T OMAWAC

ValidationManual

Version v8p4
December 1, 2022
 Contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.1 A word of caution 7
1.2 Validation layout 7

2 Presentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1 General 9
2.2 Capabilities 9
2.2.1 Application domain of the model TOMAWAC . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2.2 Wave interactions with other physical factors . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2.3 The physical processes modelled in TOMAWAC . . . . . . . . . . . . . . . . . . . . . . . 11

3 3D coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.1 Purpose 13
3.2 Description of the problem 13
3.3 Geometry and Mesh 13
3.3.1 Wave conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3.4 Results 15

4 Coupling Wind . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.1 Purpose 19
4.2 Description of the cases 19
4.3 Results 19

5 Manche . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
5.1 Purpose 23
5.2 Description of the problem 23
5.3 Results and Comments 24
6 Next_Comput . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
6.1 Purpose 28
6.2 Description of the problem 28
6.3 Results 29

7 Porous . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
7.1 Purpose 30
7.2 Description of the problem 30
7.3 Physical parameters 30
7.4 Geometry and Mesh 30
7.5 Initial and Boundary Conditions 31
7.6 Numerical parameters 31
7.7 Results 31

8 Rip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
8.1 Results 32

9 Spheric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
9.1 Purpose 35
9.2 Description of the problem 35
9.3 Results 36

10 Triplets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
10.1 Purpose 38
10.2 Description of the problem 38

11 Veget . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
11.1 Purpose 40
11.2 Description of the problem 40
11.3 Physical parameters 40
11.4 Geometry and Mesh 40
11.5 Initial and Boundary Conditions 41
11.6 Numerical parameters 41
11.7 Results 41

12 Angular Spread . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
12.1 Purpose 42
12.2 Description of the problem 42
12.3 Results 42

13 Bottom friction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
13.1 Purpose 45
13.2 Description of the problem 45
13.3 Geometry and mesh 45
13.4 Numerical parameters 46
13.5 Initial and Boundary Conditions 46
13.6 Results 46

14 Calais . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
14.1 Purpose 47
14.2 Description of the problem 47
14.3 Results and Comments 47

15 Dean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
15.1 Purpose 52
15.2 Description of the problem 52
15.3 Reference 52
15.4 Physical parameters 52
15.5 Geometry and Mesh 52
15.6 Initial and Boundary Conditions 53
15.7 Numerical parameters 53
15.8 Results 53

16 Deferl_bj78 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
16.1 Purpose 55
16.2 Description of the problem 55
16.3 Physical parameters 56
16.4 Geometry and Mesh 56
16.4.1 Initial and Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
16.5 Numerical parameters 56
16.6 Results 57
16.6.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

17 Fetch Limited . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
17.1 Purpose 59
17.2 Reference experiments 60
17.3 Geometry of the domain and bathymetry 60
17.3.1 Mesh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
17.3.2 Spectro-angular discretization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
17.4 Initial and Boundary conditions 61
17.4.1 Numerical parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
17.5 Results - infinite depth 62
17.5.1 Wave growth depending on the fetch for different wind velocities. . . . . . . . 62
17.5.2 Non-dimensionnal variances and peak frequencies. . . . . . . . . . . . . . . . . . . 63
17.5.3 Variance spectrum for a constant wind. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
17.5.4 Finite depth TOMAWAC results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
17.6 Conclusion 67
17.6.1 Infinite Depth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
17.6.2 Finite depth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
17.6.3 Some Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

18 Impose spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
18.1 Description of the problem 72
18.2 Numerical parameters 73
18.3 Results 74

19 Opposing current . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
19.1 Purpose 78
19.2 Description of the problem 78
19.3 Reference 78
19.4 Physical parameters 78
19.5 Geometry and Mesh 78
19.6 Initial and Boundary Conditions 79
19.7 Numerical parameters 80
19.8 Results 80

20 Reflection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
20.1 Purpose 81
20.2 Description of the problem 81
20.3 Geometry and Mesh 81
20.4 Initial and Boundary Conditions 81
20.5 Results 81
20.6 Test of rotated boundary 82

21 Shoal: Submerged elliptical mound . . . . . . . . . . . . . . . . . . . . . . . . . 84

21.1 Purpose 84
6

21.2 Reference experiments 84

21.3 Test-case description 85
21.3.1 Geometry of the domain and bathymetry . . . . . . . . . . . . . . . . . . . . . . . . . . 85
21.3.2 Meshes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
21.3.3 Initial and Boundary conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
21.4 Numerical parameters 87
21.5 T OMAWAC Results 87
21.6 Conclusion 92

22 Turning wind . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
22.1 Purpose 93
22.2 Description of the problem 93
22.3 Physical parameters 93
22.4 Geometry and Mesh 93
22.5 Initial and Boundary Conditions 93
22.6 Numerical parameters 94
22.7 Results 94

23 Whirl current . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
23.1 Purpose 96
23.2 Description of the problem 96
23.3 Initial and Boundary Conditions 97
23.4 Geometry and Mesh 98
23.5 Numerical parameters 99
23.6 Results 99
23.7 Conclusion 102

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
 1. Introduction

1.1 A word of caution

This document contains information about the quality of a complex modelling tool. Its purpose
is to assist the user in assessing the reliability and accuracy of computational results, and to
provide guidelines with respect to the applicability and judicious employment of this tool. This
document does not, however, provide mathematical proof of the correctness of results for a
specific application. The reader is referred to the License Agreement for pertinent legal terms
and conditions associated with the use of the software.
The contents of this validation document attest to the fact that computational modelling of com-
plex physical systems requires great care and inherently involves a number of uncertain factors.
In order to obtain useful and accurate results for a particular application, the use of high-quality
modelling tools is necessary but not sufficient. Ultimately, the quality of the computational re-
sults that can be achieved will depend upon the adequacy of available data as well as a suitable
choice of model and modelling parameters.

1.2 Validation layout

This validation is presented hereafter using a validation sheet form, each sheet detailing the
physical concepts involved, the physical and numerical parameters used and comparing both
numerical and reference solutions. Then, each sheet displays the following informations:

- Purpose & Problem description : These first two parts give reader short details about
the test case, the physical phenomena involved and specify how the numerical solution
will be validated;

- Reference : This part gives the reference solution we are comparing to and explicits the
analytical solution when available;

- Physical parameters : This part specifies the geometry, details all the physical parame-
ters used to describe both porous media (soil model in particularly) and solute character-
istics (dispersion/diffusion coefficients, soil ≡ pollutant interactions...);

- Geometry and Mesh : This part describes the mesh used in the T OMAWAC computation;

- Initial and boundary conditions : this part details both initial and boundary conditions
used to simulate the case ;
8 Chapter 1. Introduction

- Numerical parameters : this part is used to specify the numerical parameters used (adap-
tive time step, mass-lumping when necessary...);

- Results : we comment in this part the numerical results against the reference ones, giving
understanding keys and making assumptions when necessary.
 2. Presentation

2.1 General
T OMAWAC is a scientific software which models the changes, both in the time and in the spa-
tial domain, of the power spectrum of wind-driven waves and wave agitation for applications
in the oceanic domain, in the intracontinental seas as well as in the coastal zone. The model
uses the finite elements formalism for discretizing the sea domain; it is based on the compu-
tational subroutines of the TELEMAC system as developed by the EDF R&D’s Laboratoire
National d’Hydraulique et Environnement (LNHE). T OMAWAC is one of the models making up
the TELEMAC system The acronym T OMAWAC being adopted for naming the software was
derived from the following English denomination:
TELEMAC-based Operational Model Addressing Wave Action Computation
T OMAWAC can be used for three types of applications:

• Wave climate forecasting a few days ahead, from wind field forecasts. This real time type
of application is rather directed to weather-forecasting institutes such as Météo-France,
whose one mission consists in predicting continuously the weather developments and, as
the case may be, publishing storm warnings.

• Hindcasting of exceptional events having severely damaged maritime structures and for
which field records are either incomplete or unavailable.

• Study of wave climatology and maritime or coastal site features, through the application
of various, medium or extreme, weather conditions in order to obtain the conditions nec-
essary to carry out projects and studies (harbour constructions, morphodynamic coastal
evolutions, ...).

2.2 Capabilities
2.2.1 Application domain of the model T OMAWAC
T OMAWAC is designed to be applied from the ocean domain up to the coastal zone. The limits
of the application range can be determined by the value of the relative depth d/L, wherein d
denotes the water height (in metres) and L denotes the wave length (in metres) corresponding
to the peak spectral frequency for irregular waves.
The application domain of T OMAWAC includes:
 10 Chapter 2. Presentation

• the oceanic domain, characterized by large water depths, i.e. by relative water depths
of over 0.5. The dominant physical processes are: wind driven waves, whitecapping
dissipation and non-linear quadruplet interactions.

• the continental seas and the medium depths, characterized by a relative water depth
ranging from 0.05 to 0.5. In addition to the above processes, the bottom friction, the
shoaling (wave growth due to a bottom rise) and the effects of refraction due to the
bathymetry and/or to the currents are to be taken into account.

• The coastal domain, including shoals or near-shore areas (relative water depth lower
than 0.05). For these shallow water areas, such physical processes as bottom friction,
bathymetric breaking, non-linear triad interactions between waves should be included.
Furthermore, it could be useful to take into account the effects related to unsteady sea
level and currents due to the tide and/or to the weather-dependent surges.

Through a so-called finite element spatial discretization, one computational grid may include
mesh cells among which the ratio of the largest sizes to the smallest ones may reach or even
exceed 100. That is why T OMAWAC can be applied to a sea domain that is featured by highly
variable relative water depths; in particular, the coastal areas can be finely represented.
The application domain of T OMAWAC does not include the harbour areas and, more generally,
all those cases in which the effects of reflection on structures and/or diffraction may not be
ignored.
A first version of a diffraction model is available in T OMAWAC and is able to represent some
diffraction effects. The model presents still some limits. It is highly recommended to use
phase-resolving models when a detailed simulation of diffraction effects is required (e.g. harbor
agitation).

2.2.2 Wave interactions with other physical factors

Several factors are involved in the wave physics and interact to various extents with the waves
changing their characteristics. The following main factors should be mentioned:

• bathymetry and sea bottom geometry (bottom friction, refraction, surf-breaking, non-
linear effects of interactions with the bottom, sand rippling...)

• atmospheric circulation (wind and pressure effects)

• tide pattern (variation of currents and water heights),

• three-dimensional oceanic circulation currents,

• over/underelevations caused by exceptional weather events, resulting in sea levels varia-

tions up to several meters (storm, surges).

The fine modelling of the interactions between these various physical factors and the waves
is generally rather complex and several research projects are currently focused on it. Within
the application domain as defined in the previous paragraph, T OMAWAC models the following
interactions:

• wave-bathymetry interaction: the submarine relief data input into T OMAWAC are con-
stant in time, but the sea level can change in time. In addition to the effects of the sea
level variations in time, T OMAWAC allows to take into account refraction, shoaling, bot-
tom friction and bathymetric breaking. T OMAWAC simulations can take into account
some diffraction effects.
 2.2 Capabilities 11

• wave-atmosphere interaction: this interaction is the driving phenomenon in the wave

generation, takes part in energy dissipation processes (whitecapping, wave propagation
against the wind. . . ) and is involved in the energy transfer. To represent the unsteady
behaviour of this interaction, T OMAWAC requires 10 m wind fields (specification of the
couple of horizontal velocity components) with a time step matched to the weather con-
ditions being modelled. These wind fields can be provided either by a meteorological
model or from satellite measurements.

• wave-current interaction: the sea currents (as generated either by the tide or by oceanic
circulations) may significantly affect the waves according to their intensity. They mod-
ify the refractive wave propagation direction, they reduce or increase the wave height
according to their propagation direction in relation to the waves and may influence the
wave periods if exhibiting a marked unsteady behaviour. In T OMAWAC, the current field
is provided by the couple of horizontal components of its average (or depth-integrated)
velocity at the nodes of the computational grid. T OMAWAC allows to model the frequency
changes caused either by the Doppler effect or by the unsteady currents, as well as by an
heterogeneous current field.

