Wave Energy Resources Along the European Atlantic Coast
Ocean wave energy has become the focus of governments and energy companies over the past decade. In spite of its unpredictability, this untapped source of energy appears to be a sustainable alternative to traditional sources of energy such as thermic and nuclear energies, or hydropower, all of which pose significant environmental and geopolitical problems. Open to the Atlantic Ocean at latitudes between 35°N and 65°N, the Atlantic Coast of Europe is blessed with one of the highest wave powers in the world—estimated to be between 33 and 76 kW/m wave crest. The European Commission has taken a proactive attitude towards encouraging and promoting the development of marine renewable energy during the near future. In this context, the European transnational project EnergyMare was commissioned to investigate the potential of marine renewable energy resources on the European Atlantic Coast as well as test innovative measurement techniques and promote the development of test sites. The targeted wave energy resources were assessed via a 10-year hindcast, using state-of-the-art spectral wave models WaveWatch III and SWAN set up on unstructured meshes or fine-resolution regular grids. The hindcasts were combined to simultaneously provide a holistic view of the wave energy distribution across the European continental shelf and fine-resolution maps of specific areas, in particular around archipelagos and complex coastlines, where wave characteristics can be affected by the presence of small islands, headlands, or irregular bathymetry, and at wave energy test sites. The domain size and timescale of the hindcasts enable a comprehensive description of the wave climate along the European Atlantic Coast, both in terms of its distribution and its seasonal and interannual variations. In particular, a comparison of wave activity at various coastal locations shows its dependence on latitude and arguably its more significant dependence on exposure to open Atlantic waters. Wave activity during the winter months is clearly predominant, but dominant peak activity was also occasionally observed during spring and autumn. In spite of increased winter wave activity over the past couple of years, data are insufficient to enable conclusions to be made about a persistent trend in the international wave climate. Continental-scale mapping of wave energy resources together with fine-resolution mapping of coastal areas provides an overview of the wave resources to help identify the best areas for energy or test sites. Such mapping also provides information about local wave characteristics and resources that can be used for diminishing installation risks or optimising a site by selecting the most appropriate devices or array configurations. In addition to evaluating wave resources, fine estimates of energy yield from a site may require a good understanding of the wave interaction in an array of converters where significant wave interference may be induced. Finally, long-term trend estimates or periodic re-evaluations of wave resources to address potential wave climate change will probably be necessary to achieve sustainable wave energy exploitation.
KeywordsMarine renewable energy Wave resources European Atlantic Coast Spectral wave modelling WaveWatch III SWAN
Requirements for reducing industrial carbon dioxide emissions, together with the depletion of fossil fuel reserves and the need for more secure energy, drive governments and energy industries to diversify their energy sources and consider more sustainable resources. Renewable energy has become a credible alternative to fossil fuels for meeting the increasing energy demand of industrialised societies. Most renewable energy already produced is from hydraulic, wind, and solar power. In contrast, marine energy represents a tiny proportion of current energy production. Yet, the Atlantic Coast of Europe has one of the most important marine renewable energy resources in the world in terms of tidal range such as in the Severn Estuary (UK) or Saint-Michel Bay (France), tidal stream such as in the Pentland Firth (UK) or near Alderney Island (UK/France), and waves.
Conscious of Europe’s abundant renewable energy resources, the European Commission has adopted a proactive attitude to encourage, promote, and develop the production of energy from these resources. The European Union Renewable Energy Directive has set a binding target to all member states of providing 20% of energy supply with renewable energy by 2020 and at least 27% by 2030 (European Commission 2015). Each European country has proposed a National Renewable Energy Action Plan (NREAP 2010) submitted under the Article 4 of Renewable Energy Directive 2009/28/EC, setting their national targets for renewable energy in accordance with their national energy consumption rates and available resources and aligning them with the European targets. Note that the National Renewable Energy Action Plan (NREAP 2010) gives a comprehensive strategy but does not differentiate between the different types of marine energy, and they encompass other marine energy sources such as thermal or osmotic energy.
