Impact of Climate Change on Wave Energy Resource in the Mediterranean Coast of Morocco

: The increasing demand for energy and the impacts generated by CO 2 emissions make it necessary to harness all possible renewable sources of energy, like wave power. Nevertheless, climate change may generate signiﬁcant variations in the amount of wave energy available in a certain area. The aim of this paper is to study potential changes in the wave energy resource in the Mediterranean coast of Morocco due to climate change. To do this, wave datasets obtained by four institutes during the Coordinated Regional Climate Downscaling Experiment in the Mediterranean Region (Med-CORDEX) project are used. The future conditions correspond to the RCP4.5 and RCP8.5 scenarios from the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. The results show that projected future wave power is very similar to that of the present considering the whole area, although at some speciﬁc points there are slight changes that are more evident for the RCP8.5 scenario. Another remarkable result of this study is the signiﬁcant increase of the temporal variability of wave power in future scenarios, in particular for RCP8.5. This will be detrimental for the deployment of wave energy converters in this area since their energy output will be more unevenly distributed over time, thus decreasing their e ﬃ ciency. R.C. M.M. and C.M.; data curation, R.C., P.L. and L.M.; writing—original draft preparation, J.P.S.; writing—review and editing, J.P.S., M.M. and P.L.; visualization, J.P.S. and R.C.; supervision, J.P.S. and M.M.; project administration, J.P.S. and C.M.


Introduction
Energy demand is continuously growing due to the increase in population and economic activity. At the same time, the use of traditional energy sources increases the emissions of greenhouse gases that contribute to climate change. In emergent countries without reserves of fossil fuels, such as Morocco, the combination of both factors makes it extremely necessary to promote the development of renewable energy sources to reduce the carbon footprint and their dependence on the import of energy and commodities.
Bearing this in mind, Morocco has undertaken an ambitious program to reduce its reliance on conventional energy sources [1]. As shown in several studies, the country has a high potential for developing renewable energies, in particular solar [2,3] and wind [4,5] energy, which at present represent 6.5% and 11.2% of the total installed power in the country [6]. Other renewable sources such as hydropower, biomass and geothermal are also expected to increase their participation in Morocco's directions are E and northeast (NE) [70] although, at some points, waves from west-northwest (WNW) and west (W) also become significant [72]. Regarding the future climate, the exact evolution of atmospheric patterns due to climate change is hard to predict because of the coexistence of many interacting processes [73]. Nevertheless, many studies project wind intensification over northern Europe and weaker wind in southern Europe and the Mediterranean [62,74,75]. As a consequence, most studies suggest a reduction in the number of Mediterranean cyclones [73], although there is no consensus on whether the frequency of intense cyclones will increase or decrease [66,76]. This expected reduction of the number of future storms in the Mediterranean will lead to a general decrease of wave heights, especially during winter, while in summer an opposite trend is projected [73].
This study uses the wave outputs obtained from four climate models provided by the Centro Euro-Mediterraneo sui Cambiamenti Climatici (CMCC, Italy), the Centre National de Recherches Météorologiques (CNMR, France), the Goethe-Universität Frankfurt am Main (GUF, Germany) and the Laboratoire de Météorologie Dynamique (LMD France), as part of the numerical simulations produced within the Coordinated Regional Climate Downscaling Experiment in the Mediterranean Region Med-CORDEX project [90]. These simulations consider the new Representative Concentration Pathways (RCPs) adopted by the Intergovernmental Panel on Climate Change (IPCC) for its Fifth Assessment Report [91]. These pathways describe different future climates depending on the evolution of greenhouse gases emissions during the 21st century, and are named with a number that indicates the radiative forcing (i.e., the difference between incoming and outgoing energy in the atmosphere) in 2100, expressed in W/m 2 . The two scenarios selected are one intermediate (RCP4.5) and one pessimistic (RCP8.5). The first one assumes emissions reaching a peak in 2040 and then Regarding the future climate, the exact evolution of atmospheric patterns due to climate change is hard to predict because of the coexistence of many interacting processes [73]. Nevertheless, many studies project wind intensification over northern Europe and weaker wind in southern Europe and the Mediterranean [62,74,75]. As a consequence, most studies suggest a reduction in the number of Mediterranean cyclones [73], although there is no consensus on whether the frequency of intense cyclones will increase or decrease [66,76]. This expected reduction of the number of future storms in the Mediterranean will lead to a general decrease of wave heights, especially during winter, while in summer an opposite trend is projected [73].
This study uses the wave outputs obtained from four climate models provided by the Centro Euro-Mediterraneo sui Cambiamenti Climatici (CMCC, Italy), the Centre National de Recherches Météorologiques (CNMR, France), the Goethe-Universität Frankfurt am Main (GUF, Germany) and the Laboratoire de Météorologie Dynamique (LMD France), as part of the numerical simulations produced within the Coordinated Regional Climate Downscaling Experiment in the Mediterranean Region Med-CORDEX project [90]. These simulations consider the new Representative Concentration Pathways (RCPs) adopted by the Intergovernmental Panel on Climate Change (IPCC) for its Fifth Assessment Report [91]. These pathways describe different future climates depending on the evolution of greenhouse gases emissions during the 21st century, and are named with a number that indicates the radiative forcing (i.e., the difference between incoming and outgoing energy in the atmosphere) in 2100, expressed in W/m 2 . The two scenarios selected are one intermediate (RCP4.5) and one pessimistic (RCP8.5). The first one assumes emissions reaching a peak in 2040 and then declining, attaining a radiative forcing of 4.5 W/m 2 in 2100. On the contrary, in the RCP8.5 scenario the emissions rise throughout all the century, reaching a radiative forcing of 8.5 W/m 2 in 2100.
The datasets used herein consist of 20-year-long time series of wave parameters (H s , peak period T p and wave direction θ) on a three-hour basis describing the present (1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005) and the future (2081-2100) climates for the two analyzed scenarios (RCP4.5 and RCP8.5). Each dataset is named with the acronym of the corresponding institute. These time periods were selected because they are the same used by the IPCC in its Fifth Assessment Report (AR5), and they fit the goal of the paper for comparing possible modifications in the wave power resource between present and future conditions based on IPCC scenarios.
For this study, the wave power is assessed at 14 grid points covering the Moroccan Mediterranean coast, as shown in Figure 1; the geographical coordinates, water depth and distance to the coast for each point are indicated in Table 1. Table 1, the points used to carry out the study are located in water depths greater than 100 m, so it is assumed that they are located in deep water. Although for the longest wave periods this condition is not fulfilled, the number of wave data involved is small (<5% in the worst case) and the error incurred by assuming this condition is small compared to the inherent errors associated to the use of numerical model data. Therefore, to assess the theoretical wave power, the following deep-water expression is used:

