (2010) 160175
Corresponding author.
E-mail address: RSilvaC@iingen.unam.mx (R. Silva).
0378-3839/$ see front matter 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.coastaleng.2009.10.003
Contents lists available at ScienceDirect
Coastal Engineering
j our nal homepage: www. el sevi er. com/ l ocat e/ coast al eng
conditions. The mechanisms involved in the generation and extinction
of these morphological features are still not well understood. Generally
speaking, with relatively lowwaves the sediment is brought shoreward
in the propagation region, building a monotonic beach prole. The
circulation induced within an embayment by wave breaking redis-
tributes part of this sediment along the shore and also forms shoals at
the end of the outgoing rips. The changing bathymetry modies the
wave propagation pattern and, therefore, the wave height and the angle
at which the waves break, with subsequent changes in current eld
direction and intensity.
The concept of dynamic equilibrium in headland-bay beaches
should not therefore isolate shoreline behaviour fromshoal formation
and the possible interrelations between these features should be
examined. In this paper, the role played by shoals in the general
circulation of a beach is analyzed. It is shown that the development of
the shoals decreases the intensity of the velocity eld and eventually
stops it. Moreover, if the water depth over the shoals is shallow
enough, wave breaking inside the bay can even invert this process,
sweeping the shoals away.
2. Governing equations for mean ows
It is well knownthat the oscillatory motionin wind wave time scales
can induce free surface motions and current elds of a longer time scale
than the corresponding wave period. The equations governing these
mean ows are the shallow water equations, obtained from time and
depth averaged NavierStokes equations. In a Cartesian horizontal
coordinate systemwiththe (x, y) axes lying over the still water level and
the z-axis pointing upwards, they can be written as

t
+
uh
x
+
vh
y
= 0 1
uh
t
+
u
2
h
x
+
uvh
y

hu
x

x
+
hu
y

y
_ _
=

wx

bx

gh

x
+ hfv
1

S
xx
x
+
S
yy
y
_ _
2
vh
t
+
uvh
x
+
v
2
h
y

hv
x

x
+
hv
y

y
_ _
=

wy

by

gh

y
+ hfu
1

S
yx
x
+
S
yy
y
_ _
3
where is the free surface elevation, u and v are the velocity vector
components, respectively, h is the total depth (h=h
s
+) with h
s
being
the still water level. is the eddy viscosity,
wi
the surface stresses,
bi
is
the bed friction stresses (i =x, y) and f is the Coriolis parameter. S
ij
are
the components of the radiation stress tensor that represent the excess
momentum uxes associated with the oscillatory wave motion.
The gradients of the radiation stress are the driving forces of the
mean ows. These forcing terms depend on the wave propagation
pattern and in a rst approach they are usually specied and provided
as an input, therefore neglecting the interaction between the mean
and the instantaneous quantities. Even with this simplication, the
resolution of the whole set of equations can be very complicated,
especially over real domains with complex boundaries and bathy-
metric contours.
In the next section an overviewis given of the analytical approaches
to estimate circulation induced by wave breaking with emphasis on
semi-elliptic beaches. Next, a numerical model that comprises both a
wave propagation model (Silva et al. 2005) and the estimation of the
mean ows valid for arbitrary bathymetric contours over complex
domains is summarized.
3. Analytical solutions
Over half a century ago Iribarren (1947a) pointed out the im-
portance of the alongshore variation of the set-up induced by waves
to drive currents in a permanent regime and derived a formula based
on that of Chezy to estimate the magnitude of the current. He applied
these ndings to explain the growth of a spit in the estuary of
Fuenterraba (Iribarren, 1947b) and the development of a crenulate
beach bounded by a jetty at Cape Higuer. Elsewhere, Shepard and
Inman (1950) described the inuence of rip currents on incident
waves, nearshore circulation of water, magnitude and direction of
sediment transport and consequently the shape of shorelines. Later
McKenzie (1958) underlined the relationship between rip current
ows and wave-forcing parameters.
Since the introduction of the radiation stresses by Longuet-Higgins
and Stewart (1964), several attempts have been made to estimate the
circulation patterns induced by wave breaking. In this context, the
analytical resolution of the simplied equations over certain regular
domains, although less accurate than numerical models, has given
insight into the behaviour of mean ows under different forcing
conditions depending on the geometry of the considered domain.
With this aim, the solutions provided by Bowen (1969a) and
Longuet-Higgins (1970) described the mean water level and the
longshore current over an innitely long beach with straight and
parallel contours with the radiation stresses calculated for a mono-
chromatic wave train approaching the shore at an oblique angle.
Battjes (1974) addressed this case but with irregular waves instead.
Their solutions showed that the obliquity of wave fronts at breaking is
responsible for an alongshore component of the momentum ux into
the surf zone.
Also on a straight beach Bowen (1969b) and Dalrymple (1975)
showed that alongshore variations of the breaking wave height will
cause a variation in wave set-up along the shore. Under such conditions
feeder currents will ow away from zones of high waves and toward
zones of low waves where they converge and move seaward as rip
currents.
A more complex geometry was analyzed by O'Rourke and Le Blond
(1972) who studied the circulation patterns in a semi-elliptic bay and
found that the rip currents are mainly driven by the oblique angle of
incidence. Also, Baquerizo et al. (2002) and Baquerizo and Losada
(2002) presented a quasi-analytical approach for the circulation
induced by wave breaking in a semi-elliptic beach, and showed that
for this type of beach, the circulation inside the surf zone is mainly due
to the angle at breaking, while the effect of the gradient of the wave
height strengthens or weakens the velocity values without modifying
the circulation pattern.
In this paper the analytical model of Baquerizo and Losada (2002)
is applied to study the nearshore current eld for headland-bays with
a shoreline form resembling a semi-elliptic beach as formed within
the shelter of two headlands. Appendix A summarises the model
assumptions and the resolution procedure.
4. Numerical approach
For an arbitrary beach form with irregular bathymetric contours,
the numerical resolution of equations requires the coupling of two
numerical models. For the estimation of the wave climate a modied
mild-slope equation, in its full form, is used given that reection and
Table 1
Beach conguration and climatic parameters.
Beach conguration x
s
(m) y
s
(m) p m
0
s
b
(rad)
A 1200 1100 0.25 0.25 1

