Ocean Sci.

, 13, 673–690, 2017

https://doi.org/10.5194/os-13-673-2017
© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

Modelling of sediment transport and morphological evolution under

the combined action of waves and currents
Guilherme Franz1 , Matthias T. Delpey2 , David Brito3 , Lígia Pinto1 , Paulo Leitão4 , and Ramiro Neves1
1 MARETEC, Instituto Superior Técnico, Universidade de Lisboa, Av. Rovisco Pais, 1049-001, Lisboa, Portugal
2 Centre
Rivages Pro Tech, SUEZ, 2 allée Théodore Monod, Bidart, France
3 ACTION MODULERS, Estrada Principal, no. 29, Paz, 2640-583 Mafra, Portugal
4 HIDROMOD, Rua Rui Teles Palhinha, no. 4, Leião, 2740-278 Porto Salvo, Portugal

Correspondence to: Guilherme Franz ([email protected])

Received: 24 February 2017 – Discussion started: 21 March 2017

Revised: 17 July 2017 – Accepted: 25 July 2017 – Published: 7 September 2017

Abstract. Coastal defence structures are often constructed ter conditions and during events of highly energetic waves.
to prevent beach erosion. However, poorly designed struc- The MOHID code is available in the GitHub repository.
tures may cause serious erosion problems in the downdrift
direction. Morphological models are useful tools to predict
such impacts and assess the efficiency of defence structures
1 Introduction
for different scenarios. Nevertheless, morphological mod-
elling is still a topic under intense research effort. The pro- The morphological features of the coastal zone depend on the
cesses simulated by a morphological model depend on model sediment characteristics and the combined action of waves
complexity. For instance, undertow currents are neglected in and currents. Wind waves are the main energy source for
coastal area models (2DH), which is a limitation for simulat- most beaches. Particularly in the surf zone, waves may in-
ing the evolution of beach profiles for long periods. Model duce considerable changes in mean sea level and strong cur-
limitations are generally overcome by predefining invariant rents (Longuet-Higgins, 1970a, b, 1983). The pattern of surf
equilibrium profiles that are allowed to shift offshore or on- zone currents varies with the angle of waves approaching
shore. A more flexible approach is described in this paper, the shore as well as with bathymetric heterogeneities, lead-
which can be generalised to 3-D models. The present work ing to longshore and rip currents. Sediment is usually car-
is based on the coupling of the MOHID modelling system ried shoreward during low wave conditions, mainly due to
and the SWAN wave model. The impacts of different de- the asymmetry of waves in shallow waters (Myrhaug et al.,
signs of detached breakwaters and groynes were simulated in 2004). The sediment accumulated during these periods may
a schematic beach configuration following a 2DH approach. be eroded very rapidly under high wave conditions during a
The results of bathymetry evolution are in agreement with the major storm. Following these energetic events, the bottom
patterns found in the literature for several existing structures. profile may recover its initial shape only if the longshore
The model was also tested in a 3-D test case to simulate the transport of sediment during the storm is low. Different struc-
formation of sandbars by undertow currents. The findings of tures such as breakwaters and groynes can prevent the along-
this work confirmed the applicability of the MOHID mod- shore movement of sediment (e.g. Dally and Pope, 1986).
elling system to study sediment transport and morphological Consequently, serious erosion problems in the downdrift di-
changes in coastal zones under the combined action of waves rection may arise from the construction of these structures.
and currents. The same modelling methodology was applied Morphological models are useful tools to assess the impact of
to a coastal zone (Costa da Caparica) located at the mouth protection structures, enabling us to consider different wave
of a mesotidal estuary (Tagus Estuary, Portugal) to evaluate conditions and structure designs.
the hydrodynamics and sediment transport both in calm wa-

Published by Copernicus Publications on behalf of the European Geosciences Union.

674 G. Franz et al.: Modelling of sediment transport and morphological evolution

The complexity of morphological models ranges from approach may be more appropriate in order to consider the
coastal profile models to 2- or 3-D models. Actually, mor- effect of non-uniform grain-size distributions on the overall
phological models are usually a set of different models or sediment transport, through morphological models that ac-
modules, depending on the chosen approach. Here the focus count for multiple sediment fractions. Grain-size sorting is
is on the coupling of a spectral wave model with a phase- generally observed along the cross-shore beach profile, as
averaged hydrodynamic and sediment transport model. Spec- well as in the longshore beach direction (Komar, 1998). Fur-
tral wave models offer a representation of the physical pro- thermore, the method can be generalised to 3-D models for
cesses related to the generation, propagation, and dissipation a better representation of sandbar slopes and to update the
of waves (e.g. Booij et al., 1999). The wave-induced forces shoreline position.
computed by a wave model can be provided to a hydrody- This paper is divided into five sections. A brief descrip-
namic model in order to simulate wave-related phenomena, tion of the effect of waves in the nearshore hydrodynamics
such as wave set-up, wave-induced currents, and mixing. On and sediment transport is given in Sect. 2. The numerical
the other hand, the hydrodynamic model can return water modelling approach is presented in Sect. 3. The methodol-
levels and currents to the wave model (e.g. Warner et al., ogy was verified for different test cases (Sect. 4). Firstly, the
2008). Additional processes can be considered in the hy- morphological evolution of a schematic beach was evaluated
drodynamic model, such as wind action, tidal motion, and for different designs of coastal defence structures (detached
river discharges. The transport of suspended sediment may breakwaters and groynes) following a 2DH model configu-
be simulated by an advection/diffusion model. The mech- ration. The model was applied later in a 3-D configuration
anism of erosion/deposition of sediments is controlled by for the same schematic beach to verify the development of
the bed shear stress induced by currents and waves. The sandbars. Finally, the numerical modelling methodology was
bathymetry evolution resulting from the total sediment trans- applied to assess the hydrodynamics and sediment transport
port (suspended load and bed load) affects the patterns of under extreme wave conditions in a coastal zone (Costa da
currents and waves. Due to the interdependence of the phys- Caparica) located at the mouth of a mesotidal estuary (Tagus
ical processes involved in sediment dynamics, all of these Estuary, Portugal). The main conclusions found from these
models must be coupled. test cases are discussed in Sect. 5.
The hydrodynamics inside the surf zone is influenced by
important 3-D effects. Due to the absence of 3-D processes
(e.g. undertow), coastal area models (2DH) fail to reproduce 2 Background
a consistent evolution of beach profiles for long periods. This
shortcoming can be overcome by predefining invariant equi- The effect of breaking waves on the mean sea level (wave
librium profiles. In this case, the equilibrium profile only set-up) has been known since the laboratory measurements
shifts offshore or onshore depending on the overall sediment performed by Saville (1961), confirmed further by Bowen et
balance along the profile, similarly to a coastal profile model al. (1968). This tilt of the mean sea level is explained by the
(e.g. Kriebel and Dean, 1985; Kristensen et al., 2013). As the horizontal flux of momentum carried by waves or, equiva-
sediment transport in the swash zone is usually neglected in lently, by the radiation stress, a vertically integrated momen-
large-scale 2DH models, this approach also has the advan- tum flux whose gradient balances the wave set-up (Longuet-
tage of updating the shoreline position. However, limitations Higgins and Stewart 1962, 1964). In the surf zone, wave
arise when a structure is present in the surf zone. Moreover, heights and orbital velocities decrease towards the shore due
the processes responsible for sandbar evolution are not con- to wave breaking. As a consequence, the radiation stress also
sidered. Attempts to simulate sandbar dynamics have been decreases, resulting in a force directed towards the shore.
performed generally with cross-shore 2-D (2DV) and quasi- This force is balanced by a hydrostatic pressure gradient that
3-D models (e.g. Drønen and Deigaard, 2007; Ruessink et increases the mean sea level onshore.
al., 2007). Nevertheless, the proper reproduction of sandbar Breaking waves can also drive strong currents in the surf
migration is still an active topic of research (e.g. Dubarbier et zone, which are important for sediment transport and mor-
al., 2015). Inaccuracies in the cross-shore sediment transport phological evolution in the coastal zone. The horizontal mass
may degrade the coastal profile, which is a restriction for the transport associated with waves, or Stokes drift, is oriented
simulation of long-term morphological evolution. shoreward and vertically sheared, being more intense at the
In this work, we test a more flexible approach to overcom- surface (e.g. Ardhuin et al., 2008). As a result, mass conser-
ing the 2DH model limitations in order to simulate morpho- vation in the nearshore is satisfied by a seaward transport in
logical evolution for long periods. Instead of fixing an equi- the lower part of the water column, called undertow, which
librium profile to update the bathymetry and shoreline posi- has an important role in sandbar formation. The undertow is
tion, we defined a maximum slope that when surpassed gen- strongest in steep beaches and may be insignificant for mod-
erates sediment transport in the downslope direction. Thus, erate beach slopes, where circulation tends to break up into
rather than extrapolating erosion or deposition fluxes over rip currents (Longuet-Higgins, 1983). Also, obliquely break-
the entire profile, only individual grid cells are affected. This ing waves generate longshore currents (Longuet-Higgins,