As shown in
where P is the wave power (kW/m), H s is the significant wave height, T e is the energy period, ρ is the seawater density (taken as 1025 kg/m 3 ) and g is the acceleration of gravity. The energy period can be obtained from the spectral moments of order 0 (m 0 ) and −1 (m −1 ) as follows: However, as indicated by Cornett [92], wave data are often given in terms of significant wave height H s and either mean period T z or peak period T p , but not energy period T e . Most of the times this period must be obtained from other variables if, as in this case, the spectral moments are unknown. In terms of the peak period, one way to estimate it is: where α is a coefficient whose value depends on the wave spectrum features (0.86 for a Pierson-Moskowitz (PM) spectrum and increasing towards 1 with decreasing spectral width) [92]. Taking into account that mixed sea states prevail in this area (combination of sea and swell waves) [73], wave spectra are rather wide [73]. Nevertheless, since the Mediterranean Sea is a closed basin, the fetches are limited and the fully developed sea states of the PM spectrum cannot be reached. Therefore, as suggested by [92], amongst others, a value of T e = 0.9T p is selected to determine the wave energy resource. Taking into account Equations (1) and (3), and using all the datasets indicated in Section 2.2, the 20-year-long time series of wave energy resource at each point of Figure 1 is obtained for the present and future (two scenarios). Moreover, by considering the wave direction, the directional distribution of wave power is calculated for whole datasets. This distribution is of particular interest when the wave energy is going to be harvested using nonpoint absorber devices.
The temporal variability of the total wave energy at different time scales is also very important when assessing the energy resource, since it usually varies significantly throughout the year and, taking it into account, is fundamental for evaluating the performance of wave energy converters. As pointed out by several authors [93], an estimate based only on mean annual values may result in wrong decision making. When selecting the location for WEC installation, sites with a more regular time distribution of the wave energy flux are preferable to those with very irregular and unsteady wave conditions, since they are more reliable and efficient for electricity production.
To assess the temporal variability of the wave energy at a specific location, the coefficients proposed by [92] are used: the seasonal variability index (SV) and the monthly variability index (MV). SV is defined as SV = P s1 − P s4 P year (4) where P S1 is the average wave power for the highest-energy season, P S4 is the average wave power for the lowest-energy season and P year is the average annual wave power. The greater the value of SV, the larger the seasonal variability of the wave power. On the other hand, MV is defined as where P M1 is the average wave power for the highest-energy month and P M12 is the average wave power for the lowest-energy month. Both indices are also assessed at each point and for all the available datasets.