3
cos v
B 1200 1100 0.25 0.00 1

3
cos v
C 1200 900 0.25 0.25 1

3
cos v
161 R. Silva et al. / Coastal Engineering 57 (2010) 160175
diffraction are the governing processes (Silva et al. 2005). Next, the
shallow water equations with the radiation stress tensor included for
nearshore circulation, are solved with an adaptation of the numerical
method presented by Rogers (2001).
4.1. Wave climate estimation
The mild-slope and Boussinesq-type equation models represent
rather sophisticated techniques to describe the nearshore wave climate.
Boussinesqequations offer the most complete descriptionof the physics
Fig. 1. Angle of incidence at breaking.
Fig. 2. Stream functions and velocity elds for Cases A, B and C.
162 R. Silva et al. / Coastal Engineering 57 (2010) 160175
for fairly long waves. However, they are restricted to slowly varying
bottoms and their use over large areas for engineering applications is
not recommended because of their complexity. Under such circum-
stances, it is preferable to use mild-slope models that have shown
robustness when applied to areas where diffraction and reection are
dominant, as happens in headland-bay beaches.
Since the basic mild-slope equation was published by Berkhoff
(1973), several versions have been derived. Disregarding the effect of
the evanescent modes and assuming an impermeable bottom, in
terms of engineering application, the solution presented by Cham-
berlain and Porter (1995) and implemented numerically by Silva et al.
(2005) would seem to be one of the most robust solutions. It retains
second order terms and does not have any limitation on bottom
variations. In Appendix B, a detailed description of this version is
given.
4.2. Wave-driven current estimation
The NavierStokes equations provide a complete description of the
uid motion, but are inherently nonlinear and impossible to solve
analytically except for simplied cases. Reynolds (1895) time-
averaged the NavierStokes equations to produce apparently simpler
equations that still require additional assumptions such as closure sub
models. Due to the complexity of the physical processes resulting in
nearshore wave-induced currents, no numerical solutions to predict
fully coupled wavecurrent interaction are known to the authors, as
explained by Rogers (2001).
Lamb (1879) was the rst to consider analytical solutions of the
linearised two-dimensional depth-integrated shallow water equations
(SWE), describingowinwater bodies where horizontal characteristics,
such as horizontal circulation, rip currents and longshore currents,
Fig. 3. Photograph of Niembro beach a) at high tide and b) at low tide.
163 R. Silva et al. / Coastal Engineering 57 (2010) 160175
prevail over the depth inuence. Nowadays however, computer
solutions of the complete nonlinear SWE are a relatively common
place (e.g. Ebersole and Darlymple, 1979), but despite the substantial
work done on the mathematical description of ows in shallow
water and coastal hydrodynamics, the picture is still incomplete with
important physical processes such as wave breaking, turbulence and
bed frictiononly approximately represented. The procedure used in this
work to solve Eqs. (1)(3) is given in Appendix C.
5. Numerical results
In this section the analytical results are presented giving an
impression of the general circulation under different conditions, so
that the role of the parameters involved may be studied. There follows
a eld case in which the numerically computed hydrodynamics in
Acapulco Bay, Mexico are described.
5.1. Analytical model
The circulation pattern induced by wave breaking is analyzed rst
for three different congurations of the geometry and the sea climate
conditions (see Table 1). The shoreline planform is specied by the
values of the x
s
, y
s
and the relative surf zone width,
=
u
s
u
b
u
s
4
where u
s
and u
b
are the eccentricity parameters of the shoreline and
the breaker line, respectively.
All the cases correspond to a semi-elliptic bay without shoals and
with a plane sloping surf zone, with depth given by
ds; v = m
0
1p cos2vs; s0; 1; v0; : 5
In Eq. (5), v is the coordinate that sweeps the semi-ellipse from the
negative part of the major semiaxis to the positive part in an
anticlockwise direction, and s indicates the relative distance to the
shoreline (see Appendix A). The forcing sea climate is dened by the
variation of the water depth inside the surf zone d=d(s,v), together with
the wave angle at the breaker line,
b
=
b
(v). The maximum values of
the wave height at breaking are of the order of magnitude of 1 m.
Cases A and B correspond to an almost circular beach with
semiaxes x
s
=1200 m, y
s
=1100 m and relative surf zone width
=0.25, while Case C is a atter semi-ellipse with the same major
semiaxis and surf zone width and minor semiaxis, y
s
=900 m. The
angle at breaking is the same for all cases (see Fig. 1, solid line).
Fig. 2 shows the isolines of the streamfunctions and the velocity elds
obtained for the three cases, where two main circulation cells with water
obliquely entering the surf zone withanoutgoing ripcurrent at the axis of
the embayment can be identied. Maximum velocities of about 0.05 m/s
areobtained. Despitetheabsenceof lateral friction, non-zerovelocities are
found in the propagation region because of continuity requirements. Case
Ashows stronger velocities inthe outgoing ripcurrent thanCaseB. For the
smaller minor semiaxis, the cells are atter, the central rip weakens and
the surf zone lateral longitudinal currents are stronger. InCaseB, there are
no gradients of the wave height along the wave breaker line and the