Ocean Sci., 13, 673–690, 2017 www.ocean-sci.net/13/673/2017/

G. Franz et al.: Modelling of sediment transport and morphological evolution 675

1970a, b) and, consequently, longshore sediment transport.

Although nearshore sediment dynamics are dominated by
Z I I
∂
wave action, tidal motion can also play an important role, v H dV = − v H (v.n) dA + υT (∇ (v H ) .n) dA
∂t
alternately moving the breaker zone and shoreline position V A
I Z
A
shoreward and seaward, which may prevent the development 1
− p.nH dA + 2 × v H dV + F , (1)
of longshore bars in the surf zone (e.g. Levoy et al., 2000). ρ
A V
where V represents the control volume, v H = (u, v) the hor-
3 Numerical model izontal velocity vector, v = (u, v, w) the velocity vector, n
the normal vector to the bounding surface (A), nH the nor-
The present work is based on the coupling of the MO-
mal vector related to the horizontal plane,
R η υT the turbulent
HID modelling system (Leitão, 2003; Leitão et al., 2008)
viscosity, ρ the water density, p = g z ρdz + patm the wa-
and the SWAN wave model (Booij et al., 1999). The MO-
ter pressure, g the gravitational acceleration, patm the at-
HID code organisation follows an object oriented strategy
mospheric pressure, η the water level,  the earth rotation
that permits the integration of different scales and processes.
vector, and F the external forces, which include the wave-
Herein, the focus is given to the hydrodynamics and sedi-
induced force (gradient of the radiation stress) computed by
ment transport in the nearshore area. Thus, we considered the
the wave model. The wave-induced force was considered
processes related to the hydrodynamics, turbulence, advec-
in the MOHID hydrodynamic module in previous studies
tion/diffusion of suspended sediment, erosion/deposition of
to compute the effects of waves on sea level (Malhadas et
sediments, bed load sediment transport, and morphological
al., 2009), water renewal in a coastal lagoon (Malhadas et
evolution. A brief description of the most important aspects
al., 2010), and coastal water dispersion in an estuarine bay
of the hydrodynamic model for this work is presented in this
(Delpey et al., 2014).
section, followed by the main novelties implemented in the
The spatial discretisation is performed by following the
MOHID code: a new method to calculate the bed load trans-
finite-volume method. The water level and vertical velocity
port under the combined effect of currents and waves; and a
are computed through the continuity equation integrated over
bed slope correction considered to overcome the 2DH model
the entire water column or applied to each control volume, re-
limitations and to update the shoreline. This work represents
spectively. The equations are solved through the alternating
an extension of the development of the morphological model
direction implicit (ADI) method in an Arakawa C-grid struc-
previously described in Franz et al. (2017).
ture. A generic vertical discretisation allows implementation
The SWAN wave model represents the processes of wave
of different types of vertical coordinates (e.g. Sigma or Carte-
generation, propagation, refraction, shoaling, non-linear
sian) (Martins, 2001). The turbulent viscosity is computed
(quadruplet and triad) wave–wave interactions, and dissi-
differently for the horizontal and vertical directions. The hor-
pation (whitecapping, bottom friction, and depth-induced
izontal turbulent viscosity is defined as a constant value,
breaking). More information can be found in the docu-
based on the grid resolution and a reference velocity, or as a
mentation of the SWAN wave model (http://swanmodel.
function of horizontal velocity gradients, based on Smagorin-
sourceforge.net/). Both MOHID and SWAN can be run with
sky (1963). The vertical turbulent viscosity is computed by
multiple processors using shared (OpenMP) or distributed
the Global Ocean Turbulence Model (GOTM), which is cou-
(OpenMPI) memory architectures. In this work, we imple-
pled to MOHID and consists of a set of turbulence-closure
mented MPI directives in the MOHID module responsible
models (Buchard et al., 1999; Villarreal et al., 2005). Effects
for the calculations of bed load transport and morphologi-
of wave breaking on vertical turbulence can be taken into
cal evolution (sediment module), which can consider multi-
account through surface boundary conditions (Delpey et al.,
ple sediment classes. In addition to speeding up morpholog-
2014).
ical changes by an acceleration factor, this new development
To solve the Navier–Stokes and continuity equation, ap-
reduces the computational time required for modelling bed
propriate boundary conditions are required for the lateral
evolution over long time periods.
(e.g. land and open sea), surface, and bottom boundaries.
3.1 Hydrodynamic model MOHID has the option of a great variety of open boundary
conditions of several types: Dirichlet, Neumann, radiation,
The MOHID hydrodynamic module solves the Navier– cyclic, relaxation (or nudging), etc. Some boundary condi-
Stokes equations, considering the incompressibility, hydro- tions can be combinations of the types enumerated, e.g. a
static, Boussinesq, and Reynolds approximations (Martins, combination of radiation with nudging (Blumberg and Kan-
2000; Leitão, 2003): tha, 1985). Open boundary conditions (OBCs) can be im-
posed by prescribing the values of a specific variable (Dirich-
let boundary condition). This condition is commonly applied
in coastal models to impose tidal levels when the correspon-
dent barotropic velocities are not available. On the other

www.ocean-sci.net/13/673/2017/ Ocean Sci., 13, 673–690, 2017

676 G. Franz et al.: Modelling of sediment transport and morphological evolution

hand, following a Neumann boundary condition, the gradient of boundary cells can be represented in MOHID by a wet-
of a specific variable is imposed instead of a prescribed value. ting/drying scheme (Martins et al., 2001).
Assuming a null gradient condition, the value of a variable at At the surface, fluxes of momentum from wind action and
a boundary point is equal to the value at an adjacent interior wave breaking can be considered (Delpey et al., 2014). At the
point. When the shoreline location and bathymetry are uni- bottom, the method proposed by Soulsby and Clarke (2005)
form, e.g. in schematic cases, cyclic boundary conditions can to compute the bed shear stress was implemented in this
be applied. work, consisting of a steady component due to currents to-
A relaxation scheme can be applied as an OBC by assum- gether with an oscillatory component due to waves. In a lam-
ing a decay time that increases gradually from the boundary inar flow, the combined bed shear stress is a simple linear
to infinite after a defined number of cells (see Martinsen and addition of the laminar current-alone and wave-alone shear
Engedahl, 1987; Engedahl, 1995): stresses. However, in turbulent flows this addition is non-
linear and the mean and oscillatory components of the stress