Targets of renewable energy source (RES) as a percentage of gross final consumption, and marine energy contribution (installed capacity and gross electricity generation) for the five Atlantic European countries in 2020 and 2015 (NREAP 2010)
% RES (S2020)
The transnational project EnergyMare was commissioned to investigate the potential of marine renewable resources along the European Atlantic Coast, to test innovative monitoring techniques and to promote the development of test sites. Through a collaborative partnership, existing computing and monitoring resources have been combined to produce a comprehensive picture of the wave resources on the European Atlantic shelf that provides both a holistic description of the wave climate and detailed maps of sites of potential interest.
These coastal and bathymetric irregularities can influence the wave patterns near the coast and consequently introduce small-scale variability in wave energy distribution. Getting refined estimates at the areas of interest is therefore of prime interest for the wave industry, stakeholders, and regulators.
A description of wave resources can be obtained from long-term hindcasts of spectral wave models and can be supported by monitoring data. Third-generation spectral models have become the state of the art in wave modelling. Few of these models have emerged from the same fundamental equations and processes. The most common are the WAve Model (WAM), WaveWatch III, Simulating WAves Nearshore (SWAN) model, TELEMAC-based Operational Model Addressing Wave Action Computation (TOMAWAC), and MIKE21-SW. A more comprehensive description of these models is given in a previous chapter by Robertson (2016).
The first comprehensive description of wave energy resources in Europe was provided by the Wave Energy Resource Atlas (WERATLAS) project (Pontes 1998) funded by the JOULE/THERMIE programme. The WAM was used to calculate the wave parameters, significant wave height Hs, mean wave energy period Te, peak period Tp, mean direction θm, and energy flux per unit crest length Pw over the European continental shelf (49°W–45°E; 26.5°N–73°N). The values were calculated based on 85 data points of which 41 were in the Atlantic Ocean. Data collection, analysis, and interpretation were performed over the period from 1987 to 1994 for the Atlantic Ocean. WERATLAS gave the first wave hindcast at a synoptic scale but on a coarse grid (Pontes 1998; Pontes et al. 1998). The planning and development of energy sites requires finer characterisation of the wave energy to optimise the cost/benefits by selecting the most appropriate devices and array layouts. In particular, a finer wave resource characterisation could be needed in the presence of irregular coastline or complex bathymetry, which can affect the wave characteristics over short distances.
To complete a fine-resolution assessment of the wave resources at a local scale, state-of-the-art wave models were applied on unstructured or high-resolution structured meshes (see Bertotti and Cavaleri 2012). Venugopal and Nemalidinne (2015) set the spectral wave module MIKE21-SW of the MIKE21 modelling suite (DHI 2007) on an unstructured grid over the UK/Scotland waters, with fine-resolution characterisation down to 0.0005 square degrees in the Orkneys and Pentland Firth waters. The boundary conditions for this model were taken from predictions of a large-scale MIKE21-SW model extending over the North Atlantic Ocean and the North Sea (10°N–70°N and 75°W–10°E). The UK model was run for short periods during 2011 and 2012, and it was successfully validated against wave buoy data recorded at five locations around Scotland. The validation produced correlation coefficients that were higher than 0.96, for the significant wave height. The spatial distribution of the mean significant wave height and wave power around Scotland was found to be consistent with the atlas of UK renewable energy (ABP MER 2008), but with a much higher resolution, because the maps from the atlas are based on a 12 km model grid resolution in coastal areas. The wave power distribution was lowest on the eastern coasts of Scotland and highest on the western coasts, where the mean was estimated to be between 40 to 45 kW/m and the maximum values were estimated to be up to 650 kW/m near the Hebrides and Shetlands shores during January–December 2010.
Iglesias et al. (2009) investigated the wave energy potential in Galicia using the nearshore spectral wave model SWAN (Booij et al. 1999) on a 200 × 200 m grid. The open boundary conditions were provided by a large-scale WAM (18°N–69°N and 60°W–9°W) with a coarser resolution of 0.25° (approx. 30 km). The model was run over the period from 1996 to 2005 to derive a hindcast for the wave climate and wave power at various sections of the Galician coast (for locations, see Figs. 2, 13, and 15g)—Costa de la Muerte (Iglesias and Carballo 2009), Cape San Adrian to Cape Ortega (Iglesias et al. 2009), and Estaca de Bares area (Iglesias and Carballo 2010c). The model was extended to the north coast of Spain to the Asturias coastal region (Iglesias and Carballo 2010a) and the Bay of Biscay (Iglesias and Carballo 2010b). These fine-scale models applied near an irregular coastline and in the presence of a highly variable bathymetry showed that the wave energy potential can be doubled or halved over distances of only few kilometres, due to refraction, shoaling, bottom friction, sheltering, and diffraction from islands and headlands.