Results for the Present Situation
In this section, the average annual wave power at each point, as well as its directional distribution and temporal variability, are assessed for the four available datasets. The values obtained for wave energy power are in the range of those estimated in other studies [38,41,46], in particular for CMCC, GUF and the multimodel ensemble. On the contrary, the CNRM data greatly overpredict wave power (except in the western part of the study area), while LMD data slightly underpredict it. With respect to the spatial distribution, the most energetic points are those located in the central stretch of the study area (P7 to P11) with mean wave power greater than 4.4 kW/m, and up to 5 kW/m for the ensemble. In the western part of the analyzed coast (points P1 to P6), the obtained values are smaller (between 2.7 and 3.6 kW/m for the ensemble), with the exception of P5 that only has 1.1 kW/m. In the easternmost zone (P12 to P14), the wave power is comparable to that of the western part (between 2.6 and 3.3 kW/m for the ensemble). Although there is a remarkable uncertainty due to the intermodel variability, the four models follow similar patterns at all points. kW/m, and up to 5 kW/m for the ensemble. In the western part of the analyzed coast (points P1 to P6), the obtained values are smaller (between 2.7 and 3.6 kW/m for the ensemble), with the exception of P5 that only has 1.1 kW/m. In the easternmost zone (P12 to P14), the wave power is comparable to that of the western part (between 2.6 and 3.3 kW/m for the ensemble). Although there is a remarkable uncertainty due to the intermodel variability, the four models follow similar patterns at all points.

Directional Average Wave Power
Besides the total wave power available at a certain point, it is important to know its directional distribution because this may condition the type of WECs that can be used. This distribution has been obtained for the multimodel ensemble by averaging the wave energy of the four models in each direction. The results for present wave conditions are shown in Figure 3. At the points located in the western area (P1 to P4), the most energetic waves come from the E, whereas, when moving towards the east, waves coming from W and WNW become more important, being the prevailing directions at almost all the points. At the same time, waves from the E turn to ENE (points P8 to P11) or NE (points P12 to P14).
In summary, in the study area almost all the energy is presently concentrated in two directional sectors: the first is from E to NE and the second is from west-southwest (WSW) to WNW. Figure 4 shows the monthly variability of present wave power, while Figure 5 illustrates its seasonal distribution. As in the case of the annual wave power, there is a large intermodel variability, with the four models showing a different temporal distribution of wave power. The results from the CMCC model are compatible with those observed in other Mediterranean regions, with winter and autumn being the most energetic seasons, and summer the mildest period. On the contrary, the other three models show that the most energetic season is spring. Another remarkable feature of the model results is that wave energy is more evenly distributed throughout the year, with smaller differences than expected between seasons. Thus, the wave power ratio between the most energetic and the less energetic seasons is smaller than 2, while in other areas of this region it can reach up to 7.4 [59]. Considering the intermodel ensemble, about 30% of the wave power in the area corresponds to spring, 25% to winter, 24% to autumn and 21% to summer. These values represent a contrast with those found in other areas of the Mediterranean (e.g., [37]), where the seasonal distribution of wave power shows a larger variability.