1t
P t+1t = P ∗ + P ext − P ∗ , (2) are enhanced beyond the values of the laminar case. The
Td mean bed shear stress is used for determining the friction
where P is a generic property, P ∗ is the property value calcu- acting on the current, whereas the maximum shear stress is
lated by the model, P ext is the reference value of the property, used to determine the threshold of sediment motion. The tur-
1t is the model time step, and Td is the relaxation timescale. bulence generated by the skin friction acts directly on bottom
Radiation methods can also be used to impose the OBCs, sediment grains (Einstein, 1950), contrarily to that related to
which allow the propagation of internal disturbances on wa- bed forms. Thus, the threshold of sediment motion depends
ter levels through the open boundaries. These disturbances on the grain-related bed shear stress.
can be caused, for example, by the wave forces. MOHID has
3.2 Sediment transport
two types of radiation conditions (Leitão, 2003), based on
Blumberg and Kantha (1985), Eq. (3), and Flather (1976), The transport of sand is divided into suspended and bed load,
Eq. (4): in which the sand particles are in frequent contact with the
∂η 1 bed. The suspended sediment transport is computed by re-
+ (C E .n) 1η = (ηext − η), (3) solving the advection/diffusion equation. This approach is
∂t Td
more realistic than considering empirical equations based on
q − qext = (ηext − η) (C E .n) , (4)
the instantaneous bed shear stresses, as the suspended load is
where η and q are the water level and specific flow, respec- not in equilibrium with the instantaneous bed shear stresses
tively; ηext and qext are the imposed values of η and q at open in unsteady flows. The net upward flux of suspended sand
boundary points; C E is the celerity of internal water√level depends on the equilibrium concentration near the bottom,
disturbances or the celerity of external waves (CE = gh). estimated by empirical equations available in the literature,
When Td is approximated to infinity in the Eq. (3), the OBC extrapolated following the Rouse profile to the middle of the
becomes totally passive, which means that the water levels at near-bed layer, which in 2DH mode means the middle of the
the boundary points are computed only from the internal wa- water column. The adopted methodology was described pre-
ter levels. An approximated null relaxation timescale means viously in Franz et al. (2017), converging for different num-
that the water level is imposed as ηext . The Flather radiation bers of vertical layers.
condition is mostly used in nested model domains, when ve- The bed load transport under the combined effect of cur-
locities and water levels are known at the open boundaries. rents and waves is computed following the semi-empirical
Many of the OBCs used by MOHID considered the concept formulation of Soulsby and Damgaard (2005). The formu-
of an external (or reference) solution. These solutions can be lation was derived for current plus sinusoidal and asymmet-
provided via input file or via a one-way nesting of a chain rical waves, as well as asymmetrical waves alone. Amoudry
of models. This last process is used to downscale from large- and Liu (2010) obtained a generally good agreement compar-
scale domains to local ones (e.g. Franz et al., 2016). ing the results of Soulsby and Damgaard (2005) formulations
In the case of land points, the closed boundary condition with a sheet flow model, concluding that it can be imple-
is imposed as null fluxes of mass and momentum in the per- mented in both intrawave and wave-averaged models in or-
pendicular direction. However, for the parallel direction two der to study sediment transport. The parallel 8k and normal
boundary conditions can be used in MOHID: no-slip and 8⊥ components of the non-dimensional bed load transport
free-slip (Leitão, 2003). In the no-slip condition, the normal vector in relation to the current direction are
and parallel velocities in the land boundary are assumed to
be equal to zero. This generates a persistent gradient in the
velocity components parallel to the land boundary and, con-
sequently, a persistent sink associated with the horizontal tur-
bulent diffusion of momentum. The covering and uncovering

Ocean Sci., 13, 673–690, 2017 www.ocean-sci.net/13/673/2017/

G. Franz et al.: Modelling of sediment transport and morphological evolution 677

Davis, 2011), respectively; Hw is the significant wave height;

Lw is the wavelength; kw = 2π/Lw is the wave number; and
8k = max 8k1 , 8k2 ; if θmax > θcr , (5a) h is the water depth. When waves propagate into shallow wa-
8k1 = k81 θmk82 (θm − θcr )k83 , (5b) ters, γw becomes more significant, as the term 1/sinh3 (kw h)
1/2 increases when decreasing h. At a given depth, γw tend to
8k2 = k81 (0.9534 + 0.1904 cos 2∅) θw θm be higher for longer incident waves. This effect may be
 
3/2 explained by the fact that longer waves propagate over a
+ k81 0.229γw θw cos ∅ , (5c)
larger distance in limited water depth (h < L/2) than shorter
waves. Thus, they are more affected by wave–bottom inter-
action and, as a consequence, more asymmetric than shorter
0.1907θw2


8⊥ = k81 (θm sin 2∅ + 1.2γw θw sin ∅) ; waves at the same depth.
3/2 3/2
θw + (3/2) θm To compute the sediment fluxes between grid cells, the
if θmax > θcr , (6) components of the non-dimensional bed load transport vector
are rotated to the grid referential (u-axis and v-axis):
where θm and θw are the time-mean and oscillatory part of the 8u = 8k cos ∅c − 8⊥ sin ∅c , (9a)
bed shear stress (non-dimensional), respectively; k81 , k82 ,
and k83 are calibration coefficients that allow one to repre- 8v = 8k sin ∅c + 8⊥ cos ∅c , (9b)
sent different equations for the bed load transport found in where ∅c is the angle between the horizontal velocity vec-
the literature (e.g. Amoudry and Souza, 2011); ∅ is the an- tor and the u-axis. Thus, the bed load transport vector, q =
gle between the wave propagation and current direction; γw (qu qv ), in mass units (kg m−1 s−1 ) is equal to
is a factor that represents the wave’s asymmetry; θcr is the h i1/2
critical non-dimensional bed shear stress, which depends on qu = ρs 8u g (ρr − 1) d 3 , (10a)
sediment diameter and bed-material gradation. The bed load h i1/2
transport is null (8k = 8⊥ = 0) if θcr is greater than or equal qv = ρs 8v g (ρr − 1) d 3 , (10b)
to the maximum non-dimensional bed shear stress (θmax ):
where ρs is the sand particle density (kg m−3 ); ρr is the rela-
θmax = max (θmax1 , θmax2 ) , (7a) tive density (ρs /ρ); and d is the sand representative diameter
 (m).
θmax1 = (θm + θw (1 + γw ) cos ∅)2
1/2 3.3 Bed slope correction
+(θw (1 + γw ) sin ∅)2 , (7b)
 Wave action induces a shoreward sediment transport that has
θmax2 = (θm + θw (1 − γw ) cos (∅ + π))2 no counterpart in 2DH models, leading to sand accumulation
1/2 in the nearshore and increasing the steepness of the beach
+(θw (1 − γw ) sin (∅ + π ))2 , (7c) profile. Actually, undertow currents are responsible for a sea-
ward sediment transport, which may generate sandbars. Di-
The bed load transport vector, 8 = (8k , 8⊥ ), is enhanced verse opposing forces are responsible for creating an equi-
in the presence of waves. Regarding the symmetrical case librium profile, which depends on sediment characteristics
(γw = 0), the effect of the wave’s asymmetry results in an and wave heights (Dean, 1991). To account for the neglected
additional increase in the normal component (8⊥ ), whereas forces in 2DH models, we defined a maximum slope (αmax )
the parallel component (8k ) can be increased or reduced, de- that when exceeded induces sediment transport in the downs-
pending on the angle (∅) between the wave propagation and lope direction. This artificial sediment transport may act as
current direction. The asymmetry factor (γw = θw,2 /θw,1 ) is the undertow, transporting sediment seaward. The mass of
defined as the ratio between the bed shear stress due to the sand (M) in the sediment column and, consequently, the
wave’s second harmonic (θw,2 ) and basic harmonic (θw,1 ), bathymetry are updated when the bottom slope (α) is larger
set to a maximum value of 0.2 (Soulsby and Damgaard, than αmax :
2005). Considering the quadratic friction law to determine t+1 t
Mi,j = Mi,j − 1M, (11a)
the magnitude of the wave’s bed shear stress, the asymmetry
factor is computed as t+1 t
Mi,j +1 = Mi,j +1 + 1M, (11b)