2.2.3 The physical processes modelled in T OMAWAC

Those interactions being taken into account by T OMAWAC have been reviewed and a number of
physical events or processes have been mentioned in the previous paragraph. These processes
modify the total wave energy as well as the directional spectrum distribution of that energy
(i.e. the shape of the directional spectrum of energy). So far, the numerical modelling of these
various processes, although some of them are now very well known, is not yet mature and keep
on providing many investigation subjects. Considering the brief review of physical interactions
given in the previous paragraph, the following physical processes are taken into account and
digitally modelled in T OMAWAC:
—> Energy source/dissipation processes:
• wind driven interactions with atmosphere. Those interactions imply the modelling of
the wind energy input into the waves. It is the prevailing source term for the wave energy
directional spectrum. The way that spectrum evolves primarily depends on wind velocity,
direction, time of action and fetch (distance over which the wind is active). It must be
pointed out that the energy which is dissipated when the wind attenuates the waves is not
taken into account in T OMAWAC.

• whitecapping dissipation or wave breaking, due to an excessive wave steepness during

wave generation and propagation.

• bottom friction-induced dissipation, mainly occurring in shallow water (bottom grain size
distribution, ripples, percolation...)

• dissipation through bathymetric breaking. As the waves come near the coast, they swell
due to shoaling until they break when they become to steep.

• dissipation through wave blocking due to strong opposing currents.

—> Non-linear energy transfer conservative processes:
• non-linear resonant quadruplet interactions, which is the exchange process prevailing at
great depths.

• non-linear triad interactions, which become the prevailing process at small depths.
12 Chapter 2. Presentation

—> Wave propagation-related processes:

• wave propagation due to the wave group velocity and, in case, to the velocity of the
medium in which it propagates (sea currents).

• depth-induced refraction which, at small depths, modifies the directions of the wave-ray
and then implies an energy transfer over the propagation directions.

• shoaling: wave height variation process as the water depth decreases, due to the reduced
wavelength and variation of energy propagation velocity.

• current-induced refraction which also causes a deviation of the wave-ray and an energy
transfer over the propagation directions.

• interactions with unsteady currents, inducing frequency transfers (e.g. as regards tidal
seas).

• diffraction by a coastal structure (breakwater, pier, etc. . . ) or a shoal, resulting in an

energy transfer towards the shadow areas beyond the obstacles blocking the wave prop-
agation. The current version of the diffraction model implemented in T OMAWAC is able
to represent qualitatively some diffraction effects.

It should be remembered that, due to the hypothesis adopted in paragraph 2.2.1 about the
T OMAWAC application domain, the reflection (partial or total) from a structure or a pronounced
depth irregularity is not addressed by the model.
 3. 3D coupling

3.1 Purpose
This test case has been created to test the coupling between T OMAWAC and T ELEMAC -3. Four
different coupling are tested here:

• In the first one that we will call the classical one, the forces due to waves are constant
along the depth of water. This coupling is very close to the coupling that is made with
telemac2d.

• In the second coupling that we will call 3D coupling, all the quantities are dependant of
the depth. This coupling is closer to what happen in reality. For more detail about the
second coupling the reader can read [31].

• The third coupling is like the first one but T OMAWAC and T ELEMAC -3 are defined on
different domains and meshes using TEL2TOM technique (see [7] for a detailed descrip-
tion). The mesh of T OMAWAC is twice coarser and a little bit larger.

• The forth coupling is a 3D coupling where T OMAWAC and T ELEMAC -3 are defined on
different domains and meshes.

3.2 Description of the problem

We took the geometry from the classical test case ’littoral’, a coupling case between T OMAWAC
and T ELEMAC -2 and Sisyphe

3.3 Geometry and Mesh

The beach is 1000 m long, 200 m wide with a regular mesh with elements size of the order
∆x = 20 m and ∆y = 5 m The beach slope (Y=200m) is 5%. The water depth along the open
boundary (Y=0) is h=10m We use a trianglular regular grid. The mesh is as shown on Figure
3.1

3.3.1 Wave conditions

Incoming waves (waves height , period and directions) are imposed offshore at y = 0, such that
Hs = 1 m, Tp = 8 s. The Jonswap spectrum is used. The waves direction is 30 deg relative to
the y−axis.
14 Chapter 3. 3D coupling

Figure 3.1: 2D mesh of the domain in the two first coupling.

Figure 3.2: 2D mesh of the domain in the two last coupling. T ELEMAC -3 up and T OMAWAC
bottom

⇒ Offshore (Y=0): Offshore wave imposed/no littoral current/no set up

Tomawac: The wave height is imposed on the offshore boundary (5 4 4) (Hs=1m), for a wave
period (Tp=8s).

Telemac2D: The current and free surface are imposed to 0 along the offshore boundary (5 5 5).
 3.4 Results 15

3.4 Results
The results are presented Figures 3.4 (Velocity U on a vertical plan) 3.5 (velocity on the bottom)
3.3(Wave heigth Hm0) with the classical coupling.
On Figures 3.6, 3.7 and 3.8, we present the results of the 3D coupling.

Figure 3.3: Wave Heigth calculated by T OMAWAC.

Figure 3.4: Celerity U on a vertical plan at x=500.

16 Chapter 3. 3D coupling

Figure 3.5: Celerity U on the bottom calculated by T ELEMAC -3 d.

Figure 3.6: Wave Heigth calculated by T OMAWAC with the 3D coupling.

3.4 Results 17

Figure 3.7: Celerity U on a vertical plan at x=500.

Figure 3.8: Celerity U on the bottom calculated by T ELEMAC -3 d with the 3D coupling.
18 Chapter 3. 3D coupling

Figure 3.9: Wave Heigth calculated by T OMAWAC with a claccical coupling on different meshes

Figure 3.10: Wave Heigth calculated by T OMAWAC with a real coupling coupling on different
meshes
 4. Coupling Wind

4.1 Purpose
This case has been created to ensure that in case of a coupling between T ELEMAC -2 d or
T ELEMAC -3 d and T OMAWAC, the wind transmitted as a coupling variable is well taken into
account.

4.2 Description of the cases

Several case of coupling are tested. The first one is inheritaded from case littoral but to simplify
we erased the coupling with (sisyphe/gaia) (steering file t2d_littoral.cas). From this case, we
also test the tel2tom technic. We remind here that this technic allows us to couple telemac2d
and tomawac on different meshes and different domain. First we test the technic on the same
mesh (steering file t2d_same.cas) and then on different meshes (steering file t2d_different.cas).
In the case of different mesh tomawac is calculated on a coarser mesh but a little bit bigger.
In a second time, we are testing the coupling with wind in 3 dimension. Note that the wind is still
in 2D since T OMAWAC uses a 2D wind. The first case is a 2D coupling in 3 dimension (steering
file t3d_littoral.cas) and the second case is a real 3D coupling (steering file t3d_3Dcoupling.cas)
In all case, we build the reference with a wind exchanged by file (written by T ELEMAC - A nd
read by T OMAWAC). Note that the reference can not be constructed anymore, since after the
validation, the coupling with the use of a wind file in T OMAWAC is forbidden. Then we verify
that the wind exchanged by coupling leads to a result equal to the result of the reference. Note
that in the case of different mesh, the wind written on the mesh of T OMAWAC is written by
another telemac2d steering file (t2d_diffwind.cas). In that case the boundary conditions are not
the same as for littoral as it would necessitate a dedicated user fortran for that mesh and it is not
the point of this test.
On a practical point of vue, since a wind read by T OMAWAC is interpolated on time step between
2 read, we had to construct our results with a coupling period of 1, so that the wind transmitted
by coupling has the same actualisation.

4.3 Results
We present the results obtained for hm0 in all case, but the most important result is that there is
no difference if the wind is exchanged by file or by coupling. This validates the developpements
done for all the coupling.
20 Chapter 4. Coupling Wind

Figure 4.1: Hm0 in classical 2D coupling.

Figure 4.2: Hm0 with tel2tom and same mesh.

4.3 Results 21

Figure 4.3: Hm0 with tel2tom and different mesh.

Figure 4.4: Hm0 in the case 3D littoral.

22 Chapter 4. Coupling Wind

Figure 4.5: Hm0 in the case 3D coupling.

 5. Manche

5.1 Purpose
This test case has been exhibited from an old version (V5P4), the reference case at this time
was called r2d.V1P3. It has got mainly one interest, to compare results from old versions to the
new version.

5.2 Description of the problem

We simulate the storm that occured in 1990 in Manche. This test was a test described in the
validation of T OMAWAC 1.0 [5]. Because of many changes in the code it had been abandonned.
It now works with the new version but not in parallel because of the fortran user that sets the
boundary.
We discretize with 25 frequencies and 12 directions. The minimal frequency is 0.04177248 and
the frequential ratio 1.1. The mesh is shown on picture 5.1, as we are interested in storm in the
Manche the mesh is finer around the manche. The time step is 300s and we simulate during 14
days.

Figure 5.1: Mesh of the sea.

 24 Chapter 5. Manche

In fact we are going to work with spherical coordinate so the ’real’ bathymetry will be the one
showed on Figure 5.2. In the code if the bathymetry is more than -10 m, it is set to -10 m.

Figure 5.2: Real Bathymetry used in the code.

We take into account the wind through a file that gives the wind in all point of the mesh every
6 hours. On this point, in the original file the time was given with a deprecated format, so we
took the wind from the results of the old simulation which was given in a new format (number
of seconds after 0).
Nowadays time format is given through a number YYYYMMDDHHMMNN, and it used to be
given by MDDHH. As we kept the fortran user to describe boundaries conditions one will read
this format in the keyword DATE DE DEBUT DU CALCUL = 1161500 for 1990 01 16 15:00:00

5.3 Results and Comments

If one run the validation with old results, one will see a difference of 30% wave heigth result,
it is only on a few points (see figure5.3) and it does not spread on all the results, it can be
explained by a bug corrected in characteristic method on boundaries some years ago.
When we observe global results, we would say that there are no difference between reference
and new result (see 5.4 5.5 and 5.6). To run validation we have updated the reference, it is a
result of the version 7.2
As the calculation is made in spherical coordinates but the initial mesh is done in cartesian
coordinates. T OMAWAC calculates a conversion through Fortran user. The result is a little bit
different if we take the conversion from the coordinate of the old version because of simple
precision in the coordinate. That is the reason of the difference between last comparison when
running validation.
5.3 Results and Comments 25

Figure 5.3: In colour points where diff of wave height is more than 1 meter.

Figure 5.4: Wave height for last version.

26 Chapter 5. Manche

Figure 5.5: Wave height for coordinate changed outside the code.

Figure 5.6: Wave height for V1P3 version.

5.3 Results and Comments 27

Figure 5.7: Mean Direction for last version.

Figure 5.8: Mean direction for reference file.

 6. Next_Comput

6.1 Purpose
The aim of this case is to test a computation that follows another one. We start from a calculated
state at a time and we simulate some new time step.

6.2 Description of the problem

The simulation is the same as the shoal test case so one will read that test case for physical
description. We start here from a shoal simulation until time 600s. The result of this previ-
ous simulation is inside the file indicated by the keyword PREVIOUS COMPUTATION FILE
(FICHIER DU CALCUL PRECEDENT in french). This file has been created by adding the
keyword GLOBAL RESULT FILE (FICHIER DES RESULTATS GLOBAUX in french) in the
shoal steering case. We simulate for 600 other seconds. We can compare the result to the one
obtained by shoal test case with 1200 seconds of simulation.

Figure 6.1: Mesh of the domain.

 6.3 Results 29

6.3 Results
After 600 s, the difference with the full shoal simulation is null. The reference file is the result
of shoal with 12000s.

Figure 6.2: Wave heigth hm0.

 7. Porous

7.1 Purpose
This test case has been created to test the dissipation due to a porous media. That is only a test
that checks non regression solution. In order to test the subroutine of the trunk the test has not
any fortran user. This implies to have porous media in all the domain of calculation.

7.2 Description of the problem

We simulate the dissipation due to porous media in a square. The domain is a square defined by
the points (0,0) and (100,100). Since we are using the domain set in qporos, The porous media
is a square between points (40,39) and (50,61).

7.3 Physical parameters

Simulations were carried out with a water depth h = 2.0 m and a wave heigth of 0.56 m at the
incident wave boundary. The media defined in qporos has a porosity rate of 0.8 a damping mass
coefficient of 2.5, and a linear friction coefficient of 1.

