Continental Shelf Research 96 (2015) 1–15

Contents lists available at ScienceDirect

Continental Shelf Research

journal homepage: www.elsevier.com/locate/csr

Research papers

The contribution of short-waves in storm surges: Two case studies

in the Bay of Biscay
Xavier Bertin a,n, Kai Li a, Aron Roland b, Jean-Raymond Bidlot c
a
UMR 7266 LIENSs CNRS-Université de La Rochelle, Institut du Littoral et de l’Environnement, 2 rue Olympe de Gouges, 17000 La Rochelle, France
b
BGS IT&E GmbH, D-64297 Darmstadt, Pfungstaedter Strasse 37, Germany
C
European Centre for Medium Range Weather Forecasts, Reading, United Kingdom

art ic l e i nf o a b s t r a c t

Article history: This study investigates the contributions of short waves in storm surges through the hindcast of two
Received 31 January 2014 storms that hit the central part of the Bay of Biscay recently. Despite displaying comparable wind speed
Received in revised form and directions in the study area, these two storms induced different storm surges and sea states. Xynthia
17 December 2014
(27–28th of February 2010) was characterized by large (up to 7 m significant wave height Hs) and short-
Accepted 6 January 2015
period waves and induced an exceptional storm surge, locally larger than 1.6 m. The second storm,
Available online 14 January 2015
Joachim (15–16th of December 2011), was characterized by very large (up to Hs 4 10 m) and long-period
Keywords: waves but only induced a storm surge almost two times lower. To investigate these differences, a new
Storm surge unstructured grid and fully coupled modeling system is applied, with a spatial resolution fine-enough to
Unstructured grid
adequately represent the surf zones over most of the study area (25 m). The analysis of the modeling
Surface stress
results and the available field observations reveals firstly that the exceptional surge during Xynthia
Wave setup
Friction velocity originated from young and steep waves, enhancing surface stress. This particular sea-state is explained
Xynthia by the abnormal track of Xynthia, which restricted the fetch to a few hundred km. The wave radiation
Joachim stress gradient locally induced setup larger than 0.4 m along the coastlines fully exposed to ocean waves,
while wave setup in the range 0.1–0.2 m was also shown to develop regionally and to propagate in
sheltered harbors. Comparatively, wave-enhanced bottom stress appears to be a second-order process
and has a more limited impact on storm surges.
& 2015 Elsevier Ltd. All rights reserved.

1. Introduction Haiyan in the Philippines (2013) remind us of the necessity of

being able to predict storm surges accurately, although the phy-
Tropical cyclones and extra-tropical storms making landfall in sical processes controlling these phenomena remain only partly
coastal zones are among the most costly natural disasters (Smith, understood.
1996; Nicholls et al., 2007). Historically, the main part of the ma- The effects of atmospheric pressure gradients and winds on
terial and human losses in coastal zones is associated with coastal sea-level were recognized early (e.g. Doodson, 1924) and were
flooding rather than direct wind effects (Zhang et al., 2008). integrated in pioneer modeling approaches (e.g. Jelesnianski,
Coastal flooding predominantly occurs in low-lying zones under 1965). Following these early quantitative approaches, many stu-
dies have shown that the wind-induced surface stress was the
the concomitance of a large storm surge and a high spring tide,
dominant process at coastal zones bordered by large and shallow
although the importance of this concomitance also depends on the
shelves (Flather, 2001; Rego and Li, 2010; Kennedy et al., 2012).
ratio between the storm surge and the local tidal range. For in-
Tide–surge interactions were also shown early to be significant at
stance, the more than 8 m storm surge induced by hurricane Ka- some locations (e.g. Proudman, 1957; Rego and Li, 2010; Idier et al.,
trina in 2005 (Blake, 2007) would have induced a major flooding 2012). By contrast, the contribution of short-waves in storm surges
in the microtidal coastlines of Louisiana, whatever the tidal phase has only been investigated more recently. Thus, for a long time, it
might have been. The major catastrophes that occurred over the has been common practice to compute the wind surface stress
last decade, such as Katrina in the Gulf of Mexico (2005), Nargis in based on bulk formula (Eq. (1)):
the Bay of Bengal (2008), Sandy in the New York area (2012) and 2
τs = ρa Cd U10 (1)

n
Corresponding author. Fax: þ 33 546458274. where ρa is the air density, U10 is the 10 m wind speed and Cd is a
E-mail address: [email protected] (X. Bertin). drag coefficient corresponding to the sea roughness that increases

http://dx.doi.org/10.1016/j.csr.2015.01.005
0278-4343/& 2015 Elsevier Ltd. All rights reserved.
2 X. Bertin et al. / Continental Shelf Research 96 (2015) 1–15

linearly with the wind speed for low to moderate winds (e.g. contrasting case studies to investigate the contribution of short
Smith and Banke, 1975; Pond and Pickard, 1998). However, based waves in storm surges. The section following this introduction
on the pioneer work of Charnock (1955), Stewart (1974) proposed describes the data and the fully coupled modeling system used in
that the sea roughness should also depend on the wave age for a this study. Section 3 describes the studied area and both storms.
given wind speed. The dependence of the surface stress on the sea The next section presents the modeling results in terms of atmo-
state was then corroborated in many studies (Donelan et al., 1993; spheric forcing, sea states and storm surges. Section 5 discusses
Mastenbroek et al., 1993; Brown and Wolf, 2009; Sheng et al., the contribution of the three main wave-induced processes in
2010; Olabarrieta et al., 2012). More recently, field measurements storm surges: wave-enhanced surface and bottom stress and the
under extreme winds showed that the sea roughness could reach a gradients of radiation stress. Finally, the main findings of this
maximum or even decrease due to wave-induced streaks of foam study are summarized and some perspectives are given in the
and sprays for winds larger than 35–40 m/s (Powell et al., 2003; conclusion.
Takagaki et al., 2012; Holthuijsen et al., 2012). In shallow water,
orbital motions associated with short-wave propagation can also
enhance bottom stress, thereby usually reducing storm surges (e.g. 2. Methods and data
Nicolle et al., 2009). Nevertheless, the impact of this phenomenon
on the surge peak is often found to be limited (Xie et al., 2003) 2.1. The storm surge modeling system
while storm surge predictions are not always clearly improved
when accounting for this process (e.g., Jones and Davies, 1998). In 2.1.1. General overview
the nearshore, wave dissipation induces gradients of radiation In this study, we applied the numerical modeling system Semi-
stress (Longuet-Higgins and Stewart, 1964) that drive a setup implicit Eulerian–Lagrangian Finite Element (SELFE; Zhang and
easily reaching several tens of centimeter during storms. However, Baptista, 2008; Zhang et al., 2011), which now includes modules to
the proper representation of this phenomenon requires employing simulate water quality (Rodrigues et al., 2011), oil spills (Azevedo
a very fine spatial resolution (e.g. of the order of 10 m), which et al., 2014) and sediment transport (Pinto et al., 2012). Recently, a
poses a serious computational challenge when simulating storm full coupling was undertaken with the spectral wave model Wind
surges at regional scale. This difficulty probably explains why only Wave Model II (hereafter WWMII, Roland et al., 2012). The two
a limited number of studies have successfully accounted for wave codes share the same unstructured grid and the same domain
setup at regional scale (e.g. Dietrich et al., 2010). decomposition, which makes this modeling system very compu-
This study investigates the contribution of short waves in storm tationally efficient and allows for massive parallel techniques.
surges based on the hindcast of two extra-tropical storms that More details regarding coupling can be found in Roland et al.
recently hit the central part of the Bay of Biscay (France): Xynthia (2012). The unstructured grid used in this study employs 201,701
on the 27–28th of February 2010 and Joachim on the 15–16th of nodes (385,980 elements) and its resolution ranges from 30,000 m
December 2011. This study builds on a preliminary hindcast of the in the deep Ocean and far from the study area to 25 m along the
storm surge associated with Xynthia, which used an offline cou- shoreline of the study area (Fig. 1). Such a fine resolution together
pling between a circulation model and a spectral wave model and with the coupling strategy allow for a proper representation of
a coarse spatial resolution that prevented the representation of nearshore wave-induced processes, which constitutes a major
nearshore wave-induced processes (Bertin et al., 2012). Although improvement compared to the preliminary hindcast of Xynthia
both storms displayed comparable wind speed and directions over described in Bertin et al. (2012). The computational grid is boun-
the study area, they induced different storm surges and sea states. ded by the shoreline, which means that the coastal areas flooded
The main purpose of this paper is to take advantage of these during Xynthia are not represented. Bertin et al. (2014) conducted