Uw,2
2 
3 π Hw
2 where
γw = = , (8)
Uw,1 4 Lw sinh3 (kw h) 1M = 1zb Aρs (1 − n) , α > αmax (12a)
1M = 0, α ≤ αmax (12b)
where Uw,2 and Uw,1 are the amplitude of near-bed wave-
orbital velocity for the second harmonic and basic harmonic in which n is the sediment porosity, A is the grid cell area,
of Stokes second-order wave theory (e.g. Greenwood and t is an index symbol for time, and i, j are index symbols to

www.ocean-sci.net/13/673/2017/ Ocean Sci., 13, 673–690, 2017

678 G. Franz et al.: Modelling of sediment transport and morphological evolution

identify the grid cell (i – line number; j – column number). 4 Test cases
Considering the u-direction, the bed change in one time step
(1zb ) is computed as 4.1 Coastal defence structures


1zb = min (|α| − αmax ) 1x, 1zbmax , α > 0 (13a) The morphological evolution of a schematic beach was simu-
lated to assess model results, considering the beach response


1zb = min − (|α| − αmax ) 1x, −1zbmax , α < 0 (13b)
for different designs of detached breakwaters and groynes.


where α = zbi,j +1 − zbi,j /1x, 1x is the cell width, zb is Constant wave conditions were defined along the offshore
the distance from the bed to a reference height (e.g. the hy- boundary (1.5 m of wave height, 8 s of peak wave period, and
drographic zero), and 1zbmax is a threshold to avoid numer- 15◦ of peak wave direction), following the JONSWAP spec-
ical instabilities due to large shockwaves. Similar equations trum. The hydrodynamic model was applied in 2DH mode,
are used in the v-direction. Different values of the maximum considering the vertically integrated wave-induced forces.
slope (αmax ) can be defined in wet and dry cells. This method The MOHID domain was defined as 2 km cross-shore by
is based on that of Roelvink et al. (2009), previously applied 3 km alongshore, whereas the SWAN domain was defined
to simulate dune erosion. The method was implemented in as 2 km cross-shore by 9 km alongshore (3 km larger on each
the MOHID code in terms of mass evolution, in order to side of the MOHID domain). The grid resolution was equal
be applied in the future considering multiple sediment frac- in both models, ranging from 50 m × 50 m to 10 m × 10 m.
tions. The shoreline position is also updated following this A larger domain for the wave model was considered to
approach. Other authors have used the same approach to sim- avoid inaccuracies in the lateral boundaries (shadow zones),
ulate the migration of a tidal inlet (Nahon et al., 2012; Fortu- as incident wave energy was imposed only along the off-
nato et al., 2014). shore boundary. To prevent discontinuities in the SWAN
bathymetry, the part of the domain not covered by the MO-
3.4 Model coupling HID domain was updated with the depths of the MOHID
cross-shore boundaries. The bathymetry evolution was al-
The wave-induced force (radiation stress) computed by the lowed only after a warm-up period, considering a morpho-
SWAN wave model is provided to the MOHID hydrody- logical acceleration factor of 365. This means that 1 day
namic model in order to simulate the wave-induced currents of simulation time is equivalent to 1 year of morphological
and wave set-up. Fields of significant wave height, wave pe- changes. The wave forcing was updated in MOHID, as the
riod, wavelength, wave direction, and maximal orbital veloc- bathymetry and water levels were updated in SWAN, with a
ity near the bottom are also provided by SWAN to MOHID in constant frequency of 5 min (or 30 h of morphological evolu-
order to compute the bed shear stress and sediment transport. tion).
On the other hand, the MOHID hydrodynamic model returns The open boundary condition was defined as a null gradi-
fields of water level and current to SWAN. The water level ent for the sediment concentrations in the water column, as
variation caused by the tidal motion changes the breaker zone well as for the sediment mass evolution at the bed (or, equiv-
and shoreline position, affecting waves and sediment trans- alently, for bathymetry). A null gradient condition was also
port. The morphological evolution modifies the currents and imposed at the open boundaries for the normal and tangen-
waves. Thus, the bathymetry changes computed by MOHID tial current velocities. The radiation condition of Blumberg
are also updated in SWAN. The different fields computed by and Kantha (1985) was imposed for water level, Eq. (3),
SWAN must be updated in MOHID, and the different fields assuming a passive condition at the cross-shore boundaries
computed by MOHID must be updated in SWAN, with an ad- (Td = 1e32 s) and an active condition at the offshore bound-
equate frequency for each application depending on the vari- ary (Td = 1e−12 s). Thus, the water level was imposed at
ability of forcing conditions and the speed of morphological the offshore boundary as equal to the initial condition (zero
changes. in this case) to maintain the average water level inside the
The coupling between the MOHID modelling system and model domain; otherwise, it would continuously decrease.
the SWAN wave model was performed by files transferring The effect of lateral friction in land boundaries was con-
through tools developed in the Fortran language to convert sidered for a better representation of the flow around the
the results to the appropriate format and another tool de- groynes. The horizontal viscosity was set as 1.0 m2 s−1 . The
veloped in the Python language to automatically manage sand granulometry was uniform, with a diameter of 0.2 mm.
the runs of the tools and models. At this time, we have fo- The initial bathymetry was defined by considering an equi-
cused on model results instead of numerical efficiency. How- librium profile of the form h = βy 2/3 (Dean, 1991), where β
ever, considering the domain decomposition parallelisation is a constant set to 0.12 m1/3 , and y is the distance to the
approach implemented using MPI directives and the morpho- shoreline. The average slope is approximately 1 : 60 in the
logical acceleration factor, the computational time required first 200 m from the shoreline, decreasing seaward. The max-
to simulate the presented test cases was feasible through the imum slope (αmax ) was defined as 1 : 50 for the bed slope
use of a regular computer with six cores. corrections. The effectiveness of the bed slope correction

Ocean Sci., 13, 673–690, 2017 www.ocean-sci.net/13/673/2017/

G. Franz et al.: Modelling of sediment transport and morphological evolution 679

(a) (b) (c)

Figure 1. Initial bathymetry (a) and bathymetries after 9 years of morphological evolution considering the bed slope correction (b) and
without the bed slope correction (c). The thicker isoline in the bathymetry represents the shoreline, whereas the remaining ones represent the
1, 2, 3, and 4 m isobaths.