7.4 Geometry and Mesh

The computational domain was composed of a flat (slope = 0.0) 2D grid with an aspect ratio of
1 (cross-shore direction):10 (along shore direction). The calculation grid size was set as 2.0 m
in the wave propagation direction. There are 906 nodes and 1500 triangles.
The area where Porous media is taken into account is defined by the file Zone.txt’

Figure 7.1: Mesh of the domain.

 7.5 Initial and Boundary Conditions 31

7.5 Initial and Boundary Conditions

The initial significative heigth is 0.015 m with an initial frequency of 1 Hz and an initial peak
factor of 3.3 and a mean direction of 90◦ and a initial directionnal spread of 0. The boundary
conditions are given by a jonswap spectrum with a boundary pic factor of 3.3Hz a significative
heigth of 0.5656 m and a boundary frequency pic of 1 Hz.

7.6 Numerical parameters

The time step was of 1s and the duration of the computation of 1200 s. The spectro-angular
mesh has 36 directions and 6 frequencies. The frequential ratio was of 1.01 and the minimum
frequency of 0.9705 Hz.

7.7 Results
We can see on Figure 7.2 the impact of porous media compared to Figure 7.3

Figure 7.2: Wave heigth hm0 with porous media.

Figure 7.3: Wave heigth hm0 without porous media.

 8. Rip

This case test rip current

Figure 8.1: Bathymetry and mesh

8.1 Results
8.1 Results 33

Figure 8.2: Celerity U on a vertical plan along y=9.

Figure 8.3: Celerity U on a vertical plan along y=13.6.

34 Chapter 8. Rip

Figure 8.4: Celerity U on a vertical plan along x=11.8.

Figure 8.5: USTOKES on a vertical plan along y=13.6.

 9. Spheric

9.1 Purpose
This test case has been made to test spherical coordinate.

9.2 Description of the problem

The case look like channel sea but we intentionnaly put an infinite depth.
We discretize the spectrum with 31 frequencies and 36 directions. The minimal frequency is
0.04 and the frequential ratio 1.1.
The time step is 180s and we simulate during 50mn.
We take into account the wind through a file that gives the wind in all point of the mesh every
36 hours, so in our case we can consider constant wind (see Figure 9.2). As for current, it is
taken into account through a fileand is update every 1800s so every 10 time steps(see Figure
9.3).

Figure 9.1: Mesh of the domain.

 36 Chapter 9. Spheric

Figure 9.2: Wind.

Figure 9.3: Current.

9.3 Results
We present Figure 9.4 the wave height obtained after 50mn.
9.3 Results 37

Figure 9.4: Wave height hm0.

 10. Triplets

10.1 Purpose
This test case has been created to test the extend the covering by using option of triad interac-
tions that were not used in other cases.

10.2 Description of the problem

We simulate non linear transfers between triads. In the first case we use Lumped Triad Approx-
imation model (LTA) and in the second case we use the SPB model

Figure 10.1: Mesh of the domain.

10.2 Description of the problem 39

Figure 10.2: Wave height for Lumped Triad Approximation model.

Figure 10.3: Wave height for SPB model.

 11. Veget

11.1 Purpose
This test case has been created to test the vegetation dissipation without any fortran user. That
is only a test that checks non regression solution. It refers to the model of [28] but since the
frequency is not monochromatic the result can not be compared to the results they obtained.

11.2 Description of the problem

We simulate the dissipation due to vegetation in a channel.
The random wave transformation model for a flat bottom by Mendez and Losada (2004) is
expressed as follows.
Hrms,0
Hrms =
1 + β̃ x
with
1 sinh3 kαh + 3 sinh kαh
β̃ = √ C̃D bv Hrms,0 k
3 π (sinh 2kh + 2kh) sinh kh

where Hrms,o is the value of root mean square wave height at the wave boundary x=0.

11.3 Physical parameters

Simulations were carried out with a water depth h = 2.0 m and root mean square wave height
Hrms,o (0.4 m) at the incident wave boundary. The vegetation height was taken as equal to the
water depth (αh = 2.0m), the plant area per unit height was bv = 0.04m, the number of plants
per unit area was N = 10units/m2 , and the bulk drag coefficient was C̃D = 1.0. The vegetation
was present in the entire computational domain.

11.4 Geometry and Mesh

The computational domain was composed of a flat (slope = 0.0) 2D grid with an aspect ratio of
1 (cross-shore direction):10 (along shore direction). The calculation grid size was set as 2.0 m
in the wave propagation direction. There are 906 nodes and 1500 triangles.
 11.5 Initial and Boundary Conditions 41

Figure 11.1: Mesh of the domain.

11.5 Initial and Boundary Conditions

The initial significative heigth is 0.015m with an initial frequency of 1 Hz and an initial peak
factor of 3.3 and a mean direction of 90◦ and a initial directionnal spread of 0. The boundary
conditions are given by a jonswap spectrum with a boundary pic factor of 3.3Hz a significative
heigth of 0.5656m and a boundary frequency pic of 1

11.6 Numerical parameters

The time step was of 0.1s and the duration of the computation of 480s. The spectro-angular
mesh has 36 directions and 6 frequencies. The frequential ratio was of 1.01 and the minimum
frequency of 0.0951Hz.

11.7 Results
We do not compare results with an experiment as the choice of not having any fortran user pre-
vent from being in the same condition but [2] obtained good result agreements when comparing
to [28]

Figure 11.2: Wave heigth hm0.

 12. Angular Spread

12.1 Purpose
This test has been created to check a new functionnality in initial spectrum and boundary spec-
trum. Classically the function given at the boundary (or initial spectrum) is of the form:

S( f , θ ) = J( f )G(θ ) (12.1)

with f the frequency, θ the angular dependance, J is often taken as a Jonswap function, and
G(θ ) is a function of cos for example in [29],

2s θ − θ0
 
G(θ ) = G0 cos (12.2)
2

where θ0 denotes the principal wave direction, G0 is a constant so that the integration of G over
θ is unity. S is the spread parameter that sizes the angle of directions. In the three first option of
BOUNDARY (or INITIAL) ANGULAR DISTRIBUTION FUNCTION, s is a constant. Here we
check a fourth option proposed by Goda in [15], where s depends on frequency too.
  5
 smax f

: f ≤ fp
f
s=  p −2.5 (12.3)
 smax f
 : f > f p
fp

Where f p is the peak frequency, and smax should be 10, 25 and 75 for wind waves, swell with
short decay distance, and swell with long decay distance. In our cas we take an intermediary
value GODA COEFFICIENT FOR ANGULAR SPEADING = 35. The default value is 25.

12.2 Description of the problem

The problem is the same as the test case Shoal.

12.3 Results
12.3 Results 43

Figure 12.1: Mesh and bathymetry of the domain.

Figure 12.2: Wave heigth hm0.

44 Chapter 12. Angular Spread

Figure 12.3: Spectrum imposed at the left bounday of the domain.

 13. Bottom friction

13.1 Purpose
This test case aims at testing the effect of friction coefficients used in T OMAWAC, one acting
upon energy whereas the other one acts upon the mean directional frequency. This test case
is an inheritage of a comparison between Cowadis and T OMAWAC. There is no measure to
validate the test, it is only a non regression testing.

13.2 Description of the problem

The waves are let freely propagate in a large body of water. The swell should loose some
energy and its frequency should change only under the effect of friction bottom (the boundary
conditions may be ignored because the domain is very wide).
The expression that is used for computing the friction in T OMAWAC is as follows:
 
σ
Qb f (θ , ω) = −Γ F(θ , ω) (13.1)
g. sinh(kd)

Where Γ, the BOTTOM FRICTION COEFFICIENT, is taken equal to 0.038, its default value.

13.3 Geometry and mesh

The computational domain is a 100 km by 25 km rectangle and the flat bottom elevation is taken
equal to 5m.
The mesh is made with 492 nodes (80 on the boundary) and 902 triangles.

Figure 13.1: Mesh of the domain.

 46 Chapter 13. Bottom friction

13.4 Numerical parameters

Time duration is 30000 s, time step is equal to 100 s, the spectro-angular mesh has 24 angles
and 25 frequences spread on a geometric progression common ratio 1.1 with a minimum of
0.056447393.

13.5 Initial and Boundary Conditions

For both conditions, we take a Jonswap spectrum with a 1 m significant wave heigth, a pic
frequency of 0.1 The angular distribution function follows a cos2s θ distribution with an angular
spreading of 6 and a mean direction of 0.

13.6 Results
In this case we just verify that we still have the same results as the ones obtained during the
comparison with Cowadis [26].

Figure 13.2: MEAN FREQUENCY Fm01

 14. Calais

14.1 Purpose
This test case has been exhibited from an old version. It has got mainly one interest, to compare
results from version 5P8 to the new version.

14.2 Description of the problem

We simulate waves around the harbour of Calais coming from offshore. At the boundary, the
significative heigth is 4m the frequency peak 0.1 and the direction of 135 degree.
We discretize with 23 frequencies and 24 directions. The mesh is shown on picture 14.1

Figure 14.1: Mesh of the neighborood of Calais.

14.3 Results and Comments

If one run the validation with an old version of the code, one will see a difference of 11 degree
on direction in results, 11 Watt on Power and 0.5 m on wave heigth. But as we can see on figure
14.2, 14.3, 14.4,this is not the case when observing global results. Anyway those differences
48 Chapter 14. Calais

come from changes made in the characteristics, so now we take as the reference the results
computed with version 7.2.

Figure 14.2: Wave height for last version.

Figure 14.3: Wave height for V6p0 version.

14.3 Results and Comments 49

Figure 14.4: Wave height for V5p8 version.

Figure 14.5: Direction for last version.

50 Chapter 14. Calais

Figure 14.6: Direction for V6P0 version.

Figure 14.7: Power for last version.

14.3 Results and Comments 51

Figure 14.8: Power for V5P8 version.

 15. Dean

15.1 Purpose
This test case aims to verify the developpement made to take into account the wave energy
dissipation induced by vegetation. When the ratio between vegetation height and water depth is
important, dissipation may become non negligible.

15.2 Description of the problem

In order to carry out an analysis of the influence of plant height, vegetation field width and
breaking on waves propagation, Mendez and Losada (2004) analysed the evolution of the wave
height over a Dean’s shape profile [11] defined as follows:
2
h = 0.25(300 − x) 3
Where h [m] is the water depth, 0.25 the sediment scale parameter, and x=0 is the offshore
boundary.

15.3 Reference
In this test case the prediction of the effects of vegetation is validated with the original equation
and results from Mendez and Losada [28].The reference file fom_dean.slf contains the results
that have been compared to [28] in [2].

15.4 Physical parameters

Two vegetation heights, dv = 1 m and 3 m and a single 100 m long vegetation field, from 50
to 150 m, are used. The number of plants per square meter is N = 20 units/m2 and the plant
area per unit height of vegetation is bv = 0.25 m. The bulk drag coefficient is 0.2. All these
parameters are keywords in the steering file.

15.5 Geometry and Mesh

The computational domain was composed of a flat (slope = 0.0) 2D grid with an aspect ratio of
1 (cross-shore direction):10 (along shore direction). The calculation grid size was set as 2.0 m
in the wave propagation direction. Bathymetry and mesh are shown on figure 15.1. There are
906 nodes and 1500 triangles.
The area where vegetation is taken into account is defined by the file Zone.txt’
 15.6 Initial and Boundary Conditions 53

Figure 15.1: Depth evolution.

15.6 Initial and Boundary Conditions

According to the authors, the incident wave conditions imposed to T OMAWAC on the offshore
boundary are given by Hrms,o = 2.5 m (equivalent to significant wave height Hs = 3.54 m) and
Tp = 10 s. The incident waves are uni-directional random waves as defined in the previous
section and the breaking model used is that of Thornton and Guza (1983) with γ = 0.6 (where
the parameter γ is the proportional control factor indicating the maximum water depth “Hm”
compatible with water depth “d”: ).
The significant initial wave heigth was taken equal to 3.54 m with a peak frequency of 2.2 Hz.
The angular distribution function follows a cos2s θ distribution with an angular spreading of 2,
and a mean direction of 90.

15.7 Numerical parameters

The time step is of 0.1 s and the duration of the computation of 480 s. The spectro-angular mesh
has 36 directions and 6 frequencies. The frequential ratio is of 1.01 and the minimum frequency
of 0.0951 Hz. The frequencies are filtered to keep only the fourth one (0.1 Hz) at 90 degrees to
respect the Tp = 10 s imposed to the boundaries.
The spectrum tail factor was taken at 4.