Fig. 1. Unstructured grid used in this study, showing a resolution ranging from 30,000 to 25 m.
 X. Bertin et al. / Continental Shelf Research 96 (2015) 1–15 3

a specific study on the flooding associated with Xynthia and the parameterization of these source terms can be found in Bidlot
showed that the massive flooding that occurred impacted water et al. (2007b) and Bidlot (2012). An important consequence for this
levels in estuaries significantly. However, these authors also study is that this parameterization is very sensitive to the wave
showed that this effect was negligible at the tide gauges used in age C/Un and to the roughness length Z0. A high-energy level in the
the present study (i.e. o 0.05 m). Finally, a time step of 60 s was high-frequency part of the spectrum will lead to larger values for
selected for both models. Z0 and Un, which will result in a positive feedback and reinforce
the energy levels.
2.1.2. The spectral wave model In this version of the code, we have adopted to our best
WWMII solves the wave action equation (Eq. (2)) on un- knowledge the parameterization described in Bidlot et al. (2007b).
structured grid: However, this does not correspond exactly to the latest develop-
ments carried out at the European Center for Medium range
∂N ∂ (Cgx + U) N ∂ (Cgy + U) N ∂ (CσN) ∂ (CθN) Stot Weather Forecast (hereafter ECMWF), which include for example
+ + + + =
∂t ∂x ∂y ∂σ ∂θ σ (2) modifications for shallow waters. Another limitation of our ap-
proach is that we employed the total air-side stress to force our
In this equation, N is the wave action, Cgx and Cgy are the x and y- circulation model while it should be corrected from the amount of
components of the wave group velocity, U and V are the two momentum absorbed and released by the wave field (Janssen et al.
horizontal components of the current velocity, s and θ are the 2004, 2013). The impact of this limitation as well as the latest
wave relative angular frequency and the wave direction and S
development carried at ECMWF for shallow waters will have to be
corresponds to the sum of the source terms. Source terms include
investigated in future studies.
non-linear interactions, wind growth and dissipation by white-
A four-step fractional method is employed, according to Ya-
capping, bottom friction and wave breaking, which are computed
nenko (1971). First, advection in geographic space is solved using
according to the approaches JONSWAP (Hasselmann et al., 1973)
the explicit N-Scheme (e.g. Abgrall, 2006). Advection in spectral
and Battjes and Janssen (1978), respectively. Nonlinear 4-wave
space is then solved using the Ultimate Quickest finite difference
interactions are computed using the Discrete Interaction Approx-
scheme of Leonard (1991) as done in WaveWatchIII (hereafter
imation of Hasselmann et al (1985), and the Lumped Lumped Triad
Approximation of Eldeberky (1996) in shallow water. Among these WWIII; Tolman, 2009). Finally, the source terms S are integrated in
source terms, the wind input S(s,θ)in is of key importance in this a similar manner as in WWIII. WWMII is forced by wind fields
study since the coupling with the circulation model is made originating from series of 12 h forecasts from the operational high
through the friction velocity computed from the wave model. resolution model of ECMWF (0.125°/1 h) over the whole domain.
The approach used here is based on the quasi-linear theory of Along the open boundary, WWMII is forced with time-series of
wave–atmosphere interactions and follows the pioneering work of spectra computed from a regional application of WWIII for the
Phillips (1957) and Miles (1957), later improved by Janssen (1989, North Atlantic Ocean described in Dodet et al. (2010) and Bertin
1991). This kind of approach is the one that is used in many et al. (2013). Fields of currents and water levels are provided by
operational wave forecasting systems worldwide (Bidlot et al., SELFE at each hydrodynamic time step (60 s) while WWMII pro-
2007a; Tolman, 2009) and on top of which further improvement vides SELFE with fields of wave radiation stress, friction velocities
could be made, e.g. Ardhuin et al. (2010). The formulation of the as well as wave directions and orbital bottom velocities used to
wind input source terms in the framework of the Wave Action compute bottom stress.
Equation is given as follows:
2.1.3. The circulation model
ρa ⎛ U ⁎ ⎞2
S(σ, θ),in = σ eZZ 4 ⎜ ⎟ max (cos (θ − φ), 0)2N(σ, θ) The horizontal circulation is computed with SELFE (Zhang and
ρw κ ² ⎝C ⎠ (3) Baptista, 2008), which solves the full Navier Stokes equation over
unstructured grids and was designed to address a large range of
where s is the relative wave frequency ρa and ρw the air and water
spatio-temporal scales. In this study, SELFE is used in 2DH baro-
densities, respectively. κ is the von Karman constant and βmax is a
tropic mode and the resolved equations are the following (Eqs.
non-dimensional growth parameter set to 1.22. C is the phase
(7)–(9)):
velocity, φ the wind direction at 10 m height and θ the discrete
wave direction of the considered wave packet. Z is a function il- ∂ζ → ζ

lustrating that the growth rate depends on the roughness length ∂t

+ ∇. ∫−h →u dz = 0 (7)
Z0, which inherently depends on the sea state:
⎛ gZ 0 ⎞ DU ∂ψ^ 1 ∂P A ∂ζ τSx − τBx
κ = fV + αg − −g +
Z = ln ⎜ ⎟+ Dt ∂x ρw ∂x ∂x ρw (ζ + h)
⎝ C ² ⎠ cos (θ − φ)((U ⁎/C) + Z α) (4)
1 ⎛ ∂S xx ∂S xy ⎞
where zα is a wave age tuning parameter set to 0.011. The − ⎜ + ⎟
ρw (ζ + h) ⎝ ∂x ∂y ⎠ (8)
roughness Z0 is defined as
U⁎ ⎛ Z obs ⎞ DV ∂ψ^ 1 ∂P A ∂ζ τSy − τBy
U10 = ln ⎜ ⎟ = − fU + αg − −g +
κ ⎝ Z0 ⎠ (5) Dt ∂y ρw ∂y ∂y ρw (ζ + h)
1 ⎛ ∂S yy ∂S xy ⎞
α0 U ⁎2 − ⎜ + ⎟
Z0 = ρw (ζ + h) ⎝ ∂y ∂x ⎠ (9)
1 − (τ w / τ ) (6)
where ζ is the free surface elevation, h is the bathymetry, ρw is the
where α0 is a minimum value for the Charnock coefficient, Zobs is water density, g is the acceleration due to gravity, f is the Coriolis
the height at which the wind is taken (10 m), τ = U ⁎² and τw is the parameter, PA is the sea-level atmospheric pressure, and α and ψ^
wave supported stress, an integral function of the wind input are the effective earth-elasticity factor and the earth tidal poten-
source term.. Therefore, Eqs. (5) and (6) provide an implicit tial, respectively. τB is the bed shear stress computed using a
functional dependence of U ⁎, U10 and τw/ τ. Further details on Manning friction law (Eq. (10)):
4 X. Bertin et al. / Continental Shelf Research 96 (2015) 1–15