Table 1. Detached breakwaters considered for test case scenarios. current and, consequently, longshore sediment transport in
the nearshore zone. The shoreline tends to be parallel to the
Length Distance Ratio (r) Beach response* wave crests, creating asymmetric bulges. For r equal to 0.5
100 500 0.2 Salient and 1.0, longshore currents restricted the size of the salients,
100 200 0.5 Salient preventing the connection with the breakwater. The shore-
200 200 1.0 Salient/tombolo line advances more on the updrift side for larger values of r,
200 100 2.0 Tombolo trapping sediment from the littoral drift. On the other hand,
the downdrift beach erosion increases. When a tombolo is
* Beach response according to Dally and Pope (1986).
formed, the detached breakwater affects the shoreline simi-
larly to a groyne.

was first assessed neglecting coastal defence structures. The 4.1.2 Groynes
bathymetry of the schematic beach shows minor adaptations
to the wave regime after 9 years of morphological evolution Groynes are applied to reduce the littoral drift in the surf
considering the maximum bed slope, whereas without the zone, trapping sediment on the updrift side of the struc-
bed slope correction the bathymetry reaches unrealistically ture, which may cause erosion problems on the downdrift
steep slopes and sand accumulation on the beach (Fig. 1). side. Moreover, the longshore currents are forced to deviate
into deeper water around groynes, causing sediment losses
4.1.1 Detached breakwaters from nearshore to offshore. The morphological impacts of
the groynes are a function of their length from the shore-
Detached breakwaters generate sediment transport from the line. Model results were assessed for two designs of groynes,
adjacent coast to the lee side of the structure, leading to the with lengths of 100 and 200 m (Figs. 6 and 7). As expected,
formation of a bulge or salient in the beach planform. De- greater erosion occurs on the downdrift side for a longer
pending on geometrical features of the breakwater, wave cli- groyne, as more sediment from the littoral drift gets trapped
mate, and sediment availability, the salient may become at- on the updrift side. Furthermore, offshore sediment transport
tached to the breakwater, forming a tombolo. Based on the becomes intensified in the 200 m length groyne design, as
analysis of several existing breakwater projects, Dally and the deviation of longshore currents is more important. In this
Pope (1986) found that a ratio (r) between the breakwater’s case, the retrogradation of the shoreline is similar to the case
length and a distance to the shoreline of less than 0.5 pre- of a detached breakwater in which a tombolo was formed.
vents the development of a tombolo. In contrast, the develop-
ment of a tombolo is assured if r is larger than 1.5, assuming 4.2 Sandbars formation
sufficient sediment supply. Taking these values into account,
we tested the model response for the four different detached In this test case, we verified the model capacity to gener-
breakwater designs described in Table 1. ate sandbars in a 3-D approach. The same domain and sand
Model results for a near-equilibrium planform of the granulometry (0.2 mm) as in the previous 2DH test cases
shoreline agree with the analysis of Dally and Pope (1986), were considered, but without protection structures. Two wave
demonstrating the development of a salient for r equal to heights (1.5 and 1.0 m) were defined in sequence along the
0.2, 0.5, and 1.0 (Figs. 2, 3, and 4), which become attached offshore boundary during periods of 45 days of morpholog-
to the breakwater, forming a tombolo only for r equal to ical evolution (3 h of simulation time with a morphologi-
2.0 (Fig. 5). The obliquity of waves generates a longshore cal acceleration factor of 365). The peak wave period and

www.ocean-sci.net/13/673/2017/ Ocean Sci., 13, 673–690, 2017

680 G. Franz et al.: Modelling of sediment transport and morphological evolution

Figure 2. Model results for the breakwater’s length-to-distance ratio of 0.2. Bathymetry (a, d), waves (b, e), and currents (c, f) for the initial
condition (a, b, c) and near equilibrium (d, e, f) after 9 years. The thicker isoline in the bathymetry represents the shoreline, whereas the
remaining ones represent the 1, 2, 3, and 4 m isobaths.

Figure 3. Model results for the breakwater’s length-to-distance ratio of 0.5. Bathymetry (a, d), waves (b, e), and currents (c, f) for the initial
condition (a, b, c) and near equilibrium (d, e, f) after 18 years. The thicker isoline in the bathymetry represents the shoreline, whereas the
others represent the 1, 2, 3, and 4 m isobaths.

peak wave direction were maintained constant (8 s and 15◦ ). ful for representing the sand motion induced by excessively
For the 3-D case, a larger maximum slope was defined as steep slopes.
1 : 10 for the bed slope corrections, considering that the sea- The water column was divided into five layers of a sigma
ward sediment transport due to undertow currents can now be vertical coordinate and a simple exponential approach was
represented. The maximum slope in 3-D simulations is use- followed to consider the vertical variation of the wave-

Ocean Sci., 13, 673–690, 2017 www.ocean-sci.net/13/673/2017/

G. Franz et al.: Modelling of sediment transport and morphological evolution 681

Figure 4. Model results for the breakwater’s length-to-distance ratio of 1.0. Bathymetry (a, d), waves (b, e), and currents (c, f) for the initial
condition (a, b, c) and near equilibrium (d, e, f) after 27 years. The thicker isoline in the bathymetry represents the shoreline, whereas the
others represent the 1, 2, 3, and 4 m isobaths.

Figure 5. Model results for the breakwater’s length-to-distance ratio of 2.0. Bathymetry (a, d), waves (b, e), and currents (c, f) for the initial
condition (a, b, c) and near equilibrium (d, e, f) after 18 years. The thicker isoline in the bathymetry represents the shoreline, whereas the
others represent the 1, 2, 3, and 4 m isobaths.

induced forces: an exponential decrease in the radiation flux of momentum, which remains equal to the flux given
stress is imposed from the surface to the bottom, follow- by SWAN. The effects of wave breaking on vertical turbu-
ing the same shape as the profile of the orbital velocities, lence were disregarded in this work. The idea here is only
provided by the linear wave theory. The vertical radiation to provide an approximate representation of the vertical dis-
stress profile is designed to conserve the vertically integrated tribution of wave momentum, in order to generate a general

www.ocean-sci.net/13/673/2017/ Ocean Sci., 13, 673–690, 2017

682 G. Franz et al.: Modelling of sediment transport and morphological evolution

Figure 6. Groyne with a length of 100 m from the shoreline. Bathymetry (a, d), waves (b, e), and currents (c, f) for the initial condition (a,
b, c) and near equilibrium (d, e, f) after 9 years. The thicker isoline in the bathymetry represents the shoreline, whereas the others represent
the 1, 2, 3, and 4 m isobaths.

Figure 7. Groyne with a length of 200 m from the shoreline. Bathymetry (a, d), waves (b, e), and currents (c, f) for the initial condition (a,
b, c) and near equilibrium (d, e, f) after 9 years. The thicker isoline in the bathymetry represents the shoreline, whereas the others represent
the 1, 2, 3, and 4 m isobaths.

undertow pattern. Thus, the corresponding results should be more quantitative investigations. The k-ε turbulence-closure
considered as a first qualitative evaluation of the effect of model was used to compute the vertical viscosity, with the
such an undertow in our morphological module, the latter MOHID default parameterisation, whereas the horizontal
being our focus here. It is left for further work to use a more viscosity was set to 1.0 m2 s−1 .
advanced formulation of 3-D wave–current interactions for

Ocean Sci., 13, 673–690, 2017 www.ocean-sci.net/13/673/2017/

G. Franz et al.: Modelling of sediment transport and morphological evolution 683

Figure 8. Simulated sandbars for different wave heights. The horizontal plane of the bathymetry (a, b, c) and a vertical cut with velocity
modulus and vectors (d, e, f). (a, d) Initial condition. (b, e) After 45 days of morphological evolution with 1.5 m wave height. (c, f) After an
additional 45 days of morphological evolution with 1.0 m wave height.

Figure 9. Bathymetries of the Tagus Estuary hydrodynamic model (a) and the wave model (b) created to propagate the waves until the Tagus
Estuary mouth. The domain of the Costa da Caparica model is presented by the black rectangle and the location of the port of Lisbon wave
buoy is indicated by the black triangle.