15.8 Results
The results from Mendez and Losada and T OMAWAC model are compared in Fig. 15.2 below.
Even if the test case is made with 1 m of vegetation, we present on Figure 15.2 the results
obtained for three different heigth of vegetation, 0 m, 1 m and 3 m [2]. The results show very
good agreement between the Mendez and Losada model [28] and T OMAWAC. We notice that
differences seem very small and we can thus conclude that T OMAWAC is able to reproduce
the same wave attenuation as with the random wave transformation model for breaking uni-
directional random waves.
54 Chapter 15. Dean

Figure 15.2: Comparison of Hrms evolution for numerical wave model (T OMAWAC) and ran-
dom wave transformation model (Mendez and Losada) over Dean’s shape profile.

Figure 15.3: Wave height for dv=1m.

Figure 15.4: Average Wave number for dv=1m.

 16. Deferl_bj78

16.1 Purpose
The aim of this case is to test the modelisation of depth-induced breaking. The results are
compared to laboratory data [3]

16.2 Description of the problem

Our test case corresponds to the “run 15” in Battles and Janssen [3]. It is a rectilinear beach with
a bar. The bathymetry of that beach varies as shown on Figure 16.1. The Battle and Janssen’s
breaking model is used:
F1 = θ2π=0 F0 (θ )dθ where F0 (θ ) is the first moment of the action density spectrum over fre-
R

quency.
dF1 H2
= −α1 Qb ω1 m
dt 8π
wherein Hm denotes the maximum wave height. Qb , the fraction of breaking waves or breaking
rate and α1 a numerical constant of order of 1.
Hm is given by a relationship that is derived from de Miche’s criterion:
 
γ1 γ2 k1 d
Hm = tanh
k1 γ1
wherein k1 is related to ω1 by the dispersion relationship.
Qb is assessed, according to the Battjes and Janssen’s theory as the solution of the implicit
equation:
1 − Qb 8F1
=− 2
ln Qb Hm
The “directionnal” version of that source term is based on the assumption that the breaking does
not change the directionnal distribution of energy. The source term writes:
αQb fc Hm2 F( f , θ )
Qbr ( f , θ ) = −
4 m0
The values of numerical constants α1 , γ1 et γ2 as prescribed by Battjes and Janssen are 1, 0.88
and 0.8 respectively. These values, however can and must be matched as a function of the cam-
ber of incident waves and bottom slope. To that purpose the following keyword are provided:
DEPTH-INDUCED BREAKING 1 (BJ) COEFFICIENT ALPHA for α1 , DEPTH-INDUCED
BREAKING 1 (BJ) COEFFICIENT GAMMA1 for γ1 and DEPTH-INDUCED BREAKING 1
(BJ) COEFFICIENT GAMMA2 for γ2 .
 56 Chapter 16. Deferl_bj78

16.3 Physical parameters

The Battjes and Janssen’s experiments were conducted in a channel and in signle-direction
wave conditions. We have then taken a little directionnal spread(s=32) in order to set up similar
condition.

16.4 Geometry and Mesh

The domain is a 25 m by 12 m rectangle. The depth is constant over y and varrying with x as
shown on Figure 16.1 i.e. its value is -0.616 m for x greater than 19 m varies from -0.616 m to
-0.121 m for x between 19.02 m and 9.02 m, from -0.121 m to -0.23 m for x between 9.02 m
and 4.62 m, from -0.23 m to -0.05 m between 4.62 m and 1 m and -0.05 m after.

Figure 16.1: Section depth.

The mesh is free, it is made of 648 triangles and 373 points. The boundary has 96 points.

Figure 16.2: Mesh of the domain.

16.4.1 Initial and Boundary Conditions

The initial spectrum is null.
As for the boundary conditions, a Pierson-Moskowitz-typed spectrum is prescribed at the west
of the domain with a boundary peak frequency of 0.53Hz, a significative heigth of 0.202 m and
a boundary peak factor of 3.3. The boundary principal direction is 270.

16.5 Numerical parameters

The time duration is 35 s with a time step of 0.35 s, the spectro-angular mesh has 24 angles
and 25 frequences spread on a geometric progression common ratio 1.1 with a minimum of
0.168874
 16.6 Results 57

16.6 Results
Figure 16.3 displays the significant wave heights as computed by T OMAWAC. One can see that
T OMAWAC results are quite similar to the laboratory data.
The effect of wave breaking on the mean frequency is not taken into account in this case of
T OMAWAC. As we suppose that energy dissipation does not affect the spectro-angular energy
spread. The energy transfers of the pre-breaking are not taken into account.

Figure 16.3: Comparison of the significative wave heigths.

Figure 16.4: Actual wave heigth.

Figure 16.5: Significative wave heigths for Battjes and Janssen Model.

16.6.1 Conclusion
The breaking induced dissipation of energy is properly represented in T OMAWAC. The effect
on the mean frequency can be neglected.
58 Chapter 16. Deferl_bj78

Figure 16.6: Significative wave heigths for Roelvink’s model.

Figure 16.7: Significative wave heigths for Izumiya and Horikawa’s turbulence model.

Figure 16.8: Breaking wave ration for Izumiya and Horikawa’s turbulence model.
 17. Fetch Limited

17.1 Purpose

The goal of this test-case is to validate the generation (Source/sink terms) and the growth of the
wave spectrum depending of the length of wind action (fetch) with a constant wind defined at 10
meters from the sea surface. The domain used is simple and we want to observe the evolution
of few parameters along its length, as the significant wave height, peak period and variance
spectrum. It is supposed that the wind blew long enough to have reached a steady state.

Thus, this benchmark will allow us to compare and validate the different models of wind gener-
ation (three models), whitecapping dissipation (two models), non-linear quadruplet interactions
(three models but only two are used here). Friction bottom dissipation and eventually depth-
induced breaking dissipation (four models) is also included when a finite depth is imposed. The
models used are presented in the Table 17.1.
 60 Chapter 17. Fetch Limited

Table 17.1: Different models used for this test-case. In all cases, linear wave growth term from
the formula of Cavaleri and Malanotte-Rizzoli (1981) is used.*: Discrete Interaction Approx-
imation,Hasselmann et al., 1985. **: Gaussian Quadrature Method,introduced by Lavrenov
[25] and implemented by Benoit and Gagnaire-Renou [14].

Depth-induced
Wind White Quadruplet Bottom
breaking
Models generation capping transfers Depth friction
dissipation
model model formula coeff.
model
Komen [23] &
test 1 Janssen [20, 21] DIA* inf – –
Janssen [21]
Komen [23] &
test 2 Snyder [30] DIA* inf – –
Janssen [21]
test 3 Snyder [30] Westhuysen [13] DIA* inf – –
test 4 Yan [38] Westhuysen [13] DIA* inf – –
Exact GQM**
test 5 Snyder [30] Westhuysen [13] coarse inf – –
discretization
Exact GQM**
Komen [23] &
test 6 Janssen [20, 21] coarse inf – –
Janssen [21]
discretization
Exact GQM**
Komen [23] &
test 6b Janssen [20, 21] medium inf – –
Janssen [21]
discretization
test 7a Snyder [30] Westhuysen [13] DIA* 180 0.038 –
test 7b Snyder [30] Westhuysen [13] DIA* 60 0.038 –
test 7c Snyder [30] Westhuysen [13] DIA* 30 0.038 –
test 7d Snyder [30] Westhuysen [13] DIA* 15 0.038 –
Battjes &
test 7e Snyder [30] Westhuysen [13] DIA* 5 0.038 Janssen
(V = 20m/s)

17.2 Reference experiments

In order to validate T OMAWAC results, they will be compared to different empirical formulas:

• Infinite depth:
JONSWAP [16]
CERC (1977) [9]
Wilson and Goda [37]
Kahma and Calkoen [22]

• Finite depth:
CERC (1984)[10]

17.3 Geometry of the domain and bathymetry

Domain dimension: 1000 km * 500 km
 17.4 Initial and Boundary conditions 61

test 7a 180 m
test 7b 60 m
test 7c 30 m
Depth constant along the domain:
test 7d 15 m
test 7e 5m
or infinite

17.3.1 Mesh
Several meshes have been tested. The important point highlighted by the analysis is that at the
begining of the fetch the mesh must be sufficiently fine. At the end, the "free" mesh has been
kept because it is the most used in industry and maritime applications (can be adapted to coast).

Figure 17.1: Free and refined mesh ∆xbig ≈ 25000m - ∆xmed ≈ 20000m - ∆xsmall ≈ 10000m.

17.3.2 Spectro-angular discretization

• 50 frequencies ( f1 = 0.04 Hz for U10 = 20 m/s, f1 = 0.08 Hz for U10 = 10 m/s with
geometric spacing q = 1.05)

• 36 directions uniformly distributed

17.4 Initial and Boundary conditions

Jonswap spectrum has been set as initial condition.
The domain is a part of a sea. A constant and homogeneous wind blows perpendicular to a long
and straight coastline. The boundary conditions are shown in Figure 17.2:

17.4.1 Numerical parameters

• time step: 900 s

• physical time reached: 48 h

• caracterization of the computer:

Core: Linux 2.6.32-5-amdb64
Processors: Intel(R) xeon(R) CPU E5620 @ 2.4Ghz
 62 Chapter 17. Fetch Limited

Figure 17.2: Boundary conditions.

• CPU times:
tests 1/2/3/4 test 5/6 tests 6b tests 7
CPU times 3 min 3h25 5h40 25 s

17.5 Results - infinite depth

17.5.1 Wave growth depending on the fetch for different wind velocities.
Figure 17.3 shows the significant wave heights and the peak periods of waves along the fetch
(along the axis y = 250 m ) for the test 1 (the others are not shown because they give the same
conclusions). These curves are obtained after 48 hour of wind action for a wind of 5 to 25 m/s.
It is clearly seen that the wind has a considerable impact on these values and bigger waves are
obtained with stronger wind. Also, stronger is the wind, longer is the fetch needed to reach
the steady state. It can be noticed that the significant wave height does not grow linearly but
following the squared wind velocity.

Figure 17.3: Significant wave heights and peak periods for the first test, U10 = 5 − 10 − 15 −
20 − 25m/s.
 17.5 Results - infinite depth 63

17.5.2 Non-dimensionnal variances and peak frequencies.

Here all the tests are compared, but only with a 20 m/s and 10 m/s winds. The values obtained
with T OMAWAC V6P2 are compared with the empirical formulas: JONSWAP [16], CERC
(1977)[9], Wilson [37], Kahma [22]. The non-dimensional variables are:

• Non-dimensional Fetch: x∗ = Ugx2

10

g2 m0
• Non-dimensional variance: m0 ∗ = 4
U10

U10 f p
• Non-dimensional peak frequency: f p ∗ = g

Where U10 is the wind velocity at 10 meters from the sea.

m0* en fonction de X* vitesse v=10m/s m0* en fonction de X* Vitesse = 20m/s

1,00E-001 1,00E-001

1,00E-002 1,00E-002

1,00E-003 1,00E-003
m0*
m0*

1,00E-004 1,00E-004

1,00E-005
1,00E-005
100 1000 10000 100000
100 1000 10000 100000
X* X*

M0* test1 M0* test2 M0* test3 M0* test4 M0* test5 M0* test1 M0* test2 M0* test3 M0* test4 M0* test5
M0* test6 M0* test6b m0* CERC m0*KCstable m0*KCinstable M0* test6 M0* test6b m0* CERC m0*KCstable m0*KCinstable
m0*Wilson/Goda m0*Jonswap m0*PM-A m0*Wilson/Goda m0*Jonswap m0*PM-A

Fp* Vitesse v=10m/s fp* en fonction de X* Vitesse v=20m/s

0,5
0,5

0,45 0,45

0,4 0,4

0,35 0,35

0,3
0,3
fp*
fp*(X)

0,25
0,25

0,2
0,2

0,15
0,15

0,1
0,1

0,05
100 1000 10000 100000
0,05 X*
100 1000 10000 100000
X*
fp* test 1 fp* test 2 fp* test 3 fp* test 4 fp* test 5
fp* test1 fp* test2 fp* test3 fp* test4 fp* test5 fp* test6
fp* test6b fp* CERC fp* Kcstable fp* Kcinstable f*p Wilson/Goda fp* JONSWAP fp* test 6 fp* test 6b fp* Kcstable f*p Wilson/Goda fp* CERC
fp*PM-A fp* Kcinstable fp* JONSWAP fp* PM-A

Figure 17.4: Comparison of normalized Variance M0* and normalized peak frequencies fp* for
U10 = 10m/s (left) and 20m/s (right), m0* MP-A and fp* PM-A are , respectively, the variance
and the peak frequency limits given by the revisited Pierson-Moskowitz formula from Alves
[1].