These studies emphasise the variety of spectral wave models and their range of applications. Although all of these third-generation spectral models use the same fundamental formulation to represent wave propagation—here the spectral action balance equation—they may differ in their numerical techniques or their representation of source–sink terms, for instance, for the wind input, whitecapping, or nonlinear wave–wave interactions (WISE Group 2007).
The existing wave models, developed by the EnergyMare project partners, used primarily WaveWatch III and SWAN, which were combined to provide a holistic view and a detailed description of the wave resources along the Atlantic Coast. In the next section, the commonalities and differences of the relevant spectral wave models are presented, with an emphasis on WaveWatch III and SWAN. This introduces a more specific description of each model and its validation against monitoring data when applicable. In the last section, the spatial distribution and temporal variability of wave energy resources across areas of the European Atlantic coastal waters, based on medium-/long-term hindcast, are presented in term of resource availability and risk to installations.
Spectral Wave Modelling
Overview of Models
The third-generation spectral wave models, developed in the late 1980s, allow for free development of the wave spectrum without any specific shape constraint. In particular, they include the energy transfer between resonant frequencies from quadruplet wave–wave nonlinear interactions (Haselmann 1962).
The so-called progressive wave models, such as WAM (WAMDI Group 1988) or WaveWatch III (Tollman 1990), are suitable for modelling wind waves in deep seas; they proved to be less accurate in coastal waters for reasons that are presented later. The SWAN model was specifically designed based on the initial WAM source code to improve wave prediction capability nearshore (Booij et al. 1999), but it is less suitable for modelling waves in the open ocean. From this point of view, estimating the wave energy resource both on synoptic and local scales can be challenging and may require a combination of the different models.
Besides the quadruplet wave–wave nonlinear transfer, the source/sink terms include whitecapping, bottom friction, duration, and fetch-limited growth from wind friction, which also takes into account the effect of wind gustiness and air density. Wave generation includes both linear (Cavaleri and Malanotte-Rizzoli 1981) and exponential growth (Komen et al. 1984) caused by wind stress. The model is operated on regular meshgrid. The WAM model proved to be reliable for deep water and therefore the open ocean, but the absence of shallow-water processes, such as depth-induced breaking or triad nonlinear wave interactions (Booij et al. 1999), made the model less accurate nearshore, in spite of late implementation of a depth-controlled algorithm for maximum wave energy and frequency down shifting. The impact of currents on waves is often minimal except in shallow nearshore areas, in particular inside the entrance of bays and harbours where tidal currents can be strong (Yang and Wang 2015).
a global time step in which the entire solution is propagated, which includes winds and currents;
a time step for spatial propagation; this time step is adjusted on the wave frequencies to ensure numerical stability and optimise the computing time;
a time step for the intra-spectral propagation; and
a time step to integrate the source terms, giving more accurate calculations for rapidly changing wind and wave conditions.
The default numerical scheme for wave propagation in WaveWatch III is the ULTIMATE QUICKEST scheme (Leonard 1979, 1991) implemented both for the physical space and for the directional space (θ-space). In the frequency space (σ-space), the scheme is adapted to take into account variable grid spacing, and a first-order upwind scheme is used for the lowest and highest wave numbers. These numerical techniques result in a more accurate replication of peak values and give a better representation of rapidly changing wind and wave conditions. Until recently, WaveWatch III could only be implemented on rectangular grids, but the newest version 4.18 can be set up on an unstructured grid (Roland 2009).
The spectral wave model SWAN was developed from WAM to improve the accuracy of spectral wave modelling in the nearshore zone. Therefore, it uses the same scientific background and equations, but includes additional functionality such as triad wave–wave nonlinear interactions and depth-induced wave breaking (Booij et al. 1999). Recently, SWAN has also been designed to run on unstructured grids, which provide a more suitable resolution nearshore especially in the presence of irregular coastlines (Zijlema 2010).