→ g⋅n² →2 2.2. Wave and water level observations

τB = ρw⋅ ⋅U
3
H (10)
Water levels were collected at three tide gauges available in the
where H is the total water depth and n is the Manning coefficient studied area for both storms (Fig. 2) through the REFMAR portal
set to 0.020 after calibration of the tidal model. τb can also depend (http://refmar.shom.fr/). These data consist of 10-min time series
on short waves following the approach of Soulsby (1997, Eq. (11)): over 2010–2011 and were first processed through harmonic ana-
lysis using T_Tide (Pawlowicz et al., 2002). A harmonic synthesis
⎡ ⎛ τ→ ⎞ ⎤3.2
τB = τc ⎢1 + 1.2 ⎜⎜ → → ⎟⎟ ⎥
→ → w was then performed, using the 67 main astronomic constituents.
⎢⎣ ⎝ τc + τw ⎠ ⎥⎦ (11) The constituent Sa was not considered in the tidal prediction since,
in the North-East Atlantic Ocean, it originates from a combination
where τc is the current-induced shear stress and τw is the wave- of thermo-steric and atmospheric effects (unpublished yet). The
related shear stress computed according to Swart (1974). For a storm surge was finally computed as the difference between the
consistent comparison with the case where waves are not con- observed water level and the tidal prediction obtained through
sidered in the bed shear stress, τc is computed using the Manning this procedure. Deepwater wave conditions in the Bay of Biscay
approach of (Eq. (10)). were characterized by means of a non-directional buoy operated
→
Finally, τs is the surface stress computed using a bulk formula: by Météo France and the UK met Office (Fig. 2A; Biscay), which
→2 provides hourly time-series of significant wave height (Hs), and
→
τs = ρa ⋅Cd ⋅U10 (12) mean wave period (Tm02). In addition, a Datawell buoy was de-
ployed by SHOM (French Naval Oceanographic Service) to the
where Cd is a drag coefficient computed according Pond and West of Oléron Island (Fig. 2B) during the winter 2010 and pro-
Pickard (1998). Alternatively, the surface stress can be wave- vides time series of Hs, Tm02, peak wave direction (Pdir) and Peak
dependent through the wind friction velocity Un computed by wave period (Tp). Other intermediate water depth buoy observa-
WWMII as described in Section 2.1.2:
tions were available in the Southern part of the Bay of Biscay, such
→ →2 as those of Cap ferret and Anglet, operated by CETMEF (http://
τs = ρa ⋅u⁎ (13) candhis.cetmef.developpement-durable.gouv.fr) and that of Bilbao
Over the whole domain, SELFE is forced by hourly fields of sea- operated by Puerto del Estado (http://www.puertos.es). Other
level atmospheric pressure and 10 m wind speed originating from coastal buoy observations are available in the Bay of Biscay farther
from the study area but the spatial resolution of our computational
the operational high resolution model of ECMWF. The model is
grid and the bathymetric data at our disposal prevent a consistent
forced by the astronomic tidal potential over the whole domain for
comparison with our modeling results.
the tidal constituents MM, MF, O1, K1, P1, Q1, M2, S2, N2 and K2.
Along its open boundaries, SELFE is forced by the 18 main tidal
constituents in the region (Z0, O1, K1, P1, Q1, M2, S2, N2, K2, 2N2,
3. The studied area and storms
L2, MU2, NU2, M3, M4, MS4, MN4 and M6), obtained by linear
interpolation from the regional tidal model of Pairaud et al. (2008). 3.1. Geomorphic setting
This representation of the tidal forcing together with a uniform
Manning coefficient yields good tidal predictions, with a root The study area corresponds to the Central part of the Bay of
mean square error on the total elevation ranging from 0.05 to Biscay and the geomorphology of the coast is dominated by two
0.10 m in the study area, which is slightly better than Bertin et al. big islands and several embayments and estuaries, the major being
(2012), who employed a space-uniform quadratic bottom friction. the Gironde Estuary to the South (Fig. 2). These embayments

Fig. 2. (A) Location of the study area in the Bay of Biscay and (B) detailed bathymetric map of the study area with respect to mean sea-level with the location of tide gauges
(stars), wave buoys (triangles) and meteorological stations (squares) used in this study.
 X. Bertin et al. / Continental Shelf Research 96 (2015) 1–15 5

correspond to drowned incised valley segments (Chaumillon et al., deep water significant wave height (Hs) is of the order 2.0 m and
2008) and their inner part consists of extensive intertidal mud- the mean directions and peak period are 10 s and 285°, respec-
flats, locally reaching 5 km width and bounded by dykes and tively. Winter storms can episodically produce waves of Hs larger
natural barriers. These barriers isolate extensive marshes and than 10 m. During their propagation over the broad continental
coastal plains, some of which include large areas located below the shelf, waves experience a significant attenuation of Hs and a strong
high astronomic tidal level (Breilh et al., 2013). Most of these dykes refraction, which limit wave obliquity along the coast (Bertin et al.,
were flooded and/or breached during Xynthia, which caused se- 2008). Winter-mean wave conditions in the Bay of Biscay experi-
vere human and material losses (Breilh et al., 2013; André et al., ence a significant inter-annual variability, which was shown by
2013). The continental shelf in this area is wide and ranges from Dodet et al. (2010) to be partly controlled by the North Atlantic
about 100 km to the South of the Gironde Estuary to 170 km to the Oscillation.
North of Ré Island (Fig. 2).
3.3. The studied storms
3.2. Hydrodynamic setting
The first storm, Xynthia, hit the central part of the Bay of Biscay
Tides in the study area are semi-diurnal, with small diurnal in the night of the 27th to the 28th of February 2010. Xynthia
asymmetries. Along the shelf break, the deep-water Kelvin wave originated from a low-pressure zone located in the middle of the
induces a M2 wave with amplitude of the order of 1.3 m (Le Cann, Atlantic Ocean around the latitude of the Tropic of Cancer. This
1990), which grows up to more than 1.8 m in the inner part of the depression intensified on the 27th in the morning and evolved
estuaries. towards a storm when reaching the coastlines of Portugal (Fig. 3).
Diurnal waves K1 and O1 have amplitudes of the order of Xynthia crossed the North-Western part of Spain and hit the
0.07 m and display little spatial variations. In contrast, the ampli- French border of the Bay of Biscay in the night of the 28th of
tudes of quarter diurnal waves M4, MS4 and MN4 are amplified February, where sea-level pressure (SLP) reached its minimum at
more than 10 times throughout their propagation over the central 969 hPa (Fig. 3). Southern to South-Western winds ranging from
part of the continental shelf. This interesting phenomenon was 25 to 30 m/s (hourly-mean 10 m wind speed, hereafter U10) blew
explained analytically by Le Cann (1990) and then confirmed nu- over the Southern part of the Bay of Biscay (Fig. 3). Maximum
merically by Bertin et al. (2012), who showed that the resonant instantaneous gusts reached 45 m/s at Les Baleines (Fig. 2). Such
frequency of the shelf in the central part of the Bay of Biscay was wind speed are not exceptional for the region since over the last 15
close to the frequency band of the quarter diurnal constituents. years, storms Martin (27th of December 1999) and Klaus (23th
The resulting spring tidal range at the coast exceeds 6 m in the January 2009) produced gusts over 50–55 m/s. Much more unu-
inner part of the estuaries and the associated tidal currents can sual was the track of Xynthia from SW to NE (Riviére et al., 2012).
exceed 2 m/s (Bertin et al., 2005). Wave measurements in the Bay of Biscay showed that Hs during
The Bay of Biscay is fully exposed to short-waves generated in Xynthia ranged from 6 to 7.5 m, which values are regularly en-
the North Atlantic Ocean. Due to the scarcity of wave data offshore countered during winter storms. Nevertheless, Xynthia induced an
of the study area, the long term hindcast of Dodet et al. (2010) is exceptional storm surge, estimated to more than 1.5 m in La Pallice
used to characterize the local wave climate. The annual-mean (Fig. 2), which corresponds to the largest value since the