circulation is therefore due only to obliquity at breaking. The velocity eld
has a pattern similar to that of Case A, but the currents are weaker. These
results arebythemselves explanatoryof therelativeimportanceof thebay
Fig. 4. Stream functions and velocity elds for the at bay and the shoaled bay.
164 R. Silva et al. / Coastal Engineering 57 (2010) 160175
geometry at the nearshore and the wave height and angle at breaking, in
the circulation.
At rst glance, the symmetric circulation cells would potentially
transport the sediment from the surf zone towards the bay, where
eventually, two shoals could develop. This reasoning, although lacking
scientic precision, is supported by eld observations as it can be seen
in Fig. 3 at Niembro beach where shoals show up at low tide.
Fig. 5. Bathymetric contours of the at bay and the shoals.
Fig. 6. Wave propagation patterns for the at bay and the shoaled bay with b=1 and
b=2 m. Fig. 7. Velocity elds for the at bay and the shoaled bay with b=1 and b=2 m.
165 R. Silva et al. / Coastal Engineering 57 (2010) 160175
The presence of the shoals will gradually modify the wave propa-
gation patterns and, consequently, the forcing mechanisms of the
current velocity eld and the bed contours. As a rst approach, the
circulation in a at bay with dimensions similar to that of Niembro
beach, is obtained and compared with the case of a bay with two
symmetric shoals. The semiaxes of the bay are x
s
=120 m and
y
s
=117 m with =0.25. For the last case, the wave fronts are
delayed over the shallower part of the shoals giving a variation with v
of the wave angle at breaking similar to the one shown in Fig. 1
(dashed line). Fig. 4 shows the velocity eld obtained for both the at
and the shoaled bays where a variation in magnitude of the vectors
can be appreciated.
5.2. Numerical model results
The results above lead to a further analysis, using a more complete
solution of the SWE, with the effect of shoals in the circulation in-
duced in the surf zone. Thus a semi-elliptic beach without shoals, with
semiaxes x
s
=1200 m and y
s
=900 m is considered as a reference
case and shoals with the same shape and different heights measured
by the maximum value b (b=0.5, 1, 1.5 and 2.0 m) are added to the
at bathymetry (Fig. 5). The numerical coupled solution presented
in Section 2 is applied to all the cases. Figs. 6 and 7 show the wave
propagation patterns and the circulation induced for some of the cases
of the bays with and for the one without shoals. Moreover, in order to
have a more accurate impression of the modication of the velocity
currents induced by shoals, Fig. 8 gives the maximum absolute values
of the currents in the surf zone versus the parameter v that runs along
the length of the beach.
For the bay without shoals wave breaking takes place along an
almost elliptic curve and drives a weak nearshore circulation with a
similar pattern to that obtained with the analytical model. Relatively
low shoals (b=0.5 or 1 m) gradually delay the advance of the wave
fronts at the centre of the bay. The change induced in the angle at
wave breaking increases the magnitude of the symmetric alongshore
current at the lateral zones while reducing it at the central part of the
shore. For the shoal of height b=1.5 mthe wave front breaks all along
the bay almost parallel to the depth contours, with a reduction of
about 50% of the intensity of the velocity in relation to the at shoal.
Finally, for the highest shoal (b=2 m), wave breaking takes place
over the shoals widening the surf zone towards the centre of the bay
and increasing the velocity elds.
5.3. Numerical modelling
A headland-bay beach was selected to numerically analyze near-
shore circulation under different wave climate conditions, Acapulco
Bay in Mexico. For many years the bay of Acapulco has been an
important tourist destination, noted for its calm waters and beautiful
beaches. One of its main characteristics is the low energy of the
waves near the coast and the stability of the beaches. Fig. 9 shows
the bathymetry of the bay which has been discretised into a grid
with x=y=10 m, the wave period in all test cases is 10 s and
H=1 m.
In order to achieve a complete panorama of the hydrodynamic
conditions throughout an average year, three scenarios were selected
for the simulation:
Case 1: waves coming normally from the south in summer (approx
3 months/year)
Case 2: waves coming from SW in winter (approx 3 months/year).
Case 3: waves coming from SW-S during transition (approx
6 months/year).
Fig. 8. Maximum current velocities along the surf zone.
Fig. 9. Bathymetry of Acapulco Bay (depths in meters).
166 R. Silva et al. / Coastal Engineering 57 (2010) 160175
The wave height maps and the free surface contours for Case 1 are
presented in Figs. 10 and 11. The interaction between the bay mor-
phology and the wave incidence angle produces an energy con-
centration in Playa Condesa leaving the rest of the bay almost without
waves. In the northern part of the domain some reected patterns are
clearly visible. The free surface contours clearly exhibit the energy
dissipation in the north and south of the bay as well as in the channel
between Playa Caleta and Isla Roqueta. Fig. 12 shows the computed
wave-driven currents, which are only signicant in front of the cliffs
and, to a lesser extent, very close to Playa Condesa inside the bay. A
detailed image of this is presented in Fig. 13, showing the position of
two islets.
When the incident waves come from the SW (Case 2), the
dynamics inside the bay change, as shown in Fig. 14, where the energy
concentration has moved to Playa Icacos and almost no waves arrive