The option of implementing SWAN on either structured or unstructured meshgrid has been a major improvement to the model. Unstructured meshgrids were essentially used in finite element models such as TOMAWAC (Robertson 2016). Tuomi et al. (2014) tested the response of a spectral wave model (WAM) on a structured grid over an archipelago in the Gulf of Bothnia (Baltic Sea). They demonstrated that the model overestimated the wave energy propagating through the archipelago mainly because of slight inaccuracies in the representation of refraction and dissipation effects. The predictions are improved by using finer grids but at considerable expense of computing time. Unstructured grids allow for fine-resolution mapping nearshore and a better representation of convoluted coastlines, resulting in improved predictions as shown in the wave resource assessments of Robertson et al. (2014, 2016) conducted on the western coast of Vancouver Island, Canada. Implicit schemes are generally used with unstructured grid to avoid stability restrictions imposed by the Courant–Friedrichs-Lewy criterion. For the equation discretisation, SWAN uses an implicit Euler scheme solved by a three-direction sweep Gauss–Seidel relaxation technique to ensure a convergence of the solution for all grid points (Zijlema 2010). To avoid instability problems, the model does not allow triangular elements that have an angle wider than 143°.
Different numerical schemes are used for discretising the equations, depending on the type of simulation, node location, and user choice. First, a fully implicit first-order upwind scheme, which is robust and unconditionally stable but introduces numerical diffusion, was implemented (Booij et al. 1999). This scheme provided an accurate enough solution for wave propagation in the geographical space, but the spectral space required higher accuracy. More advanced schemes were introduced to improve the accuracy. For instance, nonstationary computations use the Stelling and Leendertse scheme (Stelling and Leendertse 1992), except next to boundary nodes where the first-order upwind scheme applies. The Stelling and Leendertse scheme is known to have such a small numerical diffusion that it can generate a so-called garden sprinkler effect due to spectral resolution over large grid intervals. To counteract that effect without increasing the numerical diffusion too much, a diffusion tensor can be added to the propagation equation (Booij and Holthuijsen 1987; Tolman 2002). In a more recent development, limiters were introduced to reduce local errors due to excessive transfer of energy in the spectral propagation by refraction or frequency shifting where the bathymetry representation is too coarse (Dietrich et al. 2013).
Model Set-up and ValidationModel Set-up and Validation
To obtain wave statistics along the European Atlantic Coast, wave hindcast modelling was carried out using WaveWatch III and SWAN over five distinct coastal zones: Scotland (UK), Ireland, France, Galicia (Spain), and Portugal.
SeaZone—1 arc second, for coastal areas and around archipelagos where the grid has the finest resolution;
SeaZone—30 arc seconds, for nearshore areas covering most of the Scottish shelf; and
GEBCO (General Bathymetric Chart of the Oceans)—1 arc minute, for the remaining offshore areas mainly near the northern, north-western, and eastern boundaries where the grid is coarser.
The bathymetry data were then averaged over the Voronoï cells related to the grid and integrated with the model.
Wind forcing uses 10 m elevation wind obtained from ECMWF reanalysis data on a 0.75° grid interval and 3-hour time interval. Sensitivity tests performed using the National Aeronautics and Space Administration’s Modern Era Retrospective-analysis for Research and Applications (MERRA) reanalysis wind data at the same elevation but with finer spatial (0.5°) and temporal (1-hour) resolution did not show much difference in the model results. However, the use of this fine-resolution data significantly increased the computational time, so the simulations were carried out using wind forcing from ECMWF wind data.
As presented by Gleizon and Woolf (2013) and Gleizon and Murray (2014), swells can travel long distances and influence wave energy even in the centre of the domain of a model of that scale. Wave boundary conditions were obtained from the predictions of a large-scale model (WaveWatch III) covering the North Atlantic basin and were specified as two-dimensional energy density spectra.