Fig. 3. Sea-level pressure (hPa) and 10 m wind speed (m/s) during Xynthia based on the operational analysis of ECMWF. The dashed line corresponds to the track of the
minimum sea level pressure.
6 X. Bertin et al. / Continental Shelf Research 96 (2015) 1–15

Fig. 4. Sea-level pressure (hPa) and 10 m wind speed (m/s) during Joachim based on the operational analysis of ECMWF. The dashed line corresponds to the track of the
minimum sea level pressure.

installation of a permanent tide gauge at this station in 1997. This originating from the operational high resolution forecast model of
large storm surge peaked at the same time as a high spring tide ECMWF were compared against field observations available during
(5.9 m tidal range in La Pallice) and the subsequent exceptional both storms in the study area (Fig. 2). The model data consist of
water level caused the flooding of extensive low-lying coastal successive hourly forecasts from 0 and 12 UTC, with a horizontal
zones, damage of more than 2 billion Euros and 47 fatalities, 41 of resolution of 16 km. This comparison reveals firstly that SLP is very
which occurred in flooded areas (André et al., 2013). Through a well predicted for both storms, with a Root Mean Square Dis-
preliminary application of a storm surge model at the scale of the crepancy (hereafter RMSD) of the order of 1.5 hPa (Fig. 5). Note
continental shelf, Bertin et al. (2012) showed that this exceptional that it was found necessary to use hourly output from ECMWF
storm surge resulted mainly from a particular sea-state during rather than 3 hourly as was originally available for Xynthia.
Xynthia, characterized by short-period waves which enhanced the Modeled wind speed and direction were also compared against
ocean surface roughness. These authors also showed that tide– measurements at the same sites and this comparison reveals that
surge interactions were limited in the study area and impacted the for both storms, wind speed is reasonably predicted with a RMSD
storm surge by less than 0.1 m during Xynthia (Bertin et al., 2012). of the order of 2 m/s (Fig. 6). However, it should be noted that the
The second storm, Joachim, hit the Bay of Biscay in the night of height at which the measurements are taken differs from the 10 m
the 15th to the 16th of December 2011. The track of Joachim reference height of the model at two stations. Thus, wind is
roughly followed the English Channel where the minimum SLP measured at 17 m at Les Baleines and 3 m at Biscay (Fig. 2). Such
evolved around 975 hPa. South-Western winds blew over a large differences directly explain the positive bias of the model at Biscay
fetch across the Bay of Biscay with U10 of the order of 25 m/s. In and the negative bias at Les Baleines. Measurements could have
the study area, maximum gusts of the order of 40 m/s were re- been adjusted assuming a logarithmic profile, however, such a
corded at Les Baleines (Fig. 2). Joachim induced a moderate storm correction implies hypothesis on the roughness length Z0, which
surge, which ranged between 0.7 and 1.05 m in the study area, determination is one of the objectives of this study. Wind direction
which values are locally two times lower compared to Xynthia. is reasonably reproduced with a RMSD ranging from 15° to 20°.
The storm lasted more than 12 h (i.e. wind speed 420 m/s) and
therefore encompassed a full tidal cycle, but peaked at high tide 4.2. Wave predictions
for moderate tidal ranges (3.75 m tidal range in La Pallice). Wave
measurements in the Bay of Biscay revealed Hs larger than 10 m Wave predictions were compared against the measurements
during Joachim. The maximum storm surge peaked during inter- available for both storms in the Bay of Biscay and model data
mediate tidal range so that damages were limited in the study comparison is shown for Biscay and Datawell Oléron stations
area. (Figs. 7 and 8). For Xynthia, this comparison reveals that Hs is well
reproduced with a RMSD ranging from 0.31 to 0.44 m, corre-
sponding to a 11–16% error once normalized by the observations
4. Modeling results (NRMSD, Table 1). It is worth noting that Hs during Xynthia were
not exceptional since two events produced equivalent wave height
4.1. Atmospheric forcing in the same week. The drops in wave peak direction and period are
also well captured, although with a 2 h delay. Finally, the mean
Fields of sea-level pressure (SLP) and 10 m winds (U10) wave period Tm02 is very well reproduced with a RMSD of 0.7–0.9
 X. Bertin et al. / Continental Shelf Research 96 (2015) 1–15 7

Fig. 5. Modeled against observed sea-level pressure at Les Baleines, Chassiron and Biscay stations during both storms.

for all stations, which yields a 10% NRMSD (Table 1, Fig.7). Both parameters range from 10 to 11 s at the storm peak. Such short-
model and observations reveal that Tm02 is temporarily higher period waves with Hs larger than 7 m correspond to a young sea
than Tp at the beginning of Xynthia while the values for these state with steep waves.

Fig. 6. Modeled against observed wind speed and direction at Les Baleines, Chassiron and Biscay stations for both storms.
8 X. Bertin et al. / Continental Shelf Research 96 (2015) 1–15

Fig. 7. Observed against modeled wave parameters at Oléron and Biscay buoys during Xynthia.

Fig. 8. Observed against modeled wave parameters at Oléron and Biscay buoys during Joachim.
 X. Bertin et al. / Continental Shelf Research 96 (2015) 1–15 9

Table 1
Statistical errors (Root Mean Square Discrepancy, Normalized Rood Mean Square Discrepancy and Bias) for the significant wave height (Hs), the peak wave period (Tp), the
mean wave period (Tm02) and the mean wave direction (Mwd) at the available buoys for Xynthia and Joachim. Na denotes non-available data.

Name Buoy ID Storm Hs Tp Tm02 Mwd

RMSD (m) NRMSD (%) BIAS (m) RMSD (s) NRMSD (%) BIAS (s) RMSD (s) NRMSD (%) BIAS (s) RMSD (deg)

Biscay 62001 Xynthia 0.35 12  0.05 Na Na Na 0.8 12 0.2 Na

Joachim 0.88 11  0.35 Na Na Na 0.7 7  0.2 Na
Oléron Xynthia 0.44 11  0.24 1.7 14  0.4 0.9 9.3  0.5 23
Joachim Na Na Na Na Na Na Na Na Na Na
Cap Ferret 62064 Xynthia Na Na Na Na Na Na Na Na Na Na
Joachim 0.59 9 0.3 Na Na Na 0.9 7.4  0.6 Na
Anglet 62066 Xynthia 0.31 14 0.12 2 22 0.38 0.8 12 0.18 Na
Joachim 0.91 16  0.34 1.2 9.5 0.9 Na Na Na Na
Bilbao 62024 Xynthia 0.38 16 0.2 Na Na Na 0.7 11 0 19
Joachim 1.1 18  0.48 Na Na Na 0.7 7 0 8