at the northern part of the bay. Meanwhile, in the northern part of the
domain (outside the bay) an increase in the wave height is produced.
The free surface contours of waves travelling into the bay seem to
Fig. 10. Wave height pattern for Case 1.
Fig. 11. Instantaneous free water surface for Case 1.
167 R. Silva et al. / Coastal Engineering 57 (2010) 160175
Fig. 12. Computed wave-driven currents for Case 1.
Fig. 13. Bay interior detail (Case 1).
Fig. 14. Wave height pattern for Case 2.
168 R. Silva et al. / Coastal Engineering 57 (2010) 160175
enter undisturbed until they reach the beach (Fig. 15) explaining
the energy concentration. Important currents are generated only in
front of the cliffs and in Playa Icacos, and this can be seen in Fig. 16
and with more detail in Fig. 17. The magnitude of the wave-driven
currents makes this the only condition in which the beach inside the
bay is unstable.
For Case 3 (waves coming fromsouth-southwest) the morphology
and bathymetry of the bay dissipate almost all the wave energy as
seen in Fig. 18. The free surface contours presented in Fig. 19 shows
the same tendency, with very small waves within the headland-bay
beaches. In this case no signicant wave-driven currents are
generated. The numerical results are shown in Figs. 20 and 21. The
angle of incidence corresponding to Case 3 is that which occurs most
often during an average year, in spring and autumn, hence the long
lasting beach stability in Acapulco Bay.
6. Conclusions
The analytical and numerical models described here are used to
compute wave climate and wave-driven currents. They are both
Fig. 15. Instantaneous free water surface for Case 2.
Fig. 16. Computed wave-driven currents for Case 2.
169 R. Silva et al. / Coastal Engineering 57 (2010) 160175
Fig. 17. Bay interior detail (Case 2).
Fig. 18. Wave height pattern for Case 3.
Fig. 19. Instantaneous free water surface for Case 3.
170 R. Silva et al. / Coastal Engineering 57 (2010) 160175
useful and robust tools for practical applications giving results that
agree with eld observations where the absence of numerically
generated currents in a stable beach may be understood as having
insignicant net sediment transport.
In particular, the modied mild-slope equation has proved to
successfully represent wave patterns over complicated and large
bathymetries, as well as for various boundary conditions. In addition,
the numerical scheme used to solve the SWE demands very little com-
putational effort and easily deals with discontinuities inside the domain.
These tools were applied to analyze the nearshore circulation
induced by a wave breaking in a headland-bay beach of a semi-elliptic
shoreline form with emphasis in the role played by the shoals that
usually develop under mild conditions. It is observed that the shoals
modied the wave breaking pattern, inducing current velocities at
the surf zone to gradually reduce, up to 50%, approximately, provided
that wave breaking does not take place over the shoals. If the water
depth over the shoals is shallow enough to provoke wave breaking,
the surf zone widens towards the centre of the bay and increases the
current velocities.
This picture suggests that a headland-bay beach not in equilibrium
is able to respond to variations in the energetic level of the wave
conditions with changes of the submerged beach prole without
suffering signicant changes in form.
The concept of equilibrium should be understood as dynamic
equilibrium which means: changing sand from the beach face to the
shoals and vice versa, as the wave energy increases and then
decreases. The development of the shoals is geometrically similar to
the changes of a beach prole in storms followed by mild conditions.
However, hydrodynamically, headland-bay beaches without and
withshoals are different because, once the shoals grow, the maineffect
is to weaken the circulation pattern in the surf zone limiting sediment
transport and beach erosion. The retreat of the beach is usually small,
accounted for by the amount of sand stored in the shoals.
In the case of a beach prole the growth of sandbars causes the
waves to break offshore, stopping their propagation up to the beach
face, weakening the undertow currents and the sediment transport
to the beach. The usually large retreat of the beach here is accounted
for by the amount of sand which is almost completely stored in the
sandbars.
Acknowledgements
Asuncin Baquerizo and Miguel ngel Losada are indebted to the
Comisin Interministerial de Ciencia y Tecnologa (CICYT), contract
nos. HI2003-0266 and CTM2005-06583/MAR. Rodolfo Silva and Edgar
Mendoza wish to thank the Mexican Consejo Nacional de Ciencia
y Tecnologa, CONACYT, project Caracterizacin morfodinmica y
evaluacin de alternativas para la recuperacin de playas en la zona
norte de Quintana Roo under contract QROO-2003-C02-12707 for
having partially funded this investigation.
Fig. 20. Computed wave-driven currents for Case 3.
Fig. 21. Bay interior detail (Case 3).
171 R. Silva et al. / Coastal Engineering 57 (2010) 160175
Appendix A. Analytical model for elliptic beaches
Baquerizo and Losada (2002) solved the linearised equations for a
steady wave eld without lateral momentum diffusion. They also
assumed that currents are small enough so that their interaction with
the waves is negligible. This assumption is disregarded in the seaward
zone in the regions with strong outgoing rip currents that may modify
the wave propagation pattern.
Governing equations
Under such assumptions, denoted by
T
x
=
1
h
S
xx
x
+
S
xy
y
_ _
; T
y
1
h
S
yx
x
+
S
yy
y
_ _
6
and dening the vectors T