Figure 17.4 shows the different results obtained. Generally, all tests give correct results, and
stay close to the CERC, Wilson & Goda and Kahma & Calkoen curves.
The wind impact can be seen by two different behaviours. Indeed, for a 10 m/s wind, the results
are really close to the empirical formula at small fetches. As the fetch grows, the differences
between the models increrase, and the curves began to stray from the empirical formulas. In
this case, the models of tests 1, 6 and 6b give the best results. At 20 m/s, a different behaviour
is noticed: at small fetch the results are quite different from the empirical curves and it is only
when the fetch grows that they match well with the empirical formula. In this case it is the
model of test 5 which gives the best results. It can be concluded that there is a range where
T OMAWAC results are really correct whatever the models used. For strong winds, an exact
 64 Chapter 17. Fetch Limited

resolution of the quadruplet transfers, Snyder’s wind generation model and Westhuysen white
capping model (test 5) give better results at small fetch.
Thus, the analysis of Figure 17.4 shows a good matching between the T OMAWAC wave sim-
ulations and the empirical formulas. A comparison was also done with the revisited Pierson-
Moskowitz asymptotic limits from Alves and Banner [1] for fully developed wind waves, the
U10 -scaled asymptotes is added on Figure 17.4. The formula gives:

ε = (4.02 ± 0.62) ∗ 10−3

ν = (1.23 ± 0.08) ∗ 10−1

It seems that the non-dimensional peak frequency f p ∗ is a little bit over-estimated and the non-
dimensional variance is a bit underestimated. Indeed, the empirical formula given by CERC
(1984), which is used as a criterion for fully developed seas, to calculate the minimum fetch
during the minimum duration (48h), is tUmin10g = 68.8( XUmin2 g )0.67 . Applying this formula we found
10
that the fully developed seas are reached around 1117 km for a 10 m/s wind and around 1672km
for a 20 m/s wind. Thus, the fully developed sea may not be entirely reached at our last point
(X10 = 1000km), which may explain the growth of the variance spectrum for some of the tests
(see next subsection).

17.5.3 Variance spectrum for a constant wind.

In this subsection, we look at the development of the variance spectrum E( f ) along the fetch.
Z 2π
E( f ) = F( f , θ )dθ
0

The spectra are worked out for different points of fetch and for two wind speeds (10 m/s and
20 m/s):

Points fetch
Point 1 25 km
Point 2 50 km
Point 3 100 km
Point 4 150 km
Point 5 200 km
Point 6 300 km
Point 7 400 km
Point 8 800 km
Point 9 750 km
Point 10 1000 km
Dimensional variance spectrum (not shown here) gives correct results with a big growth of the
peak value with the wind.
The non-dimensional frequency is defined by: f ∗ = U10g∗ f .
and the variance spectra are normalised by the peak value of the Pierson-Moskowitz spectrum,
αg2 0.13g
which corresponds to the steady state: EPM ( fPM ) = (2π)4 f 5 exp(−5/4) with f PM = U10 and the
PM
Philipp’s constant α = 0.0081 On Figure 17.5 and 17.6, non-dimensional variance spectra can
be observed. We can notice that as the fetch grows, the variance spectrum amplitude grows and
the peak frequency declines. Moreover, the variance spectrum tends to the Pierson-Moskowitz
spectrum for some cases. The exact quadruplet transfers calculation impact can be seen on the
test 5, 6 and 6b, the shape of the curve changes, the spectrum is more peaked and the maximum
17.5 Results - infinite depth 65

Non dimensional Variance free mesh U10=10m/s t1 Non dimensional Variance free mesh U10=20m/s t1 refined
1 2
point 1 point 1
point 2 point 2
point 3 point 3
point 4 point 4
point 5 point 5
0.8 point 6 point 6
point 7 1.5 point 7
point 8 point 8
point 9 point 9
point 10 point 10
0.6
E(f)/Epm(pm)

E(f)/Epm(pm)
1

0.4

0.5
0.2

0 0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
f* f*

Non dimensional Variance free mesh U10=10m/s t2 Non dimensional Variance free mesh U10=20m/s t2
1 1
point 1 point 1
point 2 point 2
point 3 point 3
point 4 point 4
point 5 point 5
0.8 point 6 0.8 point 6
point 7 point 7
point 8 point 8
point 9 point 9
point 10 point 10
0.6 0.6
E(f)/Epm(pm)

E(f)/Epm(pm)
0.4 0.4

0.2 0.2

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
f* f*

Non dimensional Variance free mesh U10=10m/s t3 Non dimensional Variance free mesh U10=20m/s t3
1 1
point 1 point 1
point 2 point 2
point 3 point 3
point 4 point 4
point 5 point 5
0.8 point 6 0.8 point 6
point 7 point 7
point 8 point 8
point 9 point 9
point 10 point 10
0.6 0.6
E(f)/Epm(pm)

E(f)/Epm(pm)

0.4 0.4

0.2 0.2

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
f* f*

Non-dimensionnal Variance free mesh U10=10m/s t4 Non-dimensionnal Variance free mesh U10=20m/s t4
1 1
point 1 point 1
point 2 point 2
point 3 point 3
point 4 point 4
point 5 point 5
0.8 point 6 0.8 point 6
point 7 point 7
point 8 point 8
point 9 point 9
point 10 point 10
0.6 0.6
E(f)/Epm(pm)

E(f)/Epm(pm)

0.4 0.4

0.2 0.2

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
f* f*

Figure 17.5: Comparaison of normalized spectrum variance for part one of the different tests
and U10 = 10m/s (left) and 20m/s (right).
 66 Chapter 17. Fetch Limited

Non-dimensional Variance free mesh U10=10m/s t5 Non-dimensional Variance free mesh U10=20m/s t5
1 1
point 1 point 1
point 2 point 2
point 3 point 3
point 4 point 4
point 5 point 5
0.8 point 6 0.8 point 6
point 7 point 7
point 8 point 8
point 9 point 9
point 10 point 10
0.6 0.6
E(f)/Epm(pm)

E(f)/Epm(pm)
0.4 0.4

0.2 0.2

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
f* f*

Non dimensional Variance free mesh U10=10m/s t6 refined Non dimensional Variance free mesh U10=20m/s t6 refined
2
point 1 point 1
point 2 point 2
1 point 3 point 3
point 4 point 4
point 5 point 5
point 6 point 6
point 7 point 7
point 8 1.5 point 8
0.8
point 9 point 9
point 10 point 10
E(f)/Epm(pm)

E(f)/Epm(pm)
0.6
1

0.4

0.5

0.2

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
f* f*

Non dimensional Variance free mesh U10=10m/s t6b Non dimensional Variance free mesh U10=20m/s t1
2
point 1 point 1
point 2 point 2
1 point 3 point 3
point 4 point 4
point 5 point 5
point 6 point 6
point 7 point 7
1.5
0.8 point 8 point 8
point 9 point 9
point 10 point 10
E(f)/Epm(pm)

E(f)/Epm(pm)

0.6
1

0.4

0.5

0.2

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
f* f*

Figure 17.6: Comparaison of normalized spectrum variance for part2 of the different tests and
U10 = 10m/s (left) and 20m/s (right).

value is higher, and it allows to observe a spectrum peak overstepping the state value predicted
by Pierson-Moskowitz formula, it is an overshoot (test 6 for U10 = 20m/s and test 6b).
For U10 = 10m/s, the last fetch points (points 7, 8, 9 and 10) get really close spectra. We can
conclude that from a certain fetch, the fully developed sea state is reached. Althought, for the
case U10 = 20m/s, the spectrum continues to grow and the overshoot’s presence shows that this
balance state is not yet reached. Thus, T OMAWAC gives good results which can be improved
with the exact GQM quadruplet transfers model.

17.5.4 Finite depth T OMAWAC results

In this subsection, finite depths are considered: 180 m, 60 m, 30 m, 15 m and 5 m. In regards of
Miche’s criterion (1994), excepted for a 5 meters depth and V = 20 m/s, there is no bathymetric
 17.6 Conclusion 67

breaking dissipation. All the tests are done for the same generation (Snyder), white capping
(Westhuysen) and quadruplet transfer (DIA) models. Figure 17.7 shows the evolution of the
non-dimensional variance along the fetch for the different tests. In order to compare with the
CERC (1984) parametrization, non-dimensional variables are defined as:

• CERC parametrization (1984):

g∗X
X∗ = Ua
1.23
Ua = 0.71 ∗U10
g2 ∗m0
m0∗ = Ua4

It can be noticed that the non-dimensional variance decreases with depth. Indeed, bottom fric-
tion effects are more important at small depths. On the contrary, the peak frequency grows when
the depth increases because the friction has more impact on small frequency waves, which leads
to a growth of the peak frequency.
On Figure 6, we can see that for a 20 m/s wind, T OMAWAC curves are really matching with the
CERC (1984) forecasting, particularly for d = 15 − 30 − 60 m. Indeed, it is adviced in CERC
publication to use their model for a depth between 15 and 90 meters. But the results are still
good for d = 180 m and d = 5 m.
For a 10m/s wind, except for d = 5 m and d = 15m, the results are less close to the CERC curves.
At small fetches, T OMAWAC results are overvalued, and at large fetches, they are undervalued.
In CERC publication (1984), it is adviced to use an infinite model for depth over 90m. So the
CERC infinite depth model (1977) is added to the graphs and we can see that the results are
closer to it for a 10 m/s wind and d ≥ 30m.For a 20 m/s wind,at d = 180m and at large fetches,
T OMAWAC results tends to the infinite depth limit. Generally, T OMAWAC gives correct results
if you adapt the good model at the application.

17.6 Conclusion
17.6.1 Infinite Depth
This benchmark allows to verify that T OMAWAC results for simultaneous processing of different
source terms are correct. A comparison with empirical formulas validates this simulation with
generally matching results. The use of the exact GQM quadruplet transfers model allows the
visualization of overshoots, but,CPU time are increased, so it is not advisable to use this model
for big industrial cases.

17.6.2 Finite depth

Even if a 1000 km fetch with only 5 m depth is hard to study in reality, T OMAWAC results get
close to the empirical formula. Bottom friction effect seems to be well represented.

17.6.3 Some Tests

We present Figures 17.10 17.11 and 17.12 some results of other tests. The references for the
comparison are made with the version 7.2 of T OMAWAC. Those test cases having non linear
terms it can happen that a different compiler gives a different result. Raising the NUMBER OF
ITERATIONS FOR THE SOURCE TERMS will lower this difference.
68 Chapter 17. Fetch Limited

M0* profondeur finie V = 10m/s

d= 180 - 60 - 30 - 15 - 5 m
1,00E-002

1,00E-003
m0*

1,00E-004

1,00E-005
100 1000 10000 100000
X* = X*g/Ua

CERC1 CERC2 CERC3 CERC4

CERC5 M0*180 M0*60 M0*30
M0*15 M0*5 CERC inf

M0* profondeur finie V = 20m/s

d= 180 - 60 - 30 - 15 - 5
m
1,00E-002

1,00E-003
m0*

1,00E-004

1,00E-005
100 1000 10000 100000
X* = X*g/Ua
CERC1 CERC2 CERC3 CERC4
CERC5 M0*180 M0*60 M0*30
M0*15 M0*5 CERC inf

Figure 17.7: Comparaison of normalized Variance M0* for U10 = 10m/s (high) and 20m/s
(bottom) for different depths: 5, 15, 30, 60, 180 m.
17.6 Conclusion 69

Figure 17.8: Wave heigth for infinite depth.

Figure 17.9: Wave heigth for finite depth.

70 Chapter 17. Fetch Limited

Figure 17.10: Wave heigth for test4.

Figure 17.11: Wave heigth for test6.