In addition to wave generation by wind forcing, the parameterisation of the model included other source/sink processes: triad and quadruplet wave–wave nonlinear interaction; bottom friction using a JONSWAP (Joint North Sea Wave Project) formulation with a constant friction coefficient of 0.038 m2s-3, which can be applied both for wind waves and for swell conditions (Hasselmann et al. 1973); whitecapping using a nonlinear saturation-based formulation (Van der Westhuysen et al. 2007); and wave breaking. However, sensitivity tests showed that these effects are minor compared to the generation by wind forcing and the boundary conditions.
Wave monitoring period and location
More accessible are the historical data from the Centre for Environment, Fisheries and Aquaculture Science (CEFAS) WaveNet directional waverider buoys. WaveNet is a strategic wave monitoring network for the UK. North of Scotland, only three buoys were deployed from the late 1990s up to today; they are in the West Hebrides (WMO ID: 62048; referred here as South Uist), Moray Firth (WMO ID: 62046), and Dounreay (decommissioned in 2001).
Complementary wave data were provided by Lews Castle College (LCC)1 who deployed directional Datawell waverider buoys MKIII off the north-east of Lewis Island (Outer Hebrides) in 2011/12 at Bragar and Siadar. The records gave a range of statistical and spectral data.
Finally, the Environmental Research Institute (ERI)2 deployed directional buoys MKIII for 6-month periods in 2012/13 at four different locations along the northern coast of Scotland: Brim Ness, Dunnet Bay, Pentland Firth, and Wick.
The model predictions are in a good agreement with monitoring data, as revealed by the coefficient of determination R2 values, which are between 0.81 and 0.91 for the significant wave height and between 0.6 and 0.79 for the mean period (Gleizon and Murray 2014).
Model calibration process involved a substantial number of sensitivity tests on model parameters, such as wind input, wave growth formulations, whitecapping, and bottom friction coefficients. The modelled output was compared to available data collected from four wave buoys off the Irish western coast (Fig. 9). The model calibrations show good correlation with monitoring data; the coefficient of determination R2 value is 0.9 for the significant wave height Hs and the R2 value is 0.8 for the mean period Tz at almost all four sites (Atan et al. 2015).
The validation of the model predictions was done on a routine basis using data from Météo-France, CEREMA (Centre D’Étude et D’Expertise sur les Risques, L’Environnement, la Mobilité et l’Aménagement), ports and harbours for coastal validation, and IFREMER altimetry data for offshore validation. The models are also used to provide near-real-time forecasts of sea state.
The model provided a 42-year period wave hindcast, from 1958 to 2000. This extended period allowed for determining wave height for different return periods (2, 10, 25, 50, and 100 years), which can be used, for instance, for evaluating the risk to installations.
Because the model only covers a narrow band along the coast, it is primarily driven by the boundary conditions. Wave boundary conditions were obtained from a 6-hour interval hindcast from the HIPOCAS (Hindcast of Dynamic Processes of the Ocean and Coastal Areas of Europe) project (Guedes Soares et al. 2002). The model was validated against wave buoy data recorded by Puertos del Estado during February 2013 at two locations: Cabo Silleiro and Coruña (see Fig. 11). It showed close predictions of the model with the data, for both significant wave height and peak period (Gleizon et al. 2015).
a large-scale domain covering the North Atlantic Ocean (NAt) with a grid resolution of 0.5 degree;
a continental-scale domain covering the south-west part of Europe (SWE) and extending from 23°W to 0°W and 33°N to 48°N with a grid resolution of 0.25 degree; and
a coastal domain covering the Portuguese continental coast (PCC) and extending from 11.8°W to 7.4°W and 35.6°N to 42.8°N with a grid resolution of 0.05 degree.
Different sources of bathymetric data were combined to populate the various nested models with appropriate resolution. The EMODNet (European Marine Observation and Data Network3) hydrographic portal provided fine-resolution bathymetry of 7.5 arc seconds, in particular for the nested models, and was completed by 30-arc second resolution global bathymetry data SRTM30_PLUS (Becker et al. 2009) without EMODNet data.