Joachim induces more energetic waves in the Bay of Biscay, direction, the resulting storm surges in the central part of the Bay
with Hs temporarily exceeding 10 m at the storm peak. These of Biscay were up to two times larger during Xynthia than Joachim.
waves are reproduced by our modeling system with a comparable The slightly higher winds during Xynthia can only explain a small
accuracy as for Xynthia (9–18% NRMSD, Table 1, Fig. 8), although part of these differences. By contrast, the sea states were very
with a negative bias of the order of 0.4 m. The peak wave period different for both storms, with moderately large but short period
was only available at Anglet and was of the order of 14–15 s, which waves during Xynthia and very large and longer period waves
values were adequately reproduced with a RMSD of the order of during Joachim. These contrasting settings provide a unique op-
1 s. The mean wave period Tm02 was of the order of 10–11 s and portunity to investigate the contribution of short waves in storm
was reproduced with a 7% error (Table 1, Fig. 8). The MWD was surges. This section presents numerical experiments that aim at
only available at Bilbao and was very well reproduced with a better understanding the respective contributions of three wave-
RMSD of 8°. According to the model, the MWD was quite homo- induced processes: (1) wave-increased surface stress; (2) gradients
geneous in the southern part of the Bay of Biscay and was of the of wave radiation stress and (3) wave-increased bottom stress. The
order of N270°. Considering 10–20 year records at the available settings of these numerical experiments are summarized in
buoys in the Bay of Biscay, the values met during Joachim corre- Table 2.
spond to typical values for the Bay of Biscay.
5.1. Wave enhancement of surface stress
4.3. Storm surge predictions
In order to evaluate the importance of short waves on the
For both storms, modeled storm surges were computed as the magnitude of the surface stress, we compared storm surges
difference between our baseline simulation (Table 2, configuration computed with a wave-dependent formula (simulation no. 1, Ta-
1) and a tide-only simulation. Model predictions were compared ble 1) and with the bulk formula of Pond and Pickard (1998) (si-
against observations at Sables d’Olonne, La Pallice and La Cotinière mulation no. 2, Table 1). For Xynthia (Fig. 10), the wave-dependent
stations (Fig. 2). For Xynthia, a comparable accuracy can be ob- approach yields a 20–25% larger surge at Sables d’Olonne and La
served at La Pallice and La Cotinière with a RMSD of 0.12–0.15 m Cotinière and a 35% larger surge at La Pallice while this approach
while the storm surge is slightly better reproduced at Les Sables much better matches the observations. For Joachim, the two ap-
d’Olonne with a RMSD of 0.08 m (Fig. 9). In detail, the model proaches yield differences of the order of 2% at Sables d’Olonne
slightly over-estimates the surge peak at La Cotinière. For the three and La Cotinière while the storm surge is 10–15% larger at La
stations, the model displays a 0.05–0.15 m negative bias during the Pallice using a wave dependent approach (Fig. 10).
12 h period preceding the storm. For Joachim, the storm surge is The maximum difference between both simulations was also
predicted with a comparable level of accuracy although the model computed for each grid node to provide a spatial overview of these
displays a 0.05–0.10 m negative bias both at Sables d’Olonne and differences (Fig. 11). This figure confirms firstly that much larger
La Pallice (Fig. 9). The tide gauge at La Cotinière was not operating differences are obtained for Xynthia than for Joachim. For Xynthia,
during this storm. maximum differences grow from less than 0.25 m at the entrance
of the estuaries to more than 0.5 m in shallow bays while for
Joachim, differences are only significant in the inner part of the
5. The contributions of short-waves in storm surges shallow bays were they reach 0.1–0.2 m. The increase in differ-
ences from intermediate water depth (e.g. 20–30 m) to the shal-
Although both storms induced winds of comparable speed and lower part of the study area is explained by a larger contribution of
the surface stress in the overall storm surge, because the effect of
Table 2 the surface stress is inversely proportional to the water depth.
Settings of the numerical experiments performed to investigate the contribution of To better understand why differences between both ap-
wave-induced processes. proaches are much larger for Xynthia than for Joachim, time series
of energy spectra, wave age and surface stress were extracted at
Simulation Wave-dependent Gradients of wave Wave-dependent
number surface stress radiation stress bottom stress
the location of the Datawell buoy (Fig. 2B) and plotted on Fig. 12.
This comparison shows that for Xynthia, the surface stress is up to
1 Yes Yes Yes two times larger with the wave dependent approach compared to
2 No Yes Yes the bulk formula (Fig. 12). Such differences suggest that a parti-
3 Yes No Yes
4 Yes Yes No
cular sea-state occurred during Xynthia. Indeed, wave measure-
ments as well as model results showed that, a few hours before
10 X. Bertin et al. / Continental Shelf Research 96 (2015) 1–15

Fig. 9. Modeled against observed storm surges at Sables d’Olonnes, La Pallice and La Cotinière stations for storms Xynthia and Joachim.

the storm peak, Hs ranged from 4 to 6 m with Tp ranging from 6 to 7 during Xynthia, which is two times less than Joachim. Time-
8 s (Fig. 7). The mean wave period Tm02 even temporarily exceeded series of energy spectra (Fig. 12a) show that, during this period, a
Tp during this stage of the storm and the wave age dropped to lot of energy was found in the range 0.10–0.15 Hz, with spectral

Fig. 10. Observed (black circles) against computed storm surges using a wave dependent (blue) and a bulk formula (red) at Sables d’Olonne, La Pallice and La Cotinière for
storms Xynthia and Joachim. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
 X. Bertin et al. / Continental Shelf Research 96 (2015) 1–15 11

Fig. 11. Maximum difference in surface elevation between a simulation employing a wave-dependent surface stress and a surface stress computed from the bulk formula for
(A) Xynthia and (B) Joachim.

density reaching 70 m2 Hz  1. This spectral repartition strongly velocity to be very large. Hence, the particular sea state that de-
differs for Joachim, where maximum energy in this frequency veloped during Xynthia directly explains the two times larger
band reaches 30 m2 Hz  1, although the total energy was much surface stress compared to bulk formula approach. Moreover, it
larger. Such conditions during Xynthia are representative of a very can be clearly seen that during Joachim, the energy spectral re-
young sea state. The parameterization employed to compute the partition varies little throughout the storm whereas for Xynthia,
wind input source terms in the wave model (Section 2.1.2) implies there is very abrupt growth of spectral energy after a quite calm
that the high energy levels at high frequencies caused the friction period that leads to this high energy levels in the short wave range

Fig. 12. Time series of energy spectra (m2 Hz  1), wave age (dimensionless) and surface stress (Pa) for Xynthia (A) and Joachim (B).
12 X. Bertin et al. / Continental Shelf Research 96 (2015) 1–15

Fig. 13. Observed (black circles) against computed storm surges with (red) and without radiation stress (blue) at Sables d’Olonne, La Pallice and La Cotinière for storms
Xynthia and Joachim. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

of the spectra. coastlines. This problem may be enhanced during Xynthia because
The large differences in sea state between Xynthia and Joachim the very high water levels at the storm peak have caused the
are proposed to be associated with the tracks of these storms. breaking zone to end up over the dikes and dunes. The proper
Joachim followed a classical track W–E, implying a fetch of a few representation of wave breaking over such steep barriers would
thousands of km, which resulted in large and long period waves. imply employing a much finer spatial resolution (e.g. # 1 m),
By contrast, Xynthia followed an unusual SW–NE track, which which is outside the scope of this study. Secondly, several studies
restricted the fetch to a few hundreds of kilometers and resulted in suggested that for energetic conditions, wave induced-turbulence
a much younger sea state. and bottom stress related to bed return flow could increase wave
setup along the coastline significantly (e.g. Apotsos et al., 2007;
5.2. Wave-induced forces Bennis et al., 2014). Such phenomenon cannot be accounted for
properly with a 2DH approach, which would suggest that our
Wave dissipation in the nearshore induces gradients of radia- setup predictions are locally under-estimated. However, more
tion stress (Longuet-Higgins and Stewart, 1964), which drive a surprisingly, this comparison also shows the development of a
setup along the coast (Bowen et al., 1968). To investigate the regional setup inside the estuaries, this phenomenon being more
contribution of this phenomenon in the overall storm surge, the significant during Joachim. The regional bathymetry (Fig. 2) im-
difference between a simulation with radiation stress and a si- plies that wave setup locally reaching 0.2 m develops in water
mulation without radiation stress was computed for both storms depth of one or several tens of meters, which is typically outside
(Table 1, simulation 3). Surprisingly, the differences between both the breaking zones. Our interpretation is that the islands are
approaches are not negligible at the three considered tide gauges, bounded by extensive shallow shoreface where wave breaking
although being located in harbors sheltered from wave breaking. zones develop over several kilometers during storms. This setting
Maximum differences are found at La Cotinière, where they reach causes wave setup to propagate outside these large breaking
0.15 m during Xynthia and 0.2 m during Joachim. For the three zones, including the harbors where water levels are measured. A
stations, accounting for wave radiation stress improves storm similar phenomenon was already reported in small coastal lagoons
surge predictions noticeably (Fig. 13). by Bertin et al. (2009) and Dodet et al. (2013).
To provide a spatial overview of this phenomenon, the max-
imum difference between both simulations was also computed for 5.3. Wave enhancement of bottom stress
each grid node (Fig. 14). This figure reveals firstly that maximum
setup develops along the western side of the islands and shor- Several studies investigated the impact of wave-enhanced
elines exposed to the Ocean. A zoom reveals that to the North of bottom stress on storm surges (Xie et al., 2003; Nicolle et al., 2009)
Oléron Island (Fig. 2B), maximum setup locally exceeds 0.3 m although it is not really clear whether accounting for this process
during Xynthia and 0.4 m during Joachim. However, it is expected improves storm surge predictions or not (Jones and Davies, 1998).
that the maximum spatial resolution of our computational grid In this context, we took advantage of these two contrasting storms
(25 m) is not sufficient to capture the maximum setup along the to investigate the contribution of this phenomenon in storm
 X. Bertin et al. / Continental Shelf Research 96 (2015) 1–15 13