17.6 Conclusion 71

Figure 17.12: Wave heigth for test6b.

 18. Impose spectra

18.1 Description of the problem

This example shows how to impose spectra on the boundary of a nested grid in T OMAWAC. To
do so, two simulations are done. One on a larger domain, which we will call the oceanic mesh,
and another smaller mesh which fits inside the first mesh. This second mesh will be called the
Coastal mesh, and as it is often the case, it will be more refined than the Oceanic mesh. The
larger mesh will be used to force the smaller mesh. The only requirement to do this, is that all
the points on the open boundary of the smaller mesh need to be present in the larger mesh. This
is illustrated in Figure 18.1.
The geometrical parameters of the meshes can be found in Table 18.1.

Table 18.1: Geometrical parameters for the meshes used in this test case.

Oceanic Mesh Coastal Mesh

Number of elements 3676 4596
Number of nodes 1899 2460
Maximum edge length (m) 10.0 10.9
Minimum edge length (m) 4.73 0.803

The energy spectrum will be decomposed in 36 directions and 32 frequencies. The minimal
frequency is equal to 0.04 Hz and the frequential ratio is 1.055. The initial water depth is 5 m.
The modelling time step is equal to 1 s, and the simulation will run for 100 time steps (which is
enough to reach a steady state. 3 subiterations will be used to compute the source terms.
Since this is a non-regression test, several modelling options could have been chosen, and the
choices made will now be listed for reproducibility. To generate the waves a uniform wind will
blow over the whole domain. Its velocity is equal to (20, 0) m/s. The WIND GENERATION
is in accordance with WAM cycle 3 [30]. WHITE CAPPING DISSIPATION is in accor-
dance with Van des Westhuysen (2007) [13]. BOTTOM FRICTION DISSIPATION is in
accordance with WAM cycle 4 [6, 16]. DEPTH-INDUCED BREAKING DISSIPATION is
in accordance with Thornton et Guza (1983) [32]. NON-LINEAR TRANSFERS BETWEEN
FREQUENCIES follows WAM cycle 4 (DIA Method) [17]. The LINEAR WAVE GROWTH
follows Cavaleri and Malanotte-Rizzoli (1981) [8, 34].
 18.2 Numerical parameters 73

200

150
y (m)
100

50

0
200

150
y (m)

100

50

0
−200 −150 −100 −50 0 50 100 150 200
x (m)

Oceanic Mesh Coastal Mesh

Figure 18.1: Illustration of the meshes of the test case. The mesh in blue is the large Oceanic
mesh, and the mesh in red is the smaller nested Coastal mesh.

18.2 Numerical parameters

To force the Coastal simulation with the Oceanic results, the spectra of the points in the Oceanic
mesh that are on the open boundary of the Coastal mesh need to be outputed.This is done most
easily using an external text file containing a list of the coordinates of the points to output. In
the steering file of the Coastal mesh this given using the following keyword:

/--------------------------------------------------------------------/
/ WRITING SPECTRA
/--------------------------------------------------------------------/
PUNCTUAL RESULTS FILE = ’./OceanicResults.spe’
FILE WITH COORDINATES OF SPECTRA TO WRITE =
’./SpectraOutputOceanic.dat’

On the Coastal mesh, these spectra are then imposed on the open boundary. To do so, the points
along the open boundary need to have the boundary condition 5 4 4 4. The imposed spectra
file, and the coordinates of each spectra are given using the following keywords:

/--------------------------------------------------------------------/
/ IMPOSE SPECTRA ON THE OPEN BOUNDARY
/--------------------------------------------------------------------/
IMPOSED SPECTRA FILE = ’./OceanicResults_dt10.spe’
/IMPOSED SPECTRA FILE FORMAT = ’SERAFIN ’
FILE WITH COORDINATES OF SPECTRA TO IMPOSE =
’./SpectraOutputOceanic.dat’
/TIME UNIT OF IMPOSED SPECTRA FILE = 1.
 74 Chapter 18. Impose spectra

/TIME SHIFT OF IMPOSED SPECTRA FILE = 0.

18.3 Results

This test case serves mostly to illustrate how to impose spectra from a large mesh on a smaller
nested mesh. Therefore, it will only serve as a non-regression test. The first check is to find the
differences on the two-dimensional mesh with the reference file, see Table 18.2.

Table 18.2: Summary of the differences with the reference files for the 2D results. The last
column is the maximum accepted difference.

Oceanic - Reference Coastal - Reference Oceanic - Coastal

eps
scalar 2 proc scalar 2 proc scalar 2 proc
WAVE HEIGHT HM0 0 0 0 0 -0.000156 -0.000156 0.01
MEAN DIRECTION 0 0 0 0 -0.00177 -0.00177 0.01
WAVE SPREAD 0 0 0 0 -0.038 -0.038 0.01
MEAN PERIOD TMOY 0 0 0 0 0.000478 0.000478 0.01
PEAK PERIOD TPR5 0 0 0 0 0.000314 0.000314 0.01

The spectra on the nodes along the open boundary of the Coastal mesh will also be checked for
non-regression. These spectra are checked with the reference file, and between the Oceanic and
Coastal results. This last check is to ensure that what is imposed is the same as what is seen in
the code. See Table 18.3.
18.3 Results 75

Table 18.3: Summary of the differences with the reference files for the spectra results. Note,
the name of the variables are only true for the Oceanic results. The last column is the maximum
accepted difference.

Oceanic - Reference Coastal - Reference Oceanic - Coastal

eps
scalar 2 proc scalar 2 proc scalar 2 proc
F00001PT2D000002 1.15e-41 1.15e-41 8.43e-39 1.15e-41 0 0 1e-07
F00002PT2D000003 1.15e-41 1.15e-41 8.43e-39 1.15e-41 0 0 1e-07
F00003PT2D000026 1.85e-41 1.85e-41 8.38e-39 1.85e-41 0 0 1e-07
F00004PT2D000027 4.13e-41 4.13e-41 1.18e-38 4.13e-41 0 0 1e-07
F00005PT2D000028 1.86e-41 1.86e-41 8.24e-39 1.86e-41 0 0 1e-07
F00006PT2D000029 1.62e-41 1.62e-41 8.1e-39 1.62e-41 0 0 1e-07
F00007PT2D000030 1.34e-41 1.34e-41 7.81e-39 1.34e-41 0 0 1e-07
F00008PT2D000031 2.37e-41 2.37e-41 9.64e-39 2.37e-41 0 0 1e-07
F00009PT2D000032 2.64e-41 2.64e-41 8.79e-39 2.64e-41 0 0 1e-07
F00010PT2D000033 1.74e-41 1.74e-41 6.47e-39 1.74e-41 0 0 1e-07
F00011PT2D000034 1.45e-41 1.45e-41 5.86e-39 1.45e-41 0 0 1e-07
F00012PT2D000035 2.12e-41 2.12e-41 9.08e-39 2.12e-41 0 0 1e-07
F00013PT2D000036 1.46e-41 1.46e-41 5.99e-39 1.46e-41 0 0 1e-07
F00014PT2D000037 1.76e-41 1.76e-41 6.67e-39 1.76e-41 0 0 1e-07
F00015PT2D000038 2.6e-41 2.6e-41 8.81e-39 2.6e-41 0 0 1e-07
F00016PT2D000039 2.39e-41 2.39e-41 1.01e-38 2.39e-41 0 0 1e-07
F00017PT2D000040 1.34e-41 1.34e-41 7.81e-39 1.34e-41 0 0 1e-07
F00018PT2D000041 1.7e-41 1.7e-41 8.1e-39 1.7e-41 0 0 1e-07
F00019PT2D000042 2.04e-41 2.04e-41 8.24e-39 2.04e-41 0 0 1e-07
F00020PT2D000043 6.26e-41 6.26e-41 1.08e-38 6.26e-41 0 0 1e-07
F00021PT2D000044 2.12e-41 2.12e-41 8.38e-39 2.12e-41 0 0 1e-07

Nonetheless, even if the spectra along the open boundary of the Coastal mesh is the same as
in the Oceanic mesh, there are differences in the domain (due to the nature of the boundary
conditions). This is illustrated with the last columns 6 ans 7 of Table 18.2. These differences
are minimal, as can be seen in Figures 18.2-18.4, where the values along a cross-section taken
at y = 100 m are plotted.
76 Chapter 18. Impose spectra

Figure 18.2: Cross-section of the wave height Hm0 along y = 100 m for the Oceanic and Coastal
simulations. Note, only the scalar simulations are compared.

Figure 18.3: Cross-section of the wave period tmoy along y = 100 m for the Oceanic and Coastal
simulations. Note, only the scalar simulations are compared.
18.3 Results 77

Figure 18.4: Cross-section of the wave spread along y = 100 m for the Oceanic and Coastal
simulations. Note, only the scalar simulations are compared.
 19. Opposing current

19.1 Purpose
The goal of this test-case is to check the behaviour of T OMAWAC in presence of a strong oppos-
ing current. When water meets the strong adverse current, with a velocity that approaches the
wave group velocity, waves are blocked. Without any option for strong current, the amplitud of
waves is overestimated.

19.2 Description of the problem

We present here the simulation of the test case of Lai [24]. We present two different options
that prevent from overestimation.
Option 1: consider an equilibrium range spectrum (in the presence of ambient flow) applied as
an upper limit for the spectrum Hedges et al ([18])
Option 2: add a dissipative term on the right-hand side of the action balance equation [12]
For more details on formulations, one can refer to T OMAWAC documentation, paragraphs 4.2.3.8.1
and 4.2.3.8.2.

19.3 Reference
The flume experiment of Lai et al [24] investigates the transformation of the wave spectrum
on a strong negative current gradient in a flume of 8 m length and 0.75 m depth. An opposing
current flow is induced along the flume according to figure 19.1

19.4 Physical parameters

Concerning the modelling of the opposing current with the first option the Phillips’s constant in
the Pierson-Moskowitz spectrum is equal to 0.0081. For the second option of this modelling,
the dissipation coefficient is equal to 0.65, and saturation threshold value is taken to 1.7510−3 .
The white-capping dissipation is the model of Westhuysen 2008. Non linear transferts between
frequencies are calculated by the DIA method.

19.5 Geometry and Mesh

The bathymetry is as described on figure 19.2.
The mesh is made of 1701 nodes and 3200 triangles and is shown Figure 19.3.
 19.6 Initial and Boundary Conditions 79

Figure 19.1: Current of the test case.

Figure 19.2: Bathymetry of the test case.

Figure 19.3: Mesh of the domain.

19.6 Initial and Boundary Conditions

For both conditions, we take a Jonswap spectrum with a 1.9 cm significant wave heigth, a peak
frequency of 2.2. The angular distribution function follows a cos2s θ distribution with an angular
spreading of 65 and a mean direction of 90.
 80 Chapter 19. Opposing current

19.7 Numerical parameters

Time duration is 400 s, time step is equal to 0.1 s, the spectro-angular mesh has 72 angles and
36 frequences spread on a geometric progression common ratio 1.1 with a minimum of 0.25.
The option for wave growth limiter is following the Hersbach et Janssen (1999) parameterisa-
tion.

19.8 Results
We show the results obtained Figure 19.4. The two options are efficient to reduce the wave
heigth overestimation and lead to a solution closer to measurements. This shows the interest of
including these options in the case of a strong opposing current.

Figure 19.4: Heigth comparison without wave blocking and with the two options.

Figure 19.5: Heigth comparison with the two options for the last validation.
 20. Reflection

20.1 Purpose
This test case is an example of reflection effect with T OMAWAC

20.2 Description of the problem

The domain is homogeneous with the upper and the bottom face reflective.

20.3 Geometry and Mesh

The domain is a rectangle of 400 m by 100 m.

Figure 20.1: Mesh of the domain.

20.4 Initial and Boundary Conditions

The waves are coming from the right boundary with an angle of 300 ◦ , a wave height of 0.2 and
a peak period of 1.88 s.

20.5 Results
Seeing the result figure 20.2, the reflection seems to work correctly.
 82 Chapter 20. Reflection

Figure 20.2: Wave height hm0.

20.6 Test of rotated boundary

In order to check that the reflection condition works for any boundary (not parallel with axes
like in previous section), we test the same configuration on a rotated geometry of 30◦ see figure
20.3

Figure 20.3: Mesh of the domain 2.