The wave energy resource was evaluated using a hindcast covering the period from 2000 to 2010. The National Centers for Environmental Prediction (NCEP) FNL Operational Model Global Tropospheric Analyses (NCEP/NWS/NOAA/U.S. Department of Commerce 2000) was used to feed the wave models with wind intensities and directions from July 1999 on a time interval of 6 h and over a grid resolution of 1 degree. The wave boundary conditions for the SWE and PCC models were simply given by the larger scale models, NAt and SWE, respectively.
Monitoring stations along the western Iberian Peninsula coast and coefficients of determination R2 for the significant wave height (HS) and mean wave periods (Tm). The subscript near the station names indicates the institution providing the data: (a) Puertos del Estado (Spain) and (b) Instituto Hidrográfico (Portugal)
Estaca de Baresa
Jan 02–Dec 09
Cabo de Peñasa
Jan 02–Dec 09
Jan 02–Dec 09
Jan 02–Dec 09
Jan 08–Dec 09
Jan 08–Dec 09
Jan 08–Dec 09
Jan 08–Dec 09
The coefficients of determination R2 show a good correlation between predicted and measured values for the significant wave height (HS); values fell between 0.89 and 0.92, except near the southern coast at the stations of Faro and Cadiz where they have lower values around 0.8. The mean period generally shows less good correlation—the R2 values fell between 0.61 and 0.75 at most locations and were 0.2 and 0.31 at the southernmost stations of Faro and Cadiz, respectively. However, considering that areas of interest for the wave energy resources are mainly from the central to northern part of Portugal, the model was deemed sufficiently accurate for estimating the resource along this coast.
Wave Resource Assessment
This hindcast gives an overview of the wave climate over the North Atlantic basin, but the exploitation of wave energy requires a finer characterisation of wave resources in coastal areas, in particular near irregular coastlines.
Near the coast, the wave power density tends to concentrate near unsheltered headlands such as the northern tip of the Shetland Islands (Fig. 18a), the westernmost headlands of Kerry County (Fig. 18b), or at Costa de la Muerte in Galicia (Fig. 18g). The fine grid resolution along Galicia shows that the wave power density can vary substantially over short distances (few 10 s of kilometres) near irregular shores. It can be observed that the mean wave power distribution depends on latitude, but probably more significantly on exposure to open waters. As evidence, the highest wave power in European coastal waters was found off the western coasts of Ireland and was estimated to be between 50 and 65 kW/m, except within sheltered bay areas (Fig. 18b).
Wave annual statistics
Estaca de Bares
The mean values of HS, H99, TP, and P were estimated at each location. The depth and exposure to Atlantic swells were variable. The deepest locations were at Kerry, Pontevedra, Belmullet, and Nazaré because of their proximity to the continental shelf slope or to a canyon (Nazaré). Table 4 corroborates the previous observations that wave activity is related to latitude and exposure. The 99 percentile of wave height, H99, is approximately 2.5 to 3 times higher than HS, but these are annual averages. Direct observations suggest that this ratio may be dependent on interannual or seasonal variability, in particular with a distinct split between winter and summer months.
Seasonal and Interannual Variability
The seasonal variability is determined for each simulated year by separating and analysing the hindcast data into four seasons: winter (December of previous year to February of current year), spring (March to May), summer (June to August), and autumn (September to November). The seasonal average is simply obtained by taking the average of the relevant seasonal values over the period of simulation.
Wave seasonal statistics
The geographical difference in wave activity is highlighted by the winter interannual variability. In Bretagne, the highest wave activity is noted during winters 2013 and 2014. In comparison, it is significantly lower during winters 2011 and 2012. Conversely, at the uppermost latitude of the Shetlands, the highest wave activity was noted during winters 2011 and 2012, and it diminished during winters 2013 and 2014. Finally, the interannual and seasonal wave resource estimates at Belmullet appear to be more consistent throughout the period studied, exhibiting a marked lower activity during summer.
No clear interannual trend can be derived from these estimates. The seasonal variability and differences between the selected locations suggest that the local wave climate is sensitive to mesoscale wind variations (~ 1000 km) over relatively short periods of time.
Summary and Discussion
The wave resources along the European Atlantic Coast are characterised using a 7-year hindcast of high-resolution spectral wave models. The modelling domains cover almost the entire European coast from the Shetland Islands in the north to the Portuguese Algarve region in the south, except for Asturias and Cantabria on the northern coast of Spain. The extent and resolution of the models can provide detailed maps of the resource for energy site developers, regulators, and/or potential users and at the same time provide a holistic description of the resource in Europe.