Fig. 14. Maximum difference in surface elevation between a simulation including wave forces and a simulation without wave forces for (A) Xynthia and (B) Joachim.

surges. The comparison between both approaches at the three different storm surges and sea states.
selected tide gauges revealed very limited differences, with the The comparison between a wave dependent and a bulk formula
wave dependent bottom stress generally decreasing the storm parameterization to compute the surface stress revealed that the lar-
surge by 0.05 m (Fig. 15). Such differences are too weak to con- ger-than-normal storm surge during Xynthia was explained by a
clude whether a wave dependent parameterization for bottom particular sea-state, characterized by young and steep waves. This
stress improves storm surge predictions. Alternative para- particular sea-state, rather uncommon in the North-East Atlantic
meterizations, such as that of Grant and Madsen (1979) were also Ocean, enhanced the surface stress significantly and was explained by
tested but resulted in comparable predictions. In addition, bottom the particular track of Xynthia, which reduced the fetch to a few
stress parameterization also alters tidal propagation while the hundred kilometers in the Bay of Biscay. By contrast, employing a
modeled storm surge is computed by subtracting water levels wave-dependent surface stress for Joachim did not produce significant
originating from a tide-only simulation. Therefore, the observed differences although waves were much bigger than during Xynthia.
differences probably also result from tidal propagation differences These results suggest firstly that a wave-dependent surface stress
between both simulations. However, the limited sensitivity on the should be employed for storm surge predictions in small seas such as
bottom stress parameterization may be explained by the fact that the Mediterranean, the Adriatic or the Black Sea, where young sea
the selected tide gauges are connected to channels where water states similar to those produced by Xynthia are much more common.
depths rapidly reach or exceed 10–20 m, which causes bottom More generally, a wave dependent surface stress parameterization
stress to be a second order process. To better investigate this should be preferred in operational storm surge models since it did not
phenomenon, further studies will have to be carried out with deteriorate predictions for more developed sea-states.
water level measurements by much smaller water depth. The numerical experiments on wave forces revealed that dur-
ing storms, wave setup can locally exceed 0.4 m along shorelines
exposed to large waves. More surprisingly, this analysis showed
6. Conclusions and future work that a regional setup can also develop outside breaking zones and
propagate in the inner part of estuaries. This behavior, which was
This study investigated the contribution of short waves in already shown in shallow wave dominated inlets, will have to be
storm surges, based on two storms that hit recently the central verified in other coastal environments, such as barrier islands,
part of the Bay of Biscay. Despite displaying comparable wind estuaries and rias. However, it is expected that our predictions of
fields in the study area, these storms induced substantially wave setup are locally under-estimated, first due to a lack of
14 X. Bertin et al. / Continental Shelf Research 96 (2015) 1–15

Fig. 15. Observed (black circles) against computed storm surges with a Manning parameterization bottom stress (red) and a wave-dependent parameterization (blue) at
Sables d’Olonne, La Pallice and La Cotinière for storms Xynthia and Joachim. (For interpretation of the references to color in this figure legend, the reader is referred to the
web version of this article.)

spatial resolution at high tide and second due to our 2DH ap- References
proach, which prevents a proper representation of the vertical
circulation that takes place in breaking zones. Our modeling sys- Abgrall, R., 2006. Residual distribution schemes: current status and future trends.
tem is currently being improved to integrate the latest develop- Comput. Fluids 35 (7), 641–669. http://dx.doi.org/10.1016/j.
compfluid.2005.01.007.
ments on wave–current interactions in 3D (e.g. Bennis et al., 2011), André, C., Monfort, D., Bouzit, M., Vinchon, C., 2013. Contribution of insurance data
which will allow investigating the importance of these phenom- to cost assessment of coastal flood damage to residential buildings: insights
enon on storm surges. gained from Johanna (2008) and Xynthia (2010) storm events. Nat. Hazards
Earth Syst. Sci. 13, 2003–2012. http://dx.doi.org/10.5194/nhess-13-2003-2013.
Finally, the numerical experiments on a wave-dependent bottom Apotsos, A., Raubenheimer, B., Elgar, S., Guza, R.T., Smith, J.A., 2007. Effects of wave
stress were less convincing and did not allow drawing any particular rollers and bottom stress on wave setup. J. Geophys. Res. 112, C02003. http://dx.
conclusions. The little differences obtained between the various ap- doi.org/10.1029/2006JC003549.
Ardhuin, F., Rogers, E., Babanin, A.V., Filipot, J.-F., Magne, R., Roland, A., van der
proaches suggest that the tide gauges used in this study, located in Westhuysen, A., Queffeulou, P., Lefevre, J.-M., Aouf, L., Collard, F., 2010. Semi-
intermediate water depth, were not suitable to investigate this process empirical dissipation source functions for ocean waves. Part I: definition, cali-
that is dominant in shallower depths. Measurements will have to be bration, and validation. J. Phys. Oceanogr. 40, 1917–1941.
Azevedo, A., Oliveira, A., Fortunato, A.B., Zhang, Y., Baptista, A.M., 2014. A cross-scale
carried out in shallower waters to better investigate the impact of
numerical modeling system for management support of oil spill accidents. Mar.
wave-enhanced bottom stress on storm surges. Pollut. Bull. 80 (1–2), 132–147 (Original Research Article).
Battjes, J.A., Janssen, J., 1978. Energy loss and set-up due to breaking of random
waves. In: Paper presented at 16th International Conference on Coastal En-
gineering, Coastal Engineering Research Council, Hamburg, Germany.
Acknowledgements Bennis, A.C., Ardhuin, F., Dumas, F., 2011. On the coupling of wave and three di-
mensional circulation models: choice of theoretical framework, practical im-
plementation and adiabatic tests. Ocean Model. 3–4, 260–272.
The developers of SELFE are greatly acknowledged. Water level Bennis, A.C., Dumas, F., Ardhuin, F., Blanke, B., 2014. Mixing parameterization:
data used in this study originate from the portal REFMAR (http://re impacts on rip currents and wave set-up. Ocean Eng. 84, 213–227.
fmar.shom.fr/). Tidal boundary conditions were provided by Florent Bertin, X., Li, K., Roland, A., Breilh, J.F., Zhang, Y.L., Chaumillon, E., 2014. A modeling-
based analysis of the flooding associated with Xynthia, central Bay of Biscay.
Lyard from UMR 5566 LEGOS. Wave data Offshore Oléron Island were
Coast. Eng. 94, 80–89.
provided by the French Oceanographic and Hydrographic Institute Bertin, X., Chaumillon, E., Sottolichio, A., Pedreros, R., 2005. Tidal inlet response to
(SHOM) and these data were acquired in the scope of the projects sediment infilling of the associated bay and possible implications of human
activities: the Marennes–Oléron Bay and Maumusson Inlet, France. Cont. Shelf
MOUTON (funded by DGA PEA 012401) and EPIGRAM (funded by
Res. 25, 115–1131.
LEFE/IDAO and ANR, agreement ANR-08-BLAN-0330-01). Wave data Bertin, X., Castelle, B., Chaumillon, E., Butel, R., Quique, R., 2008. Longshore trans-
in the Bay of Biscay were provided by Météo France, CETMEF and port estimation and inter-annual variability at a high energy dissipative beach:
Puerto del Estado. The development of SELFE regarding wave–current Saint Trojan beach, SW Oléron Island, France. Cont. Shelf Res. 28, 1316–1332.
Bertin, X., Fortunato, A.B., Oliveira, A., 2009. A modeling-based analysis of processes
interactions was performed in the context of the project DYNAMO, driving wave-dominated inlets. Cont. Shelf Res. 29, 819–834.
funded by the French National Research Agency (Grant agreement no. Bertin, X., Bruneau, N., Breilh, J.F., Fortunato, A.B., Karpytchev, M., 2012. Importance
ANR-12-JS02-00008-01). Finally, Aron Roland benefited from an in- of wave age and resonance in storm surges: the case Xynthia, Bay of Biscay.
Ocean Model. 42, 16–30.
vited researcher grant from Region Poitou-Charente during his visit to Bertin, X., Prouteau, E., Letetrel, C., 2013. A significant increase in wave height in the
La Rochelle. North Atlantic Ocean over the 20th century. Glob. Planet. Change 106, 77–83.
 X. Bertin et al. / Continental Shelf Research 96 (2015) 1–15 15