The open boundary conditions for sediment concentra- The model was capable of representing an undertow pat-
tions in the water column and sediment mass evolution at tern and associated sediment transport that induce the for-
the bed column were defined as a null gradient, as well as mation of longshore sandbars (Fig. 8). A longitudinal cur-
the boundary conditions for normal and tangential velocities, rent is presented in the surf zone, similarly to that observed
as in the previous 2DH test cases. Considering that no struc- in 2DH. However, a cross-shore velocity component is now
tures were included in this test case, the bathymetry evolution represented by the 3-D model. Inside the surf zone, this com-
was expected to be nearly uniform along the beach. Thus, ponent is shoreward near the surface and seaward near the
a cyclic boundary condition was imposed at the cross-shore bottom. As expected, the cross-shore component has oppo-
boundaries together with a Flather radiation condition at the site directions before and after the breaking zone. The sand-
offshore boundary (Eq. 4). bar migrated seaward, changing the location of the break-
ing zone until wave heights decreased to 1.0 m after 45 days.

www.ocean-sci.net/13/673/2017/ Ocean Sci., 13, 673–690, 2017

684 G. Franz et al.: Modelling of sediment transport and morphological evolution

Figure 10. Residual velocity in the Tagus Estuary mouth at the surface (a) and near the bottom (b) obtained from the 3-D baroclinic model
for the Tagus Estuary.

Figure 11. Wave conditions for the winter of 2013/2014 obtained from the Portuguese coast wave model in the location of the port of Lisbon
wave buoy.

At this time, waves were able to propagate further without significant coastline retreat was observed in the Costa da Ca-
breaking, creating a new sandbar nearer the coast. Finally, parica in the last century. Defence structures (groynes) were
the results demonstrate the model’s potential to represent the built around the 1960s to reduce coastal erosion, resulting
formation of multiple sandbars, which are observed in many in some stability until the 2000/2001 winter, when this issue
places (e.g. Dolan and Dean, 1985; Ruessink et al., 2009). started to receive more attention from the local authorities
(Veloso-Gomes et al., 2009). The importance of the prob-
4.3 Costa da Caparica lem has augmented due to urbanisation and tourism develop-
ment. The location near to the Tagus Estuary inlet increases
The hydrodynamics and sediment transport on the southern the complexity of sediment dynamics in this zone.
coast of the Tagus Estuary mouth (Costa da Caparica) are A downscaling approach was followed to provide appro-
evaluated under extreme wave conditions by coupling the priate boundary conditions for the Costa da Caparica model,
MOHID modelling system and the SWAN wave model. A considering previous results of the wave and hydrodynamic

Ocean Sci., 13, 673–690, 2017 www.ocean-sci.net/13/673/2017/

G. Franz et al.: Modelling of sediment transport and morphological evolution 685

Figure 12. Velocity results of the Costa da Caparica model during flood (a, c) and ebb (b, d) tides at the surface (a, b) and near the bottom
(c, d) without considering the wave’s effect on the hydrodynamics.

Figure 13. Wave results of the Costa da Caparica model.

modelling system for the Portuguese coast developed by the ary conditions (water level, current velocities, salinity, and
MARETEC research group (http://forecast.maretec.org). A temperature) for the Costa da Caparica model were provided
new domain was created with 100 m × 100 m of grid resolu- by the 3-D baroclinic model for the Tagus Estuary (see Franz
tion to propagate the waves until the Tagus Estuary mouth. et al., 2014a, b), through the application of the relaxation
The bathymetries of the father hydrodynamic model and fa- scheme together with the Flather radiation condition. The
ther wave model are presented in Fig. 9, showing the domain wave model for the Portuguese coast was validated previ-
of the Costa da Caparica model. The hydrodynamic bound- ously considering the data of the port of Lisbon wave buoy,

www.ocean-sci.net/13/673/2017/ Ocean Sci., 13, 673–690, 2017

686 G. Franz et al.: Modelling of sediment transport and morphological evolution

Figure 14. Velocity results of the Costa da Caparica model during flood (a, c) and ebb (b, d) tides at the surface (a, b) and near the bottom
(c, d) considering the wave’s effect on the hydrodynamics.

among other buoys located along the Portuguese coast (see divided into 10 layers, including five layers in the first me-
Franz et al., 2014c). tre above the bottom with fixed thicknesses ranging from 0.1
The Tagus Estuary is classified as mesotidal, with an av- to 0.3 m, and five sigma layers on top. The exponential ap-
erage tidal height of 2.0 m in the mouth (Lemos, 1972). proach described in the last section was followed to represent
The tide is the main mechanism forcing the flow in the es- the vertical variation of the wave-induced forces. The ver-
tuary, determining current directions and water level varia- tical viscosity was computed by the k-ε turbulence-closure
tions (Franz et al., 2014d). The maximum velocities reach model and the horizontal viscosity was set to 5.0 m2 s−1 . The
2 m s−1 in the estuary mouth. The Tagus River is the estu- bathymetry data for the coast of the Costa da Caparica were
ary’s main freshwater source, with an annual average flow provided by the Portuguese Environment Agency (APA).
of about 300 m3 s−1 . The estuary stratification is strongly re- In a first scenario, the effects of the waves on the cur-
lated to the river flow and tidal cycle. The residual currents rents and sediment transport were neglected. Therefore, just
pattern in the Tagus Estuary mouth is characterised by a jet in the influence of hydrodynamic boundary conditions from the
the inlet channel and two adjacent vortices, causing a residual Tagus Estuary model was taken into account. The results of
recirculation into the estuary in front of the Costa da Caparica the Costa da Caparica model without the wave action demon-
(Fig. 10). strate strong velocities of up to 2 m s−1 at the surface and
The effects of the waves on the currents and sediment 0.5 m s−1 near the bed in the northern zone, with opposite di-
transport were investigated during a high-energy event in the rections on the flood and ebb tides, whereas weak velocities
winter of 2013/2014 caused by the Hercules storm. The wave are found in the remaining coast directed to the estuary in-
conditions for the period of study in the location of the port let in both situations (Fig. 12). The strong tidal currents are
of Lisbon wave buoy reached wave heights higher than 7 m deflected away from the shoreline by the presence of a long
and wave periods of up to 20 s (Fig. 11). A variable grid res- groyne.
olution was defined for the Costa da Caparica domain, rang- During the period of study, the waves propagated from off-
ing from 50 to 10 m near the coast. The water column was shore, mainly with a west-northwesterly (WNW) direction

Ocean Sci., 13, 673–690, 2017 www.ocean-sci.net/13/673/2017/

G. Franz et al.: Modelling of sediment transport and morphological evolution 687

Figure 15. Results of bed load sediment transport for the scenarios without waves (a, b) and under extreme wave conditions (c, d).

(Fig. 11). The bathymetric features (Fig. 9) cause the mod- ric heterogeneities. Thus, the results of sediment transport
ification of the wave propagation direction through refrac- should be seen as a first qualitative assessment. Moreover,
tion to a west-southwesterly (WSW) or southwesterly (SW) the morphological evolution was ignored at this stage.
direction in the nearshore (Fig. 13). The effect of the cur- The patterns of the bed load sediment transport for the sce-
rents and water level variations on wave propagation was narios without waves and under extreme wave conditions are
also considered, with an update frequency of 1 h, the same presented in Fig. 15. Along the coast of the Costa da Ca-
frequency at which the wave forcing was updated in MO- parica, the results of the bed load sediment transport are prac-
HID. The oblique angle of the waves’ incidence generates tically null when the wave action was disregarded, suggest-
a nearshore longitudinal current oriented to the estuary in- ing that the tidal currents are irrelevant for the sediment trans-
let (Fig. 14), reinforcing the velocities observed in the sce- port in this area. By contrast, simulations including wave
nario in which the wave action was neglected. The velocity forcing show a very strong bed load sediment transport due
vectors have a shoreward component at the surface, whereas to waves. However, the importance of the tidal currents for
near the bottom a seaward component is observed, caused by the sediment transport in the inlet of the Tagus Estuary is no-
the vertical variation of the wave-induced forces. Although ticeable. The littoral drift caused by the waves is deflected
the currents were intensified along the coast of the Costa da seaward by the tidal currents during the ebb tide and by the
Caparica, a small reduction of velocities can be observed in longer groynes present near the estuary inlet.
the northern part of the model domain.
Based on a few existing data in the literature, a uniform
granulometry was assumed for the Costa da Caparica model 5 Conclusions
with a diameter of 0.3 mm (Freire et al., 2006). A more repre-
sentative set of granulometry data is necessary to better char- The potential of a new modelling approach to simulate the
acterise the sediment distribution in the model domain, con- impact of different designs of coastal defence structures was
sidering that grain-size sorting is expected to occur due to demonstrated in this paper. The coupling between the MO-
the complex hydrodynamics, wave variability and bathymet- HID modelling system and the SWAN wave model can be
useful for engineering studies in order to evaluate the best