We can see on figure 20.4 that the results are simply rotated from those of figure 20.2.
Plotting the wave height on point [0,0] of first geometry and [0,0] of second geometry, we can
see that it is the same result.
20.6 Test of rotated boundary 83

Figure 20.4: Wave height hm0.

Figure 20.5: Wave height on point [0,0].

 21. Shoal: Submerged elliptical mound

21.1 Purpose
The goal of this test-case is to check the behaviour of T OMAWAC in presence of refraction,
shoaling and diffraction effects over a shoal.
A shoal causes a concentration area of wave just behind it, what makes fail numerical mod-
els of refraction based on the radius theory. This test-case allows to verify simulations from
T OMAWAC for refraction, shoaling and diffraction processes.

21.2 Reference experiments

The references of this test-case are the experiments of [35]. They worked on the refraction and
the diffraction of irregular waves over a mound using a directional spectral wave generator and
they measured the wave’s height.

Figure 21.1: Set up of Vincent and Briggs experiments.

Different irregular wave conditions were tested, varying the narrowness and the directional
spread of the spectrum. They used a TMA shallow-water spectrum (Bouws et al. 1985) and
a wrapped normal spreading function which can be close to a Mitsuyasu directional spreading
function.
 21.3 Test-case description 85

Table 21.1: Test conditions for shoal test series, Vincent and Briggs experiments.

Test Period Height σm

Case ID Type α γ
number (sec) (cm) (deg)
(a) Initial Series
1 M1 Mono 1.30 5.50 – – –
2 N1 Spec 1.30 7.75 0.01440 2 10
3 B1 Spec 1.30 7.75 0.01440 2 30
4 N2 Spec 1.30 7.75 0.00440 20 10
5 B2 Spec 1.30 7.75 0.00440 20 30

In this benchmark, only one type of wave condition is tested for the diffraction effects, the N1
one. But B1, B2, N1 and N2 are tested to evaluate the refraction and the shoaling.

21.3 Test-case description

The case is about spectral wave propagation on a submerged elliptical mound. The study will
compare the spectral significant wave heights given by T OMAWAC simulations with measure-
ments from [35]’s experiments.

21.3.1 Geometry of the domain and bathymetry

Bassin dimension: 27.43 m * 18.3 m
depth constant outside the shoal: 0.4572 m
Elliptical shoal major radius: 3.96 m
minor radius: 3.05 m
maximum height: 0.305 m
coordinates of the center: (6.10,13.72)

The elliptical shoal is defined by:

x y
a = ( 3.96 ) + ( 3.05 ) if a ≥ 1, Z f = −0.4572
√
if a < 1, Z f = −0.9144 + 0.762 1 − 0.64a

21.3.2 Meshes
Spatial discretization
Three meshes are used for the simulation:
• finer: Element size at the shoal 0.2 m ∆x/λ = 0.09
• medium: Element size at the shoal 0.4 m ∆x/λ = 0.18
• coarser: Element size at the shoal 0.8 m ∆x/λ = 0.35
Where λ is the wavelength.

Spectro-angular discretization
• 22 frequencies (0.35 - 2.85 Hz)
• logarithmic scale (∆ f / f = 0.1)
• 36 directions
 86 Chapter 21. Shoal: Submerged elliptical mound

Figure 21.2: The different meshes used.

21.3.3 Initial and Boundary conditions

The domain is initially at rest. The incident wave conditions are the following:
• the incident wave conditions are imposed at the boundary representative of the wave
generator. All the others are absorbing boundaries.

• Frequency spectrum: TMA shallow water spectrum

S( f , θ ) = E( f ) ∗ D(θ )

αg2 −5 fp f−f
exp(−0.5( σ f pp )2 )
E( f ) = 4
f exp(−1.25( )4 ) ∗ γ ∗ φ ( f , d)
(2π) f
• Directional spreading function: gaussian type

1 (θ − θm )2
D(θ ) = √ exp[− ]
2πσm 2σm2

Where:
• α: Phillips constant

• f p : peak frequency ( f p = 0.769 Hz)

 21.4 Numerical parameters 87

• γ: peak factor

0.07 if f < f p
• σ=
0.09 if f > f p

• φ ( f , d) is afactor which allows the consideration of depth effects:

 0.5w2p if wd ≤ 1
φ ( f , d) = 1 − 0.5(2 − wd )2 if 1 < wd < 2
1 if 2 ≤ wd

with wd = 2π f (d/g) 1/2

• θm : mean wave direction at frequency f

• σm : directional spreading parameter

Figure 21.3: The boundary conditions.

21.4 Numerical parameters

• time steps:

1. ∆x/λ = 0.09 → ∆t = 0.005

2. ∆x/λ = 0.18 → ∆t = 0.025
3. ∆x/λ = 0.35 → ∆t = 0.1

• CPU times:
∆x/λ = 0.09 ∆x/λ = 0.18 ∆x/λ = 0.35
B1 – 1569 s 209 s
B2 – 1387 s 211 s
N2 – 1534 s 239 s
N1 4027 s 1359 s 229 s

21.5 T OMAWAC Results

At first, simulations without diffraction effects are done in order to validate the propagation and
refraction of the wave over this complex bathymetry. To compare the simulation results and
the measurements of [35], the significant wave heights have been normalized by the incident
significant wave height.
88 Chapter 21. Shoal: Submerged elliptical mound

• B1 and B2 tests (wide directional spreading):

A good matching can be seen in these tests (see Figure 21.4). After drawing the iso-
wave-height compared to the [35]’s measurements, the wave concentration zone is well
represented. The amplification factor goes over 1.3 for B1 and 1.4 for B2. As in mea-
surements, this amplification factor is a bit higher for B2 case (higher peak factor) than
for B1 case. But simulation factors are a little bit higher than those measured.

Moreover, the computations of normalized wave height along the transect 4 match quite
well with the measurements (not shown here).

Figure 21.4: Comparaison of normalized significant wave height iso-lines for the B1 (left) and
B2 (right) tests with the measurements from Vincent and Briggs (coarser mesh).

• N1 and N2 tests (narrow directional spreading):

These tests are not as good as the B ones because only one area of big wave heights is
shown while two are highlighted in measurements. Also the amplification factors are
bigger than those measured. It can be explained by the lack of representation of the
diffraction effects. But again, the results along the transect 4 show a coherence of the
simulation and the measurements. (see Figure 21.5 for N2, and Figure 21.7 for N1).
21.5 TOMAWAC Results 89

Figure 21.5: Comparaison between normalised iso-line T OMAWAC results and Vincent mea-
surements for the case N2 (coarser mesh).

Subsequently, the computation will include diffraction effects. The diffraction coefficient Kd is
defined by the ratio of the spectral significant wave height to the spectral significant incident
wave height, Kd = HHm0_inc
m0
.

Parametrical tests have shown the influence of two non-dimensional parameters: the ratio of
the mesh size over the wavelength (∆x/λ ) and the current number. The courant number Cr is:
C ( f )∆t
Cr = g ∆xp .

where Cg ( f p ) is the group velocity associated to the sea-state peak frequency, ∆t the simulation
time step and ∆x the size of the smallest element of the mesh. The diffraction coefficients
obtained during the simulations (changing Cr and ∆x/λ ) are compared with measurements
[35] and the iso-Kd curves resulting from T OMAWAC simulations are superposed with the ones
measured by Vincent and Briggs. (see Figure 21.8).
90 Chapter 21. Shoal: Submerged elliptical mound

Figure 21.6: Comparaison between T OMAWAC results and Vincent measurements: measured
and simulated values of the diffraction coefficient along the transect 4 (see Table 21.1) of the
model.

Figure 21.7: Comparaison between T OMAWAC results and Vincent measurements: measured
and simulated iso-Kd curves over and behind the mound for the case N1.
21.5 TOMAWAC Results 91

As one can see in these two graphs (Figures 21.6 & 21.7), the diffraction effects over the shoal
are not well simulated by T OMAWAC due to a large build-up of energy. Indeed the larger Cr
is and the finer the mesh is, the larger the numerical and the energy build-up are. Moreover
with the coarser mesh (∆x/λ = 0.35) T OMAWAC is not able to capture the diffraction effects
generated by the submerged shoal. In the Figure 21.7, the simulation’s curves are too flattened
and the curves that have the shape closest to the measurements are the No diffraction ones. Also
in the iso-Kd graph, the curves do not superimpose well when diffraction is taken into account,
but for the case with only propagation and refraction the correspondance is correct. Indeed, we
can see that the iso-line shapes are closest to the measured ones.

Mainly, the build-up and the numerical noise are generated when a high resolution of the mesh
is used and they are due to the meshfree algorithm (used to compute second derivatives during
simulation). In order to improve the simulation of the diffraction effects, a spatial filter is built
to limit the energy build-up. The effect of this filter can be seen on the diffraction coefficient Kd
(see Figure 21.8). The noise and the energy build-up effects are reduced but it does not improve
the quality of the results significantly.

Figure 21.8: Comparaison between T OMAWAC results and Vincent measurements: measured
and simulated values using a filter of the diffraction coefficient along the transect 4 of the model.
 92 Chapter 21. Shoal: Submerged elliptical mound

Figure 21.9: Wave height hm0 for the last calculation.

21.6 Conclusion
This benchmark test allows to compare simulations from T OMAWAC and experimental results.
It follows that T OMAWAC provides a correct simulation of the propagation, shoaling and the
refraction but the diffraction effects are limited and not really well represented. In order to
improve the results, a smoothing filter can be applied on the spatial domain.
 22. Turning wind

22.1 Purpose
This test case shows how T OMAWAC calculate the spectrum when there is a wind that is turning.
It illustrates the phenomena of white capping and quadruplet interactions.

22.2 Description of the problem

The domain is homogeneous, The depth is infinite, the wind given at 10m is homogeneous prop-
agation step is inhibited and no boundary condition are given. So the domain of computation
can be reduced to a few points as the solution is homogeneous and finally we focus on only one
of them. The domain is finally very simple (see Figure 22.1)

22.3 Physical parameters

The modulus of the wind is constant, equal to 20 m/s. Initially the direction is set to 90 degrees,
(which means that UY is null). This is maintained till the peak frequency reaches the double of
its equilibrium value (peak frequency of Pierson-Moskowitz). This occurs after 28800 s (8h),
at this time we change the direction of wind for 30 degrees (a rotation of 60 degrees). All the
values of the wind are read in a file called wind.slf .

22.4 Geometry and Mesh

The domain is a square of 2 km. The mesh is very simple as there are 5 nodes and 4 triangles.
That also means that this case can not be run in parallel.

22.5 Initial and Boundary Conditions

Initial conditions are imposed homogeneous on the domain with the option 4, with a Phillips
constant of 0.024 an initial peak frequency of 0.3 Hz and an initial peak factor of 1. The initial
directionnal spread is set to one and the initial main direction to 90 degrees (like the wind).
Those value has been set to be identical to the Vledder study [36]
Nothing is imposed at the boundaries.
 94 Chapter 22. Turning wind

Figure 22.1: Mesh of the domain.

22.6 Numerical parameters

Time duration is set to 115200 s (32 h), time step is equal to 900 s, the spectro-angular mesh
has 12 angles and 26 frequences spread on a geometric progression common ratio 1.1 with a
minimum of 0.04177248 Hz.

22.7 Results
We present here the results only for the rotation of 60 degrees as it is the effective rotation that is
done in the case. Initially there were 2 more rotations one of 90 degrees and one of 30 degrees.
The results for those rotations are described in [4]. Celerity of wind is 20 m/s. During 28790s,
the direction is 90 degrees, then it turns to 30 degrees.
At the moment when the direction changes the established swell is going to interact with the
wave induced by the new direction of wind. Three phenomenas occur in that process.

• Wind contribution to the energy that will raise the new wave,

• Attenuation of the swell part of the spectrum,

• Non linear interactions between swell and wind sea.

The results of T OMAWAC are compared to simulations made by two differents code. EXACT-
NL and WAM-cycle 3 (see [36]). Figure 22.2 shows that there are good agreements of T OMAWAC
results compared to other simulations. One can denote that the spectrum tail factor can be sen-
sitive. During the first 4 hours a factor 4 is better but after that time a factor 5 is better.
Let us remark that those are old results and a new simulation with new linear terms might give
better results.
22.7 Results 95

Figure 22.2: Comparison of direction of wave with time after a direction change of 60 degree
at t=0 s. Simulations are made for differents spectrum tail factor and different initial spectrum.