Directional Average Wave Power
Besides the total wave power available at a certain point, it is important to know its directional distribution because this may condition the type of WECs that can be used. This distribution has been obtained for the multimodel ensemble by averaging the wave energy of the four models in each direction. The results for present wave conditions are shown in Figure 3. At the points located in the western area (P1 to P4), the most energetic waves come from the E, whereas, when moving towards the east, waves coming from W and WNW become more important, being the prevailing directions at almost all the points. At the same time, waves from the E turn to ENE (points P8 to P11) or NE (points P12 to P14).
In summary, in the study area almost all the energy is presently concentrated in two directional sectors: the first is from E to NE and the second is from west-southwest (WSW) to WNW. Figure 4 shows the monthly variability of present wave power, while Figure 5 illustrates its seasonal distribution. As in the case of the annual wave power, there is a large intermodel variability, with the four models showing a different temporal distribution of wave power. The results from the CMCC model are compatible with those observed in other Mediterranean regions, with winter and autumn being the most energetic seasons, and summer the mildest period. On the contrary, the other three models show that the most energetic season is spring. Another remarkable feature of the model results is that wave energy is more evenly distributed throughout the year, with smaller differences than expected between seasons. Thus, the wave power ratio between the most energetic and the less energetic seasons is smaller than 2, while in other areas of this region it can reach up to 7.4 [59]. Considering the intermodel ensemble, about 30% of the wave power in the area corresponds to spring, 25% to winter, 24% to autumn and 21% to summer. These values represent a contrast with those found in other areas of the Mediterranean (e.g., [37]), where the seasonal distribution of wave power shows a larger variability. The more uniform distribution of wave power throughout the year is also observed in Figures 6  and 7, in which the coefficients MV and SV that measure the monthly and seasonal power variability are shown. Although there are differences among the four models, most of the values of MV ( Figure  6) are smaller than one, except for the CMCC model in the western part of the study area and the LMD and CNMR models in the easternmost stretch. The values of the ensemble are also smaller than one at all locations, indicating that the monthly variability of the wave power in this area is relatively moderate under present conditions. This is beneficial for the potential deployment of WECs, since their energy output would be more regular throughout the year, as confirmed also by the SV values ( Figure 7) that are smaller than 0.85 for all points and models, and smaller than 0.6 at all points for the intermodel ensemble. Therefore, although the wave power potential of this area is relatively modest, its small temporal variability due to its uniform monthly and seasonal distribution favors its consideration for wave power harvesting. The more uniform distribution of wave power throughout the year is also observed in Figures 6  and 7, in which the coefficients MV and SV that measure the monthly and seasonal power variability are shown. Although there are differences among the four models, most of the values of MV ( Figure 6) are smaller than one, except for the CMCC model in the western part of the study area and the LMD and CNMR models in the easternmost stretch. The values of the ensemble are also smaller than one at all locations, indicating that the monthly variability of the wave power in this area is relatively moderate under present conditions. This is beneficial for the potential deployment of WECs, since their energy output would be more regular throughout the year, as confirmed also by the SV values ( Figure 7) that are smaller than 0.85 for all points and models, and smaller than 0.6 at all points for the intermodel ensemble. Therefore, although the wave power potential of this area is relatively modest, its small temporal variability due to its uniform monthly and seasonal distribution favors its consideration for wave power harvesting.              Figures 8 and 9 compare respectively the multimodel ensemble mean significant wave heights (Hs) and mean energy periods (Te) under present conditions and both future scenarios (RCP4.5 and RCP8.5). In the case of Hs (Figure 8), scenario RCP4.5 shows slight variations (between -1.2% and 1.7%) with increases being located in the western part of the study area. In the case of scenario RCP8.5, the model ensemble projects greater variations of the average Hs, with increases (between 0.3% and 7%) also in the western part of the studied domain and decreases (between 0.9% and 2%) in the eastern part.

Wave Parameters
In the case of Te (Figure 9), the projected future average values are very similar to those of the present conditions. Thus, in scenario RCP4.5, these periods vary between -0.2% and 1.2%, and in scenario RCP8.5, between -1% and 1.4%. Taking into account the results shown in Figures 8 and 9, it is evident that future modifications in wave power will be mainly due to changes in wave height rather than wave period.   Figures 8 and 9 compare respectively the multimodel ensemble mean significant wave heights (H s ) and mean energy periods (T e ) under present conditions and both future scenarios (RCP4.5 and RCP8.5). In the case of H s (Figure 8), scenario RCP4.5 shows slight variations (between −1.2% and 1.7%) with increases being located in the western part of the study area. In the case of scenario RCP8.5, the model ensemble projects greater variations of the average H s , with increases (between 0.3% and 7%) also in the western part of the studied domain and decreases (between 0.9% and 2%) in the eastern part.