=(T
x
,T
y
) and

b
=(
b,x
,
b,y
), Eq. (6) can be
expressed in vectorial form as

T =

b
g 7
and by applying the rotational operator to Eq. (7),

T =

b
: 8
The frictional terms are assumed to be proportional to the current
velocity eld,

b;x
=
fu
h
;
b;y
=
fv
h
9
where f is a friction coefcient that, for a sinusoidal wave, and small
current velocities compared to the maximum wave orbital velocities,
is known to depend on the total water depth and wave amplitude
(Longuet-Higgins, 1970) and can be expressed as
f =
2C
f
u
max

: 10
In Eq. (10) C
f
is a friction coefcient of the order of 0.01 (see
Longuet-Higgins, 1970) and u
max
is the maximum bottom horizontal
orbital velocity of the wave.
On the other hand, the fullment of the conservation equation
guarantees the existence of a stream function, , that allows the
velocity eld to be expressed as
u =
1
h

y
; v =
1
h

x
: 11
In terms of , the problem can be reduced to the resolution of the
following partial differential equation

f
d

1
d

_ _ _ _
=

T: 12
Inside the surf zone, the circulation is forced by T

and can be
calculated once the components of the radiation stress tensor are
specied. Outside the surf zone, there is no driving force. However,
current shows up because of continuity requirements. In these regions
the problem can be reduced to solve