The high-resolution maps show that in coastal areas the wave power can vary significantly over short distances, in particular in the presence of irregular coastlines such as in Galicia, Bretagne, and Ireland, or in the vicinity of islands and archipelagos such as those in Scotland, Cotentin, or Charente. The resource can therefore only be accurately estimated using a fine-resolution grid. Tuomi et al. (2014) showed that inappropriate grid resolution can result in insufficient attenuation of waves in archipelagos and therefore in overestimating the resource. Most spectral wave models can now operate on unstructured meshes, which should be used around complex coastlines.
The wave characteristics and power density were compared at various locations selected near sites of potential interest and fairly distributed along the coastline. The comparisons showed that the wave resource depends essentially on the latitude, but perhaps more importantly on the exposure to Atlantic open waters. Located between 51°N and 55°N and frontally exposed to the North Atlantic, western Ireland has the highest wave energy resource in Europe. The annual average power density at Belmullet (north-west) and Kerry (south-west) has been estimated to be between 60 and 65 kW/m. At these locations, the wave height 99 percentile, which can be used as an indicator of peak wave activity and therefore of potential risk to installations and maintenance operations, can reach 8.5 to 9 metres during the winter months. Annual averaged peak wave periods are estimated to be between 10 and 11 s at most of the selected locations.
The seasonal variability shows a clear and consistent difference between summer (lowest wave activity) and winter (highest wave activity) at all locations. However, no clear trend emerges in the interannual variability. From the investigated locations, the most consistent interannual wave activity was found at Belmullet.
This study provides a detailed description of wave resources along the European Atlantic Coast that investigates both their spatial distribution and temporal variability using hindcast modelling. However, the energy yield of a marine energy site does not only depend on the availability of the resource. Other factors that may influence energy yield include the array layout that can optimise or reduce the resource at a local level, the morphology of the seabed that may change, for instance with accretion or erosion of sediments in sandy areas, or simply affect the local wave patterns by inducing wave breaking, refraction and shoaling, or the local hydrodynamic conditions related, for instance, to tidal currents and water levels.
In addition, wave resources are uncertain because of their inherent unpredictability over the medium to long term, which itself depends on meteorological unpredictability and long-term uncertainties caused by climate change. Statistics from hindcast modelling probably provide the closest evaluation of the resource, which need to be regularly re-evaluated to adjust for longer term changes.
This work was funded by the European Regional Development Fund under the EnergyMare project of the Atlantic Area Programme.
The authors are grateful to INCAT (Ingeniería Civil de Atlàntico) for post-processing the model output data and for producing maps of wave height and wave power along the Galician coast, and to Thibault Coulombier and Camille Letretel for their help in providing additional data related to the French coast.
- ABP MER. (2008). Atlas of UK marine renewable energy resources: Atlas pages. A strategic environmental assessment report, Department for Business Entreprise and Regulatory Reform.Google Scholar
- Atan, R., Goggins, J., & Nash, S. (2015). A preliminary assessment of the wave characteristics at the Atlantic Marine Energy Test Site (AMETS) using SWAN. In Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France.Google Scholar
- Barstow, S., Mollison, D., & Cruz, J. (2008). The wave energy resource. In Cruz, J. (Ed.), Ocean wave energy (pp. 40). Springer.Google Scholar
- Barstow, S., Mørk, G., Lønseth, L., & Mathisen J. P. (2009). WorldWaves wave energy resource assessments for the deep ocean to the coast. In Proceedings of the 8th European Wave and Tidal Energy Conference, Uppsala, Sweden (pp. 149–159).Google Scholar