Wave Parameters
In the case of T e (Figure 9), the projected future average values are very similar to those of the present conditions. Thus, in scenario RCP4.5, these periods vary between −0.2% and 1.2%, and in scenario RCP8.5, between −1% and 1.4%. Taking into account the results shown in Figures 8 and 9, it is evident that future modifications in wave power will be mainly due to changes in wave height rather than wave period.  Figures 8 and 9 compare respectively the multimodel ensemble mean significant wave heights (Hs) and mean energy periods (Te) under present conditions and both future scenarios (RCP4.5 and RCP8.5). In the case of Hs (Figure 8), scenario RCP4.5 shows slight variations (between -1.2% and 1.7%) with increases being located in the western part of the study area. In the case of scenario RCP8.5, the model ensemble projects greater variations of the average Hs, with increases (between 0.3% and 7%) also in the western part of the studied domain and decreases (between 0.9% and 2%) in the eastern part.

Wave Parameters
In the case of Te (Figure 9), the projected future average values are very similar to those of the present conditions. Thus, in scenario RCP4.5, these periods vary between -0.2% and 1.2%, and in scenario RCP8.5, between -1% and 1.4%. Taking into account the results shown in Figures 8 and 9, it is evident that future modifications in wave power will be mainly due to changes in wave height rather than wave period.    Figure 10 compares the multimodel ensemble wave power under present conditions and both future scenarios (RCP4.5 and RCP8.5), while Table 2 presents the corresponding numerical values and the percentage of variation. This percentage was calculated as the difference between the wave power in future scenarios and at present, and divided by the latter. Positive values reflect future increases of wave power, while negative values indicate future decreases.

Wave Power
As mentioned in the preceding section, present conditions indicate moderate wave energy potential, ranging from 1.12 to 5.16 kW/m, which are similar to values found by other authors in this area. The most energetic zone is concentrated in the central stretch of the study area (points P7 to P11). For future climate conditions, scenario RCP4.5 shows a slight decrease (between -0.4% and -3.9%) of the wave resource at 10 points, particularly in the central and eastern parts of the studied domain. At the other four points (P2 to P5) there is a slight increase of the wave power (between 0.4% and 2%). For the RCP4.5 scenario, the model ensemble projects an average decrease of 1.8% of the present wave power, considering the whole studied area (average of the wave energy at all points).    Table 2 presents the corresponding numerical values and the percentage of variation. This percentage was calculated as the difference between the wave power in future scenarios and at present, and divided by the latter. Positive values reflect future increases of wave power, while negative values indicate future decreases.

Wave Power
As mentioned in the preceding section, present conditions indicate moderate wave energy potential, ranging from 1.12 to 5.16 kW/m, which are similar to values found by other authors in this area. The most energetic zone is concentrated in the central stretch of the study area (points P7 to P11). For future climate conditions, scenario RCP4.5 shows a slight decrease (between −0.4% and −3.9%) of the wave resource at 10 points, particularly in the central and eastern parts of the studied domain. At the other four points (P2 to P5) there is a slight increase of the wave power (between 0.4% and 2%). For the RCP4.5 scenario, the model ensemble projects an average decrease of 1.8% of the present wave power, considering the whole studied area (average of the wave energy at all points). Table 2. Comparison of the average wave power (obtained from the multimodel ensemble) between present and future conditions. The percentages represent the relative variation of wave power under future conditions with respect to present ones.  In the case of scenario RCP8.5, the study area shows two behaviors. In the western part (points P1 to P7) the intermodel ensemble projects an increase of wave power in the future, with variations between 0.3% and 7.9%; on the contrary, in the eastern part, future wave power projections indicate a reduction between 1.2% and 6.1%, except at point P12, where an increase of 1.5% is obtained. The average wave power of the 14 points for this scenario gives practically the same value as the average wave power of the 14 points for present conditions (with a difference of -0.1%).

Point
Although Concerning the directional distribution of the wave power, model results do not show significant changes in the future scenarios; for this reason the results are not plotted.