www.ocean-sci.net/13/673/2017/ Ocean Sci., 13, 673–690, 2017

688 G. Franz et al.: Modelling of sediment transport and morphological evolution

solution to protect the coast against erosion. The speed-up Competing interests. The authors declare that they have no conflict
of morphological changes, along with the multiprocessing of interest.
architecture, allows for the modelling of bed evolution for
long periods and for the study of several scenarios. The ac-
celeration factor of 365 allowed us to simulate many years Acknowledgements. The authors are grateful to the Portuguese
of morphological evolution of a schematic beach with con- Environment Agency (APA) for providing the bathymetry data for
stant wave conditions in a feasible computational time. The the coast of the Costa da Caparica. The first author was financed by
the Brazilian National Council for Scientific and Technological De-
bathymetry results reached an equilibrium condition, demon-
velopment (CNPq) under the Ciências Sem Fronteiras programme
strating the model stability. The acceleration factor is user-
(research grant no. 237448/2012-2). MARETEC acknowledges
defined, depending on the variability of forcing conditions the ERDF funds of the Competitiveness Factors Operational
and speed of morphological changes. The morphological Programme (COMPETE), and national funds from the Foundation
changes should not be speeded up (acceleration factor = 1) for Science and Technology (FCT) (project UID/EEA/50009/2013).
to simulate extreme events that occur within few days, as
for the case presented for the Costa da Caparica. A more Edited by: John M. Huthnance
efficient coupling method is currently being developed in- Reviewed by: André B. Fortunato and one anonymous referee
side the MOHID code to further reduce the computational
time. Moreover, the interface of the MOHID modelling sys-
tem (MOHID Studio) was developed to make the coupling
with SWAN straightforward for all users. References
The potential for modelling the evolution of sandbars was
also demonstrated in this paper. In the future, an up-to-date Amoudry, L. O. and Liu, P. L. F.: Parameterization of near-bed pro-
methodology can be applied to resolve the vertical varia- cesses under collinear wave and current flows from a two-phase
tion of wave-induced forces, as well as wave-induced ver- sheet flow model, Cont. Shelf Res., 30, 1403–1416, 10.1016/j.
tical mixing, based on e.g. the generalised Lagrangian mean csr.2010.04.009, 2010.
(GLM) theory implemented in the MOHID code (see Delpey Amoudry, L. O. and Souza, A. J.: Impact of sediment-induced
et al., 2014). The test cases shown in this paper are only a pre- stratification and turbulence closures on sediment transport
and morphological modelling, Cont. Shelf Res., 31, 912–928,
liminary demonstration of model potential, which is thought
https://doi.org/10.1016/j.csr.2011.02.014, 2011.
to be encouraging. The findings of this work confirmed the
Ardhuin, F., Rascle, N., and Belibassakis, K. A.: Ex-
applicability of the MOHID modelling system in studying plicit wave-averaged primitive equations using a gen-
sediment transport and morphological changes in coastal sys- eralized Lagrangian mean, Ocean Model., 20, 35–60,
tems under the combined action of waves and currents. https://doi.org/10.1016/j.ocemod.2007.07.001, 2008.
The application of the described modelling methodology Blumberg, A. F. and Kantha, L. H.: Open boundary condition for
to a coastal zone located near the inlet of a mesotidal estu- circulation models, J. Hydraul. Eng., 111, 237–255, 1985.
ary with strong tidal currents allowed for an assessment of Booij, N., Ris, R., and Holthuijsen, L. H.: A third-generation wave
the hydrodynamics and sediment transport in situations of model for coastal regions: 1. Model description and validation, J.
calm water conditions (no waves) and under extreme wave Geophys. Res.-Oceans, 104, 7649–7666, 1999.
conditions. Although these initial results are just a qualita- Bowen, A. J., Inman, D., and Simmons, V.: Wave “set-down” and
set-Up, J. Geophys. Res., 73, 2569–2577, 1968.
tive assessment of sediment transport, the applicability of the
Burchard, H., Bolding, K., and Villarreal, M. R.: GOTM, a general
modelling methodology to complex cases was demonstrated.
ocean turbulence model: theory, implementation and test cases,
In the future, with a more representative set of data, quan- Space Applications Institute, 1999.
titative studies could be performed, taking into account the Dally, W. R. and Pope, J.: Detached breakwaters for shore protec-
morphological evolution. Furthermore, the methodology can tion, DTIC Document, 1986.
be used to evaluate different designs of defence structures in Dean, R. G.: Equilibrium beach profiles: characteristics and appli-
order to propose a more efficient solution for the coastline cations, J. Coastal Res., 7, 53–84, 1991.
retreat and intense erosion observed in recent years on the Delpey, M., Ardhuin, F., Otheguy, P., and Jouon, A.: Effects of
coast of the Costa da Caparica. waves on coastal water dispersion in a small estuarine bay, J.
Geophys. Res.-Oceans, 119, 70–86, 2014.
Dolan, T. J. and Dean, R. G.: Multiple longshore sand bars in the
Data availability. No data sets were used in this article. upper Chesapeake Bay, Estuar. Coast. Shelf Sci., 21, 727–743,
1985.
Drønen, N. and Deigaard, R.: Quasi-three-dimensional modelling
of the morphology of longshore bars, Coast. Eng., 54, 197–215,
Code availability. The MOHID code updated with the develop-
2007.
ments performed in this work is available in the repository: https:
Dubarbier, B., Castelle, B., Marieu, V., and Ruessink, G.: Process-
//github.com/Mohid-Water-Modelling-System/Mohid.
based modeling of cross-shore sandbar behavior, Coast. Eng., 95,
35–50, 2015.