b
= 0: 13
Notice that because the diffusive terms are neglected, a disconti-
nuity in the eld velocity should occur along the offshore border of the
surf zone.
Beach conguration
Eq. (12) is used to obtain the circulation induced by wave breaking
in the semi-elliptic shaped beach shown in Fig. 22. The geometric
parameters dening the bay shape are, the semiaxes, x
s
and y
s
or
equivalently, the eccentricity parameter
u
s
= atanh
y
s
x
s
14
and the scale factor,
a =
x
s
coshu
s
: 15
The breaker line is considered to be dened by a semi-ellipse of
semiaxes x
b
and y
b
with the same scale factor, a, as the shoreline and
eccentricity parameter, u
b
. While in the surf zone the total water depth
d is assumed to vary from the depth at the breaker line to 0 at the
shoreline, the seaward region bottom is supposed to smoothly depart
from a constant value, d
0
, that is, d=d
0
+f(x,y) where ||f(x,y)|| 1.
This assumption allows variations in water depth and reduces the
problem to a Laplace equation outside the surf zone where there is no
forcing terms.
Solution of the problem
Baquerizo and Losada (2002) solved the problem by splitting the
domain into ve regions (Fig. 23) with the conditions of no ux across
the boundaries and continuity of the stream function, , and its
Fig. 22. Beach conguration.
172 R. Silva et al. / Coastal Engineering 57 (2010) 160175
normal derivative,
r
/n, at the interfaces. Inside the bay, in the surf
zone (Region A), an elliptic-hyperbolic coordinate system
x = a cosh u cos v
y = a sinhu sinv
16
gives a solution under the assumption that the relative surf zone
width is a small parameter, dened as
=
u
s
u
b
u
s
: 17
In the seaward zone (Regions C, D and E) the problem reduces to a
Laplace equation and is solved by separation of variables with the
conditions of no ux across the impermeable boundaries (, x
s
)
and (x
s
, ) and no current velocities far offshore. Inside the bay,
Region B, the solution is obtained by applying Green's second theorem
to the stream function and Green's function of the free space for
several collocation points on the boundary of the domain.
Appendix B. WAPO model equation
The modied version of the mild-slope by Silva et al. (2005) was
developed by assuming a homogeneous incompressible uid with
irrotational motion over an impermeable bottom with spatially
varying water depth h(x, y), for the three-dimensional problem. The
Laplace equation is derived from the continuity equation. Applying
Green's second identity to the depth dependency function, f(z), and
the velocity potential, (x,y,z,t)=Re{(x,y,t)f(z)}, to the Laplace
equation, and substituting the linearised kinematic boundary condi-
tion at the impermeable bottom and the linearised mixed free surface
boundary condition for variable depth, the modied mild-slope
equation can be derived. Including a dissipation factor and following
Silva et al. (2005), without loss of generality, the modied time-
independent mild-slope equation for a monochromatic wave, in its
Helmholtz form can be written as

2
h
+ K
2
c
= 0; 18
where
K
2
c
= k
2
+ iD +
rh
I
1

I
1
_

I
1
_ ; 19
rh = I
2

2
h
h +
h
h
2
I
2
h
I
3
_ _
; 20
I
1
=
0
h
f z
2
dz; I
2
=
0
h
f z
f z
h
dz; I
3
=
0
h
f z
h
_ _
2
dz: 21
In Eqs. (19)(21), f(z)=cosh k(h+z) coth kh and
h
=/x,/
y
is
the horizontal gradient operator, and (x,y) represent the horizontal
coordinates. =(I
1
)
1/2
is the normalized velocity potential, z is the
vertical coordinate measured positively upwards from the undisturbed
free surface at z=0 and is the angular frequency (=2/T), in which
T is the wave period and the wave number k is determined from
2
=kg
tanhkh. Dis the dissipation factor. The WAPOmodel only considers wave
breaking andbottomfrictiondissipation, andsoD=f
D
+f
B
. The breaking
dissipation factor, f
D
, may be expressed as Dally et al. (1985):
f
D
=
kC
k
h
1
C
G
h
H
B
_ _
2
_ _
22
where H
B
is the wave height at the breaking point, (easily evaluated
through H
B
=h, with =0.8), C
k
=0.15, and C
G
=0.4.
In practice, the wave-induced bottom boundary layer is usually
turbulent. The dissipation termis given by Kirby and Dalrymple (1994),
f
B
=
4
3
C
f
a
2
ng sinh
3
kh
23
where C
f
is the DarcyWeisbach friction factor, a is the local wave
amplitude (H/2), is the angular frequency and
n =
1
2
1 +
2kh
sin2kh
_ _
: 24
The boundary conditions used here were derived previously by
Silva et al. (2005), thus we present only a summary of them. For
solving Eq. (18), two classes of boundary conditions were considered:
(a) an open boundary; and (b) a partially reecting boundary
condition. The open boundary allows full transmission of both
incoming and outgoing waves. At its extremes, the partially reecting
boundary condition tends either to a totally reecting or a fully
absorbing boundary condition.
For simplicity, the seaward boundary condition can be treated as
an open boundary of constant depth, the landward boundary as
partially reecting, and the lateral boundaries as open or partially
reecting. The exterior bathymetry may vary only in the cross-shore
direction.
For the case of an open boundary condition a second order
parabolic approximation of the equation is used. For the seaward
boundary condition, the total potential is considered to be the sum of
the incident and the scattered waves, where the scattered waves
travel towards the seaward boundary in directions that partially
oppose the incident waves. An equivalent procedure can be applied
for the y-direction, if it is assumed that the lateral boundaries have
constant water depth and if reections from the landward side can be
disregarded. In practice however, the depth usually decreases in the
landward direction. Moreover, the plane wave formulation does
not include the effects of breaking. When this procedure is used in
conjunction with the governing equation in the interior of the
domain, with wave breaking, a discontinuity along the open boundary
results from incorrect forcing. To overcome this limitation, it is
therefore assumed that the depths at the lateral boundaries change
solely in the x-direction (as also implemented by Zhao et al. 2001).
The landward boundary condition is obtained by using the as-
sumption that the wave eld may be decomposed into wave trains
that approach and are reected from the coast. The value of the phase
shift between the incident and the reected trains is difcult to
Fig. 23. Division of the domain into regions.
173 R. Silva et al. / Coastal Engineering 57 (2010) 160175
determine, hence it is normally set as =0. Assuming that it is valid
to use Snell's law, the local angle
1
may be estimated from
sin
l
= k
0
= k
l
sin 25
where k
0
is the wave number at the seaward boundary and k
l
is the
local wave number at the location of interest.
The lateral boundary conditions can be established through the
following equation,