- Cavaleri, L., & Malanotte-Rizzoli, P. (1981). Wind wave prediction in shallow water. Theory and applications. Journal of Geophysical Research, 86C11, 10961–10973.Google Scholar
- DHI. (2007). MIKE21 SW—Spectral waves FM module User guide. Denmark: Scientific Document, Danish Hydraulic Institute.Google Scholar
- European Commission. (2015). Renewable energy progress report, Report from the commission to the European parliament, the council, the European economic and social committee and the committee of the regions, SWD (2015), p. 117.Google Scholar
- Gleizon, P., Campuzano F. J., Carracedo García, P., Gomez B., & Martinez, A. (2015). Wave energy mapping along the European Atlantic coast. In Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France.Google Scholar
- Gleizon, P., & Murray, A. (2014). Modelling wave energy in archipelagos—Case of northern Scotland. In Proceedings of the 2nd Environmental Interactions of Marine Renewable Energy Technologies, Stornoway, United Kingdom.Google Scholar
- Gleizon, P., & Woolf, D. (2013). Wave energy assessment in Scotland. In Proceedings of the 10th European Wave and Tidal Energy Conference, Aalborg, Denmark.Google Scholar
- Guedes Soares, C., Weisse, R., Carretero, J. C., & Alvarez, E. (2002). A 40 years hindcast of wind, sea level and waves in European waters. In Proceedings of the 21st International Conference on Offshore Mechanics and Arctic Engineering (OMAE 2002), Oslo, Norway.Google Scholar
- Hasselmann, K., Barnett, T. P., Bouws, E., Carlson, H., Cartwright D. E., Enke, K., et al. (1973). Measurements of wind-wave growth and swell decayduring the Joint North Sea Wave Project (JONSWAP), Deutschen Hydrographischen Zeitschrift, Suppl. A(8) n° 12, pp. 95.Google Scholar
- Leonard, B. P. (1979). A stable an accurate convective modelling procedure based on quadratic upstream interpolation. Computer Methods in Applied Mechanics and Engineering, 18, 17–74.Google Scholar
- May, V. J., Hansom, J. D. (2003). Coastal geomorphology of Great Britain, Geological Conservation Review Series (Vol. 28, pp. 754). ISBN 1 86107 484 0.Google Scholar
- NCEP/NWS/NOAA/U.S. Department of Commerce. (2000). NCEP FNL Operational Model Global Tropospheric Analyses, continuing from July 1999, Research Data Archive at the National Center for Atmospheric Research, Computational and Information Systems Laboratory, Boulder, Colorado. Retrieved February 11, 2015, from http://dx.doi.org/10.5065/D6M043C6.
- Nielsen, P. (2009). Coastal and Estuarine processes, advanced series on ocean engineering (Vol. 29). World Scientific.Google Scholar
- NREAP (National Renewable Energy Action Plan). (2010). http://ec.europa.eu/energy/en/topics/renewable-energy/national-action-plans.
- Piollé, J. F., Croizé-Fillon, D. (2012). Operation of CERSAT data centre. Ifremer report 13/2 213 213.Google Scholar
- Pontes, M. T., Athanassoulis, G. A., Barstow, S., Bertotti, L., Cavaleri, L., Holmes, B., et al. (1998). The European wave energy resource. In Proceedings of the 3rd European Wave Energy Conference, Patras, Greece.Google Scholar
- Robertson, B. (2017). Wave energy assessments: quantifying the resource and understanding the uncertainty. In Z. Yang & A. Copping (Ed.), Marine renewable energy: Resource characterization, practical energy harvest and effects on physical systems. Springer.Google Scholar
- Roland, A. (2009) Development of WWM II: Spectral wave modelling on unstructured meshes, Ph.D. thesis, Technische University at Darmstadt, Institute of Hydraulic and Water Resources Engineering.Google Scholar
- Stelling, G. S., & Leendertse J. J. (1992). Approximation of convective processes by cyclic AOI methods. In Proceedings of 2nd International Conference on Estuarine and Coastal Modelling, ASCE Tampa, Florida (pp. 771–782).Google Scholar
- WAMDI group (1988) The WAM model—a third generation wave prediction model, Journal of Physical Oceanography, 18, 1775–1810.Google Scholar
- WISE Group. (2007). Wave modeling—The state of the art. Progress in Oceanography, 75, 603–674.Google Scholar
- Yang, Z., & Wang T. (2015). Modelling wave resource characterization using and unstructured grid coastal ocean model. In Proceedings of the 11th European Wave and Tidal Energy Conference 2015, Nantes, France.Google Scholar