Ocean Sci., 13, 673–690, 2017 www.ocean-sci.net/13/673/2017/

G. Franz et al.: Modelling of sediment transport and morphological evolution 689

Einstein, H. A.: The bed-load function for sediment transportation Levoy, F., Anthony, E., Monfort, O., and Larsonneur, C.: The mor-
in open channel flows, Citeseer, 1950. phodynamics of megatidal beaches in Normandy, France, Mar.
Engedahl, H.: Use of the flow relaxation scheme in a three- Geol., 171, 39–59, 2000.
dimensional baroclinic ocean model with realistic topography, Longuet-Higgins, M. S. and Stewart, R.: Radiation stress and mass
Tellus A, 47, 365–382, 1995. transport in gravity waves, with application to “surf beats”, J.
Flather, R.: A tidal model of the northwest European continental Fluid Mech., 13, 481–504, 1962.
shelf, Mem. Soc. R. Sci. Liege, 10, 141–164, 1976. Longuet-Higgins, M. S. and Stewart, R.: Radiation stresses in water
Franz, G., Fernandes, R., de Pablo, H., Viegas, C., Pinto, L., Cam- waves; a physical discussion, with applications, Deep Sea Re-
puzano, F., Ascione, I., Leitão, P., and Neves, R.: Tagus Estuary search and Oceanographic Abstracts, Vol. 11, No. 4, Elsevier,
hydro-biogeochemical model: Inter-annual validation and opera- 1964.
tional model update. 3.as Jornadas de Engenharia Hidrográfica, Longuet-Higgins, M. S.: Longshore currents generated by obliquely
Lisbon, Portugal, Extended abstracts, 103–106, 2014a. incident sea waves: 1, J. Geophys. Res., 75, 6778–6789, 1970a.
Franz, G., Campuzano, F., Fernandes, R., Pinto, L., de Pablo, H., Longuet-Higgins, M. S.: Longshore currents generated by obliquely
Ascione, I., and Neves, R.: An integrated forecasting system of incident sea waves: 2, J. Geophys. Res., 75, 6790–6801,
hydro-biogeochemical and waves in the Tagus estuary, Proceed- https://doi.org/10.1029/JC075i033p06790, 1970b.
ings of the Seventh EuroGOOS International Conference, Lis- Longuet-Higgins, M. S.: Wave set-up, percolation and undertow in
bon, 2014b. the surf zone, P. Roy. Soc. Lond. A, 390, 283–291, 1983.
Franz, G., Campuzano, F., Pinto, L., Fernandes, R., Sobrinho, J., Malhadas, M. S., Leitão, P. C., Silva, A., and Neves, R.: Effect of
Simões, A., Juliano, M., and Neves, R.: Implementation and val- coastal waves on sea level in Óbidos Lagoon, Portugal, Cont.
idation of a wave forecasting system for the Portuguese Coast, Shelf Res., 29, 1240–1250, 2009.
Proceedings of the Seventh EuroGOOS International Confer- Malhadas, M. S., Neves, R. J., Leitão, P. C., and Silva, A.: Influence
ence, Lisbon, 2014c. of tide and waves on water renewal in Óbidos Lagoon, Portugal,
Franz, G., Pinto, L., Ascione, I., Mateus, M., Fernandes, R., Leitao, Ocean Dynam., 60, 41–55, 2010.
P., and Neves, R.: Modelling of cohesive sediment dynamics in Martins, F.: Modelação Matemática Tridimensional de escoamen-
tidal estuarine systems: Case study of Tagus estuary, Portugal, tos costeiros e estuarinos usando uma abordagem de coordenada
Estuarine, Coast. Shelf Sci., 151, 34–44, 2014d. vertical genérica, PhD thesis, Universidade Técnica de Lisboa,
Franz, G. A. S., Leitão, P., Santos, A. D., Juliano, M., and Neves, Instituto Superior Técnico, Lisbon, Portugal, 2000.
R.: From regional to local scale modelling on the south-eastern Martins, F., Leitão, P. C., Silva, A., and Neves, R.: 3D modelling
Brazilian shelf: case study of Paranaguá estuarine system, Braz. in the Sado estuary using a new generic vertical discretization
J. Oceanogr., 64, 277–294, 2016. approach, Oceanol. Acta, 24, 551–562, 2001.
Franz, G., Leitão, P., Pinto, L., Jauch, E., Fernandes, L., and Martinsen, E. A. and Engedahl, H.: Implementation and testing of
Neves, R.: Development and validation of a morphological a lateral boundary scheme as an open boundary condition in a
model for multiple sediment classes, Int. J. Sediment Res., barotropic ocean model, Coast. Eng., 11, 603–627, 1987.
https://doi.org/10.1016/j.ijsrc.2017.05.002, in press, 2017. Myrhaug, D., Holmedal, L. E., and Rue, H.: Bottom friction and
Freire, P., Taborda, R., and Andrade, C.: Caracterização das praias bedload sediment transport caused by boundary layer streaming
estuarinas do Tejo, APRH, Congresso da Água, vol. 8, 2006. beneath random waves, Appl. Ocean Res., 26, 183–197, 2004.
Fortunato, A. B., Nahon, A., Dodet, G., Pires, A. R., Freitas, M. C., Nahon, A., Bertin, X., Fortunato, A. B., and Oliveira, A.: Process-
Bruneau, N., Azevedo, A., Bertin, X., Benevides, P., Andrade, based 2DH morphodynamic modeling of tidal inlets: a compari-
C., and Oliveira, A.: Morphological evolution of an ephemeral son with empirical classifications and theories, Mar. Geol., 291,
tidal inlet from opening to closure: The Albufeira inlet, Portugal, 1–11, 2012.
Cont. Shelf Res., 73, 49–63, 2014. Roelvink, D., Reniers, A., Van Dongeren, A., de Vries, J. v. T., Mc-
Greenwood, R. and Davis, R. A.: Hydrodynamics and sedimenta- Call, R., and Lescinski, J.: Modelling storm impacts on beaches,
tion in wave-dominated coastal environments, Elsevier, 2011. dunes and barrier islands, Coast. Eng., 56, 1133–1152, 2009.
Komar, P. D.: Beach Processes and Sedimentation. Upper Saddle Ruessink, B., Kuriyama, Y., Reniers, A., Roelvink, J., and
River, N.J, Prentice Hall, 1998. Walstra, D.: Modeling cross-shore sandbar behavior on the
Kriebel, D. L. and Dean, R. G.: Numerical simulation of time- timescale of weeks, J. Geophys. Res.-Earth, 112, F03010,
dependent beach and dune erosion, Coast. Eng., 9, 221–245, https://doi.org/10.1029/2006JF000730, 2007.
1985. Ruessink, B., Pape, L., and Turner, I.: Daily to interannual cross-
Kristensen, S. E., Drønen, N., Deigaard, R., and Fredsoe, J.: Hybrid shore sandbar migration: observations from a multiple sandbar
morphological modelling of shoreline response to a detached system, Cont. Shelf Res., 29, 1663–1677, 2009.
breakwater, Coast. Eng., 71, 13-27, 2013. Saville, T.: Experimental determination of wave set-up, Proceedings
Leitao, P., Mateus, M., Braunschweig, L., Fernandes, L., and Neves, of the Second Technical Conference on Hurricanes, Report No.
R.: Modelling coastal systems: the MOHID Water numerical lab, 50, 242–252, 1961.
Perspectives on integrated coastal zone management in South Smagorinsky, J.: General circulation experiments with the primitive
America, 77–88, 2008. equations: I. The basic experiment, Mon. Weather Rev., 91, 99–
Leitão, P. C.: Integração de Escalas e Processos na Modelação do 164, 1963.
Ambiente Marinho, 2003. Soulsby, R. and Clarke, S.: Bed shear-stresses under combined
Lemos, P. A. F.: Estuário do Tejo. Administração Geral do Porto de waves and currents on smooth and rough beds, HR Wallingford,
Lisboa, Lisbon, 1972. Report TR137, 2005.

www.ocean-sci.net/13/673/2017/ Ocean Sci., 13, 673–690, 2017

690 G. Franz et al.: Modelling of sediment transport and morphological evolution

Soulsby, R. L. and Damgaard, J. S.: Bedload sediment transport in Warner, J. C., Sherwood, C. R., Signell, R. P., Harris, C. K., and
coastal waters, Coast. Eng., 52, 673–689, 2005. Arango, H. G.: Development of a three-dimensional, regional,
Veloso-Gomes, F., Costa, J., Rodrigues, A., Taveira-Pinto, F., Pais- coupled wave, current, and sediment-transport model, Comput.
Barbosa, J., and Neves, L. D.: Costa da Caparica artificial sand Geosci., 34, 1284–1306, 2008.
nourishment and coastal dynamics, J. Coast. Res., 1, 678–682,
2009.
Villarreal, M. R., Bolding, K., Burchard, H., and Demirov, E.:
Coupling of the GOTM turbulence module to some three-
dimensional ocean models, in: Marine Turbulence: Theories, Ob-
servations and Models, edited by: Baumert, H. Z., Simpson, J. H.,
and Sündermann, J., Cambridge, Cambridge University Press,
225–237, 2005.

Ocean Sci., 13, 673–690, 2017 www.ocean-sci.net/13/673/2017/