y
= F iK
c
a
0
+ i
a
1
K
c

x
2
+
a
2
K
2
c

yx
2
_ _
+ 2ik
l

l
cos
l
; on
Fyboundary:
26
Since the breaking parameter and the bottom friction dissipation
term are functions of the local wave height and are unknown
beforehand, Eq. (26) has to be solved by an iteration procedure. In the
rst iteration no dissipation term is considered, and the resulting
velocity potential is used as the initial local incident velocity potential.
In the second iteration the dissipation terms are evaluated using the
local wave height, and so on.
For the case of a partially reected boundary condition, a linear
wave train travelling from the interior to the exterior of the domain is
considered. When this wave reaches the boundary it is partially
reected into the domain and partially transmitted outside the domain.
Eq. (18) and the boundary nite difference equations are applied to
all the water grid points in the domain. The resulting banded matrix
equationis solvedusing aninexpensive bandedsolver, ina way similar
to that of Silva et al. (2005) and Maa et al. (2002). The procedure can
be summarized as follows: to conserve computer memory, the spare
matrix is stored in two matrices; one contains the complex coefcients
and the other contains the locations of each unknown; the solver
constructs a small-banded matrix equation and then follows the
standard Gaussian elimination method with partial pivoting for
forward elimination; intermediate results are saved on the hard disk,
and information is transferred between the two matrices during the
procedure; this continues until the entire banded matrix equation is
processed, and nally, back substitution begins by reading the last
saved data entry, one block at a time, and repeated until all saved
blocks have been read and processed. Silva et al. (2005) and Maa et al.
(2002) have demonstrated the efciency of this procedure.
Appendix C. Numerical model for the SWE
Following Rogers (2001), Eqs. (1)(3) are discretised spatially
using nite volumes on a rectangular grid with Roe's ux function
used to represent the nonlinear convective terms, for which the SWE
in integral form can be written as

qd +
S

fdS =

hd 27
where
q =

uh
vh

h =
0

wx

bx
= gs
0x
+ hfv
1

S
xx
x
+
S
xy
y
_ _

wy

by
= gs
0y
+ hfu
1

S
yx
x
+
S
yy
y
_ _
_

_
_

_
28
is the problemdomain, S is the boundary of , f

=f
I
f
V
is the ux
function vector, and f
I
and f
V
are the inviscid and viscous uxes,
f
I
=

uhn
x
u
2
h + g
2
+ 2h
s
= 2n
x
+ uvhn
y
uvhn
x
+ v
2
h + g
2
+ 2h
s
= 2n
y

y f
V
=

0
h
u
x
_ _
n
x
+ h
u
y
_ _
n
y
h
v
x
_ _
n
x
+ h
v
y
_ _
n
y

29
n
x
and n
y
are the Cartesian components of a vector normal to S. is
water density and s
0i
is the bed slope in direction i,
wi
are the tangential
surface stresses,
bi
are the bed shear stresses and S
ii
are the radiation
stress components; all relative to i direction, f is the Coriolis parameter.
The solution to Eq. (27) gives the values of uh, vh and at each cell
centre, for which it is more convenient to write that equation as
Vq
t
j
i
=
C
i

f
i
ds + V
i
h
i
30
where q
i
and h
i
are the cell centre values andV
i
stands for the area of cell
i. The surface integral in Eq. (30) may be evaluated via the expression

C
i

f
i
ds =

f
E
+

f
W
+

f
N


f
S
s 31
with

f
E
;

f
W
;

f N; and

f
S
being the ux vectors through each cell side. s
denotes the length of the cell side.
The estimation of the inviscid uxes was made using the Roe's ux
function
f
I
i; j
=
1
2
f
I
q

i; j
+ f
I
q

i; j
j Aj q

i; j
q

i; j
32
where q
i,j
+
and q
i,j

are the reconstructed Reimann states at the cell

interface and A is the ux Jacobian. The viscid uxes were evaluated
with a central nite difference scheme.
The numerical solution consists of computing the Riemann states
and the ux functions (inviscid and viscous) at each of the cell sides,
which, when added to the source term evaluation, gives the right
hand side of Eq. (30). For the latter, the radiation stress tensor for a
progressive wave travelling with direction from the x axis is
S =
ga
2
4
S
xx
S
xy
S
yx
S
yy
_ _
=
ga
2
4
2C
g
C
cos
2
+
2C
g
C
1
_ _
C
g
C
cossen
C
g
C
cossen
2C
g
C
sen
2
+
2C
g
C
1
_ _
_
_
_
_
_
_
_
_
_
_
_
_
:
33
C and C
g
are the wave and group celerities, respectively and a is
the local wave amplitude which as well as are taken from the
previously computed wave eld thus dening a coupled model for
wave and wave-driven currents.
The time integration is made using a second order Adams
Bashford technique dened as
q
n+1
i
= q
n
i
+
t
V
n
i
3
2
Vq
t
j
n
i