Breilh, J.-F., Chaumillon, E., Bertin, X., Gravelle, M., 2013. Assessment of static flood 17–74.
modeling techniques: application to contrasting marshes flooded during Xyn- Longuet-Higgins, M.S., Stewart, R.W., 1964. Radiation stress in water waves; a
thia (western France). Nat. Hazards Earth Syst. Sci. 13, 1595–1612. http://dx.doi. physical discussion with applications. Deep-Sea Res. 11, 529–562.
org/10.5194/nhess-13-1595-2013. Mastenbroek, C., Burgers, G., Janssen, P.A.E.M., 1993. The dynamical coupling of a
Bidlot, J.-R., Li, J.-G., Wittmann, P., Faucher, M., Chen, H., Lefevre, J.-M., Bruns, T., wave model and a storm surge model through the Atmospheric Boundary
Greenslade, D., Ardhuin, F., Kohno, N., Park, S., Gomez, M., 2007a. Inter-com- Layer. J. Phys. Oceanogr. 23, 1856–1866.
parison of operational wave forecasting systems. In: Proceedings of 10th In- Miles, J.W., 1957. On the generation of surface waves by shear flows. J. Fluid Mech.
ternational Workshop on Wave Hindcasting and Forecasting and Coastal Ha- 3, 185–204.
zard Symposium, North Shore, Oahu, Hawaii, November 11–16. Nicolle, A., Karpytchev, M., Benoit, M., 2009. Amplification of the storm surges in
Bidlot, J.-R., Janssen, P., Abdalla, S., 2007b. A revised formulation of ocean wave shallow waters of the Pertuis Charentais (Bay of Biscay, France). Ocean Dyn. 59,
dissipation and its model impact, ECMWF Technical Memorandum 509, 821–935. http://dx.doi.org/10.1007/s10236-009-0219-0.
ECMWF, Reading, United Kingdom, 27 pp. Available online at http://www. Nicholls, R.J., Wong, P.P., Burkett, V.R., Codignotto, J.O., Hay, J.E., McLean, R.F., Ra-
ecmwf.int/publications/. goonaden, S., Woodroffe, C.D., 2007. Coastal systems and low-lying areas. Cli-
Bidlot, J.-R., 2012. Present status of wave forecasting at ECMWF. In: Proceedings
mate change 2007: impacts, adaptation and vulnerability. In: Change, M.L.,
from the ECMWF Workshop on Ocean Waves. ECMWF, Reading, United King-
Parry, O.F., Canziani, J.P., Palutikof, P.J., van der Linden, Hanson, C.E. (Eds.),
dom, 25–27 June 2012.
Contribution of Working Group II to the Fourth Assessment Report of the In-
Blake, E.S., 2007. The deadliest, costliest and most intense United States tropical
cyclones from 1851 to 2006 (and other frequently requested hurricane tergovernmental Panel on Climate. Cambridge University Press, Cambridge, UK,
facts), NOAA Technical Memorandum NWS TPC. 5, 43 pp. pp. 315–356.
Brown, J.M., Wolf, J., 2009. Coupled wave and surge modelling for the eastern Irish Olabarrieta, M., Warner, J.C., Armstrong, B., Zambon, J.B., He, R., 2012. Ocean–at-
Sea and implications for model wind-stress. Cont. Shelf Res. 29 (10), mosphere dynamics during Hurricane Ida and Nor’Ida: an application of the
1329–1342. coupled ocean–atmosphere-wave-sediment transport (COAWST) modeling
Bowen, A.J., Inman, D.L., Simmons, V., 1968. Wave ‘‘set-down’’ and wave ‘‘set-up’’. J. system. Ocean Model. 43–44, 112–137.
Geophys. Res. 73, 2569–2577. Pairaud, I.L., Lyard, F., Auclair, F., Letellier, T., Marsaleix, P., 2008. Dynamics of the
Charnock, H., 1955. Wind stress on a water surface. Q. J. R. Meteorol. Soc. 81, semi-diurnal and quarter-diurnal internal tides in the Bay of Biscay. Part 1:
639–640. Barotropic tides. Cont. shelf Res. 28, 1294–1315. http://dx.doi.org/10.1016/j.
Chaumillon, E., Proust, J.-N., Menier, D., Weber, N., 2008. Incised valley morphol- csr.2008.03.004.
ogies and sedimentary-fills within the inner shelf of the Bay of Biscay (France): Pawlowicz, R., Beardsley, B., Lentz, S., 2002. Classical tidal harmonic analysis in-
a synthesis. J. Mar. Syst. 72, 383–396. cluding error estimates in MATLAB using T_TIDE. Comput. Geosci. 28, 929–937.
Dietrich, J.C., Bunya, S., Westerink, J.J., Ebersole, B.A., Smith, J.M., Atkinson, J.H., Phillips, O.M., 1957. On the generation of waves by turbulent wind. J. Fluid Mech. 2,
et al., 2010. A high resolution coupled riverine flow, tide, wind, wind wave and 417–445.
storm surge model for southern Louisiana and Mississippi: Part II — synoptic Pinto, L., Fortunato, A.B., Zhang, Y., Oliveira, A., Sancho, F.E.P., 2012. Development
description and analyses of Hurricanes Katrina and Rita. Mon. Weather Rev. and validation of a three-dimensional morphodynamic modelling system for
138, 378–404. non-cohesive sediments. Ocean Model. 57–58, 1–14.
Dodet, G., Bertin, X., Taborda, R., 2010. Wave climate variability in the North-East Pond, S., Pickard, G.L., 1998. Introductory Dynamical Oceanography. Butterworth-
Atlantic Ocean over the last six decades. Ocean Model. 31 (3–4), 120–131. Heinmann.
Dodet, G., Bertin, X., Bruneau, N., Fortunato, A.B., Nahon, A., Roland, A., 2013. Wave– Powell, M.D., Vickery, P.J., Reinhold, T.A., 2003. Reduced drag coefficient for high
current interactions in a wave-dominated tidal inlet. J. Geophys. Res. Oceans wind speeds in tropical cyclones. Nature 422, 279–283.
118, 1587–1605. http://dx.doi.org/10.1002/jgrc.20146. Proudman, J., 1957. Oscillations of tides and surge in an estuary of finite length. J.
Donelan, M.A., Dobsen, F., Smith, S., Anderson, R., 1993. On the dependence of sea
Fluid Mech. 2, 371–382.
surface roughness on wave development. J. Phys. Oceanogr. 23, 2143–2149.
Rego, J.L., LI, C., 2010. Nonlinear terms in storm surge predictions: effect of tide and
Doodson, A.T., 1924. Meteorological perturbations of sea-level and tides. Geophys. J.
shelf geometry with case study from Hurricane Rita. J. Geophys. Res. 115,
Int. 1, 124–147.
C06020.