Figure 22.3: Wave height hm0 when winds changes.

 23. Whirl current

23.1 Purpose
This test case should make it possible to check that the effects of refraction by the current are
properly taken into account by T OMAWAC 7.0. The distorsion of a wave spectrum as it reaches
a whirl current zone like those observed along the Norwegian coasts [27]

23.2 Description of the problem

This test case is derived from the Mathiesen’s results [27]. Equivalent results can be found in
[33] and [19]. The spatial domain is a 80 km-sided square (see Figure 23.1). A circular, origin-
centered whirl current is considered ; it is modelled as follows. Is tangential velocity is zero at
the origin and linearly increases up to r = r1 as per:

r
u(r) = ul for r ≤ rl ≤ r0
rl

Then, that velocity follows a Gaussian profile:

2 !
r − r0

u(r) = umax exp − for r > rl
br0

Both continuity and derivability for u lead to:

√
rl 1 + 1 − 2b2
=
r0 2

 !
rl − r0 2

ul
= exp −
umax br0

The following values were adopted in the computations: umax = 1m/s, r0 = 10km and b = 0.3.
 23.3 Initial and Boundary Conditions 97

Figure 23.1: Whirl current.

The model as developped by Mathiesen is a refraction model that computes the orthogonal
waves through a conventional ray method. The results it provides for the 0.1 Hz frequency are
displayed on Figure 23.2. Many orthogonal crossings can be observed.

23.3 Initial and Boundary Conditions

That method of rays can be implemented either forwards (as on Figure 23.2), or backwards, then
knowing the arrival point of an orthogonal, it provides the starting point. A whole wave spec-
trum can be reconstructed by multiplying the computations of this kind. The incident spectrum
as prescribed by [27] is as follows:

S( f , θ ) = S( f )D( f , θ )

In that expression S( f ) is a frequency spectrum of the classical Jonswap type, where the peak
frequency f p is set to 0.1 Hz. D( f , θ ) is a Gaussian distribution:

 2

exp − (θ −θ m)
2σ02
D( f , θ ) = √
2πσ0
 98 Chapter 23. Whirl current

Figure 23.2: Computation of rays.

θm is the mean direction of the incident waves (here θm = 0) and σ0 is given by the following
relationship.

 −2.03
f
σ0 = σ0p fp if f < f p
 1.04
f
σ0 = σ0p fp if f ≥ f p

where σ0p , the directional spread at the peak, is 25 in our case. This spectrum is imposed on
the East South and West boundaries while the North boundary is free.

The initial spectrum is null.

23.4 Geometry and Mesh

The mesh is made of 1876 nodes and 3590 triangles and is shown Figure 23.3
 23.5 Numerical parameters 99

Figure 23.3: Mesh of the domain.

23.5 Numerical parameters

Time duration is 60000 s, time step is equal to 1200 s, the spectro-angular mesh has 48 angles
and 25 frequences spread on a geometric progression common ratio 1.1 with a minimum of
0.04177248.

23.6 Results

On figure 23.4, we present the amplification factor of the heigth due to the current. We can
notice two different zones at the center of the domain where there is a strong modification of
the heigth. One zone where the heigth is raising (till 35%) when the swell is opposed to the
current, and one zone where the heigth is decreasing (till 20%) where swell and current are in
the same direction. On the other parts of the domain, modifications are less than 5%. Those
results are coherent with the ray calculus presented on Figure 23.2 since on the two zones we
denote orthogonal crossing.
In his paper, Mathiesen [27] defines 8 points shown on Figure 23.4. On these points he gives
the energy angular spread at 0.1 Hz (frequency peak of the incident spectrum). We compare
this spread to the one obtained by T OMAWAC on Figure 23.5. We can denote that the results are
very closed to Mathiesen results especially on points 4 and 7. Notice that Mathiesen took an
angular discretisation of 2.5 when our is of 7.5.
100 Chapter 23. Whirl current

Figure 23.4: Significative heigth and positions of the measurement points.

23.6 Results 101

Figure 23.5: Comparison of the angular spread at 0.1 Hz between T OMAWAC and Mathiesen
 102 Chapter 23. Whirl current

Figure 23.6: significative heigth of the last calculation

23.7 Conclusion
This test case showed on a realistic case of refraction of current that T OMAWAC gives suitable
results compared to fine results.
 [1] J.H.G.M. ALVES and M.L. BANNER. Revisiting the pierson-moskowitz asymptotic lim-
its for fully developed wind waves,. Journal of Physical Oceanography,, 33:1301 – 1323,
2003.
[2] V. Bacchi, E. Gagnaire-Renou, N. Durand, and M. Benoit. Wave energy dissipation by
vegetation in tomawac. In 21st Telemac- Mascaret User Club Grenoble. 21st Telemac-
Mascaret User Club Grenoble, October 2014.
[3] J.A. BATTJES and J.P.F.M. JANSSEN. Energy loss and set-up due to breaking of random
waves. In Proc. 16th Int. Conf. Coastal Eng., pages 569–587., 1978.
[4] M. BENOIT. Logiciel tomawac de modélisation des états de mer en éléments finis. dossier
de validation de la version 1.0. Technical Report HE-42/96/010/B, EDF-R&D-LNH, 1996.
Fiche cas-test : Evolution du spectre directionnel de houle dans un vent tournant.
[5] M. BENOIT. Logiciel tomawac de modélisation des états de mer en éléments finis. dossier
de validation de la version 1.0. Technical Report HE-42/96/010/B, EDF-R&D-LNH, 1996.
Fiche cas-test : Simulation des tempêtes observées en Manche début 1990.
[6] E. BOUWS and G.J. KOMEN. On the balance between growth and dissipation in an
extreme depth-limited wind-sea in the southern north-sea. J. Phys. Oceanogr., 13:1653–
1658, 1983.
[7] WA Breugem, E Fonias, L Wang, and A Bolle. Tel2tom: coupling telemac2d and
tomawac on arbitrary meshes. In XXVIthTELEMAC-MASCARET User Conference.
XXVIth TELEMAC-MASCARET User Conference, October 2019.
[8] L CAVALERI and P. MALANOTTE-RIZZOLI. Wind wave prediction in shallow water :
theory and applications. J. Geophys. Res., 86 (C5):10,961–10,975, 1981.
[9] CERC. Shore Protection Manual. USACE, Vicksburg, MS, 1977.
[10] CERC. Shore Protection Manual Volume 1. USACE, Vicksburg, MS, 1984.
[11] R.G Dean. Equilibrium beach profiles: characteristics and applications. Journal of Coastal
Research, 7:53–84, 1991.
[12] A.J. Van der WESTHUYSEN. Spectral modeling of wave dissipation on negative current
gradients. Coastal Eng, 58:17–30, 2012.
[13] A.J. Van der WESTHUYSEN, M. ZIJLEMA, and J.A. BATTJES. Nonlinear saturation-
based whitecapping dissipation in swan for deep and shallow water. Coastal Eng., 54:
151–170, 2007.
104 Bibliography

[14] E. Gagnaire-Renou, M. Benoit, and Ph. Forget. Ocean wave spectrum properties as de-
rived from quasi-exact computations of nonlinear wave-wave interactions. J. Geophys.
Res. C (Oceans), 115, C12, C12058, 2010. doi: 10.1029/2009JC005665.
[15] Y. Goda and Y. Suzuki. Computation of refraction and diffraction of sea waves with
mitsuyasu’s directional spectrum. Port and Harbour Res. Inst., pages 1–45, 1975.
[16] K. HASSELMANN, T.P. BARNETT, E. BOUWS, H. CARLSON, D.E. CARTWRIGHT,
K. ENKE, J.A. EWING, H. GIENAPP, D.E. HASSELMANN, P. KRUSEMAN,
A. MEERBURG, P. MULLER, D.J. OLBERS, K. RICHTER, W. SELL, and
H. WALDEN. Measurements of wind-wave growth and swell decay during the joint north
sea wave project (jonswap). Deutschen Hydrographischen Zeitschrift, 8 N 12, 1973.
[17] S. HASSELMANN and K. HASSELMANN. Computations and parameterizations of the
nonlinear energy transfer in gravity-wave spectrum. part i: a new method for efficient
computations of the exact nonlinear transfer integral. J. Phys. Oceanogr., 15:1369–1377,
1985.
[18] TS HEDGES, K ANASTASIOU, and D GABRIEL. Interaction of random waves and
currents. Waterway Port Coastal and Ocean Engineering, 111:275–288, 1985.
[19] K. P. Hubbert and J. Wolf. Numerical investigation of depth and current refraction of
waves. Journal of Geophysical Research, 96:2737–2748, 1991.
[20] P.A.E.M. JANSSEN. Wave-induced stress and the drag of air flow over sea waves. J.
Phys. Oceanogr., 19:745–754, 1989.
[21] P.A.E.M. JANSSEN. Quasi-linear theory of wind-wave generation applied to wave fore-
casting. J. Phys. Oceanogr., 21:1631–1642, 1991.
[22] K.K. KAHMA and C J. CALKOEN. Reconciling discrepancies in the observed growth of
wind-generated waves. J. Phys. Oceanogr., 22:1389–1405, 1992.
[23] G.J. KOMEN, S. HASSELMANN, and K. HASSELMANN. On the existence of a fully
developed wind-sea spectrum. J. Phys. Oceanogr., 14:1271–1285, 1984.
[24] R.J. LAI, S.R. LONG, and N.E. HUANG. Laboratory studies of wave-current interaction :
kinematics of the strong interaction. Journal of Geophysical Research, 94:16201–16214,
1989.
[25] I.V. LAVRENOV. Effect of wind wave parameter fluctuation on the nonlinear spectrum
evolution. J. Phys. Oceanogr., 31:861–873, 2001.
[26] F. Marcos, M. Benoit, and P. Thellier. Cowadis software for finite element wave propaga-
tion in coastal areas : Validation document of release 1.0. Technical report, EDF R&D,
1998.
[27] M. MATHIESEN. Wave refraction by a current whirl. Journal of Geophysical Research,
92:3905–3912, 1987.
[28] F. M. Mendez and I. J. Losada. An empirical model to estimate the propagation of random
breaking and nonbreaking waves over vegetation fields. Coast. Eng., 51:103–118, 2004.
[29] H. MITSUYASU, F. TASAI, T. SUHARA, S. MIZUNO, M. OHKUSU, T. HONDA, and
K. RIKIISHI. Observations of the directional spectrum of ocean wavesusing a cloverleaf
buoy. J. Phys. Oceanogr., 5:750–760, 1975.
Bibliography 105

[30] R.L. SNYDER., F.W. DOBSON, J.A. ELLIOT, and R.B. LONG. Array measurements of
atmospheric pressure fluctuations above surface gravity waves. J. Fluid Mech., 102:1–59,
1981.

[31] Maria João Teles. Wave-current modelling at local and regional scales. PhD thesis, IST
Técnico Lisboa, 2013. Phd supervised by António Alberto Pires Silva and Michel Benoit.

[32] E.B. THORNTON and R.T. GUZA. Transformation of wave height distribution. J. Geo-
phys. Res., 88(10):5925–5938, 1983.

[33] H.L. TOLMAN. A third-generation model for wind waves on slowly varying unsteady
and inhomogeneous depths and currents. J. Phys. Oceanogr., 21:782–797, 1991.

[34] H.L. TOLMAN. Effects of numerics on the physics in a third-generation wind-wave

model. J. Phys. Oceanogr., 22:1095–1111, 1992.

[35] C.L. VINCENT and M.J. BRIGGS. Refraction-diffraction of irregular waves over a
mound. Journal os Waterway, Port,Coastal and Ocean Engineering,, 115 N2:269–284,
1989.

[36] G.Ph. VAN VLEDDER. Directional response of wind waves to turning winds. PhD thesis,
Delft University of Technology, 1990.

[37] B W. WILSON. Numerical prediction of ocean waves in the north atlantic for december
1959. Deutsche Hydrographische Zeitschrift, 18 (3):114–130, 1965.

[38] L. YAN. An improved wind input source term for third generation ocean wave modelling.
Tech. Rep., 8, Royal Dutch Meteor. Inst, 1987.