1
2
Vq
t
j
n1
i
_ _
: 34
References
Baquerizo, A., Losada, M.A., 2002. Circulation pattern in bounded coasts. Report on
Tasks M.5.2.2, HUMOR Project.
Baquerizo, A., Losada, M.A., Ortega, J.L., 2002. Rip currents in semi-elliptic bays. Proc.
28th Inter. Conf. Coastal Eng., ASCE, v. 1, pp. 718726.
174 R. Silva et al. / Coastal Engineering 57 (2010) 160175
Battjes, J.A., 1974. Computation of set-up, longshore currents, run-up and overtopping
due to wind-generated waves. Delft University of Technology, Communications on
Hydraulics, 74-2.
Berkhoff, J.C., 1973. Computation of combined refractiondiffraction. Proc. 13th Inter.
Conf. Coastal Eng., ASCE, v. 1, pp. 471490.
Bowen, A.J., 1969a. The generation of longshore currents on a plane beach. J. Mar. Res.
27, 206214.
Bowen, A.J., 1969b. Rip currents, 1: theoretical investigations. J. Geophys. Res. 74 (23),
54675478.
Chamberlain, P., Porter, D., 1995. The modied mild-slope equation. J. Fluid Mech. 291,
393407.
Dally, W., Dean, R., Dalrymple, R., 1985. Wave height variation across beaches of
arbitrary prole. J. Geophys. Res. 90 (C6), 1191711927.
Dalrymple, R.A., 1975. Amechanismfor ripcurrent generation on opencoast. J. Geophys.
Res. 80, 34853487.
Ebersole, B., Darlymple, R., 1979. A numerical modelling for nearshore circulation
including convective accelerations and lateral mixing. Ocean Engineering Report 21,
University of Delaware. Newark, USA.
Hsu, J.R.C., Benedet, L., Klein, A.H.F., Raabe, A., Tsai, C.P., Hsu, T.W., 2008. Appreciation of
static bay beach concept for coastal management and protection. J. Coast. Res. 24 (1),
198215.
Iribarren, R., 1947a. Corrientes y transportes de arena originados por el oleaje, 1. Revista
de Obras Pblicas, 95, tomo I (2785), 195206. (In Spanish).
Iribarren, R., 1947b. Corrientes y transportes de arenas originados por el oleaje, 2.
Revista de Obras Pblicas, 95, tomo I (2786), 243250. (In Spanish).
Kirby, J., Dalrymple, R., 1994. Combined Refraction/Diffraction Model REFDIF 1,
Version 2.5. Documentation and Users Manual. Research Report No. CACR-94-22.
Center for Applied Ocean Research, Department of Civil Engineering, University of
Delaware. Newark, USA.
Lamb, H., 1879. Hydrodynamics. Cambridge University Press. 258pp.
Longuet-Higgins, M.S., 1970. Longshore currents generated by obliquely incident sea
waves-1 and 2. J. Geophys. Res. 75, 67786801.
Longuet-Higgins, M.S., Stewart, R.W., 1964. Radiation stress in water waves: a physical
discussion with applications. Deep Sea Res. 11, 529562.
Maa, J., Hsu, T., Lee, D., 2002. The RIDE model: an enhanced computer programfor wave
transformation. Ocean Eng. 29, 14411458.
McKenzie, P., 1958. Ripcurrent systems. J. Geol. 66, 103113.
O'Rourke, J.C., Le Blond, P.H., 1972. Longshore currents in a semicircular bay. J. Geophys.
Res. 77 (3), 445452.
Reynolds, O., 1895. On the dynamic theory of incompressible viscous ows and the
determination of the criterion. Philos. Trans. R. Soc. 186A, 123164.
Rogers, B., 2001. Rened localised modelling of coastal ow features using adaptive
quadtree grids. PhD Thesis, University of Oxford, 156pp.
Shepard, F.P., Inman, D.L., 1950. Nearshore water circulationrelatedto bottomtopography
and wave refraction. Trans. Am. Geophys. Union 31, 196212.
Silva, R., Borthwick, A., Taylor, E., 2005. Numerical implementation of the harmonic
mild-slope equation. Coast. Eng. 52, 391407.
Zhao, L., Panchang, V., Chen, W., Demirbilek, Z., Chhabbra, N., 2001. Simulation of wave
breaking effects in two-dimensional elliptic harbor wave models. Coast. Eng. 42,
359373.
175 R. Silva et al. / Coastal Engineering 57 (2010) 160175