Eldeberky, Y., 1996. Nonlinear Transformations of Wave Spectra in the Nearshore
Riviére, G., Arbogast, P., Lapeyre, G., Maynard, K., 2012. A potential vorticity per-
Zone (Ph.D. thesis of Delft University of Technology). The Netherlands, Delft, p.
spective on the motion of a mid-latitude winter storm. Geophys. Res. Lett. 39
203.
Flather, R.A., 2001. Storm surges. In: Steele, J., Thorpe, S., Turekian, K. (Eds.), En- (12), L12808.
cyclopedia of Ocean Sciences. Academic, San Diego, California, pp. 2882–2892. Rodrigues, M., Oliveira, A., Guerreiro, M., Fortunato, A.B., Menaia, J., David, L.M.,
Grant, W.D., Madsen, O.S., 1979. Combined wave and current interaction with a Cravo, A., 2011. Modeling fecal contamination in the Aljezur coastal stream
rough bottom. J. Geophys. Res. 84 (C4), 1797–1808. http://dx.doi.org/10.1029/ (Portugal). Ocean Dyn. 61 (6), 841–856.
JC084iC04p01797. Roland, A., Zhang, Y., Wang, H.V., Meng, Y., Teng, Y.C., Maderich, V., Brovchenko, I.,
Hasselmann, S., Hasselmann, K., Allender, J.H., Barnett, T.P., 1985. Computations and Dutour-Sikiric, M., Zanke, U., 2012. A fully coupled 3D wave–current interaction
parameterizations of the linear energy transfer in a gravity wave spectrum, II, model on unstructured grids. J. Geophys. Res., 117. http://dx.doi.org/10.1029/
parameterizations of the nonlinear transfer for application in wave models. J. 2012JC007952.
Phys. Oceanogr. 15, 1378–1391. Sheng, Y.P., Zhang, Y., Paramygin, V.A., 2010. Simulation of storm surge, wave, and
Hasselmann, K., Barnett, T.P., Bouws, E., Carlson, D.E., Hasselmann, P., 1973. Mea- coastal inundation in the Northeastern Gulf of Mexico region during Hurricane
surements of wind-wave growth and swell decay during the Joint North Sea Ivan in 2004. Ocean Model. 35 (4), 314–331.
Wave Project (JONSWAP). Deutsch. Hydrogr. Z. 8 (12), 95–102. Smith, S.D., Banke, E.G., 1975. Variation of the sea surface drag coefficient with
Holthuijsen, L.H., Powell, M.D., Pietrzak, J.D., 2012. Wind and waves in extreme wind speed. Q. J. R. Meteorol. Soc. 101, 665–673.
hurricanes. J. Geophys. Res. 117, C09003. Smith, K., 1996. Environmental Hazards: Assessing Risk and Reducing Disaster, 2nd
Idier, D., Dumas, F., Muller, H., 2012. Tide-surge interaction in the English Channel. ed.. Routledge, London.
Nat. Hazards Earth Syst. Sci. 12, 3709–3718. Soulsby, R., 1997. Dynamics of Marine Sands, a Manual for Practical Applications.
Janssen, P.A.E.M., 1989. Wave-induced stress and the drag of air flow over sea Thomas Telford, H.R. Wallingford, England, ISBN: 0-7277-2584X.
waves. J. Phys. Oceanogr. 19, 745–754. Stewart, R.W., 1974. The air–sea momentum exchange. Bound. Layer Meteorol. 6,
Janssen, P.A.E.M., 1991. Quasi-linear theory of wind wave generation applied to 151–167.
wave forecasting. J. Phys. Oceanogr. 21, 1631–1642. Swart, D.H., 1974. Offshore Sediment Transport and Equilibrium Beach Profiles.
Janssen, P.A.E.M., Saetra, O., Wettre, C., Hersbach, H., 2004. Impact of the sea state Delft Hydraulics Lab Publication, Delft, The Netherlands, p. 131.
on the atmosphere and ocean. Ann. Hydrogr. 6e série, 3 (772), 2004. Takagaki, N., Komori, S., Suzuki, N., Iwano, K., Kuramoto, T., Shimada, S., Kurose, R.,
Janssen, P.A.E.M., Breivik, O., Mogensen, K., Vitart, F., Balmaseda, M., Bidlot, J.-R., Takahashi, K., 2012. Strong correlation between the drag coefficient and the
Keeley, S., Leutbecher, M., Magnusson, L., Molteni, F., 2013. Air–sea interaction
shape of the wind sea spectrum over a broad range of wind. Geophys. Res. Lett.
and surface waves, ECMWF, Technical Memorandum. 712, 34 pp.
39, L23604.
Jelesnianski, C.P., 1965. Numerical calculation of storm tides induced by a tropical
Tolman, H.L., 2009. User Manual and System Documentation of WAVEWATCH III
storm impinging on a continental shelf. Mon. Weather Rev. 93, 343–358.
version 3.14. NOAA/NWS/NCEP/MMAB Technical Note 276, 194 p.
Jones, J.E., Davies, A.M., 1998. Storm surge computations for the Irish Sea using a
Xie, L., Pietrafesa, L.J., Wu, K., 2003. A numerical study of wave–current interaction
three-dimensional numerical model including wave–current interaction. Cont.
Shelf Res. 18, 201–251. through surface and bottom stresses: coastal ocean response to hurricane Fran
Kennedy, A.B., Westerink, J.J., Smith, J.M., Hope, M.E., Hartman, M., Taflanidis, A.A., of 1996. J. Geophys. Res. 108 (C2), 3049–3066.
Tanaka, S., Westerink, H., Cheung, K.F., Smith, T., Hamann, M., Minamide, M., Yanenko, N.N., 1971. The Method of Fractional Steps. Springer, Berlinhttp://dx.doi.
Ota, A., Dawson, C., 2012. Tropical cyclone inundation potential on the Hawaiian org/10.1007/978-3-642-65108-3.
Islands of Oahu and Kauai. Ocean Model. 52–53, 54–68. http://dx.doi.org/ Zhang, Y., Witter, R.W., Priest, G.P., 2011. Nonlinear tsunami–tide interaction in 964
10.1016/j.ocemod.2012.04.009. Prince William Sound tsunami. Ocean Model. 40 (3–4), 246–259.
Le Cann, B., 1990. Barotropic tidal dynamics of the Bay of Biscay shelf: observations, Zhang, Y.L., Baptista, A.M., 2008a. SELFE: a semi-implicit Eulerian–Lagrangian fi-
numerical modelling and physical interpretation. Cont. Shelf Res. 10 (8), nite-element model for cross-scale ocean circulation. Ocean Model. 21 (3–4),
723–758. 71–96.
Leonard, B.P., 1991. The ULTIMATE conservative difference scheme applied to un- Zhang, Z., Xiao, C., Shen, J., 2008b. Comparison of the CEST and SLOSH models for
steady one-dimensional advection. J. Comput. Methods Appl. Mech. Eng. 88, storm surge flooding. J. Coast. Res. 24 (2), 489–499.