Temporal Variability
The evolution of the temporal variability in future scenarios with respect to the present situation is plotted in Figures 11 and 12, where the mean wave power (intermodel ensemble and averaged over all the points) for each month and season is presented. In addition, in Tables 3 and 4, this monthly and seasonal variation of the wave power is summarized, indicating the magnitude of the changes.
For scenario RCP4.5, the monthly wave power in the study area ( Figure 11 and Table 3) is between 1% and 10.3% smaller for eight out of twelve months, while it increases between 2.5% and 16% for March, May, August and October. For scenario RCP8.5, the changes of the monthly wave power are generally greater, with larger wave power during six months and smaller power during the other six. There is a noticeable decrease of wave power for all winter months (between -13% and -22.3%), November (-13.1%) and part of the summer (July and August, between -4.3% to -14.3%), whereas the monthly wave power increases from March to June (between 4.5% and 24%), September (3.4%) and October (22.1%).
Analyzing the seasonal changes ( Figure 12 and Table 4), for the RCP4.5 scenario the multiensemble average shows a slight increase in the spring and autumn wave power (up to 2%), a slight reduction in summer (-1.8%) and a larger decrease in winter (-9.6%). For scenario RCP8.5, the changes follow the same pattern but are clearly magnified, with values that are around twice those for the RCP4.5 scenario, except for spring, when the variation in wave power rises from +2% to +14%. The significant decrease forecasted for winter is consistent with the projections of other studies [59,73]. In the case of scenario RCP8.5, the study area shows two behaviors. In the western part (points P1 to P7) the intermodel ensemble projects an increase of wave power in the future, with variations between 0.3% and 7.9%; on the contrary, in the eastern part, future wave power projections indicate a reduction between 1.2% and 6.1%, except at point P12, where an increase of 1.5% is obtained. The average wave power of the 14 points for this scenario gives practically the same value as the average wave power of the 14 points for present conditions (with a difference of −0.1%).
Although the mean wave power projected under present conditions (3.60 kW/m) and both future scenarios Concerning the directional distribution of the wave power, model results do not show significant changes in the future scenarios; for this reason the results are not plotted.

Temporal Variability
The evolution of the temporal variability in future scenarios with respect to the present situation is plotted in Figures 11 and 12, where the mean wave power (intermodel ensemble and averaged over all the points) for each month and season is presented. In addition, in Tables 3 and 4, this monthly and seasonal variation of the wave power is summarized, indicating the magnitude of the changes.
For scenario RCP4.5, the monthly wave power in the study area ( Figure 11 and Table 3) is between 1% and 10.3% smaller for eight out of twelve months, while it increases between 2.5% and 16% for March, May, August and October. For scenario RCP8.5, the changes of the monthly wave power are generally greater, with larger wave power during six months and smaller power during the other six. There is a noticeable decrease of wave power for all winter months (between −13% and −22.3%), November (−13.1%) and part of the summer (July and August, between −4.3% to −14.3%), whereas the monthly wave power increases from March to June (between 4.5% and 24%), September (3.4%) and October (22.1%).
Analyzing the seasonal changes ( Figure 12 and Table 4), for the RCP4.5 scenario the multiensemble average shows a slight increase in the spring and autumn wave power (up to 2%), a slight reduction in summer (−1.8%) and a larger decrease in winter (−9.6%). For scenario RCP8.5, the changes follow the same pattern but are clearly magnified, with values that are around twice those for the RCP4.5 scenario, except for spring, when the variation in wave power rises from +2% to +14%. The significant decrease forecasted for winter is consistent with the projections of other studies [59,73]. These changes in the distribution of wave energy by months and seasons are transferred to the variability coefficients MV and SV, as shown in Figures 13 and 14, in which their present and future values for the 14 points are plotted. Figure 13 shows how MV values for the RCP4.5 scenario are greater than those for the present situation, in particular in the western part of the study area (increases between 10% and 20% at points P2 to P8), while in the eastern area the increases are smaller, with even a slight reduction of the MV value at the easternmost point (P14) for this scenario. Table 3. Comparison of the average monthly wave power (obtained from the multimodel ensemble and considering all the study area) between present and future conditions. The percentages represent the relative variation of future with respect to present conditions.  These changes in the distribution of wave energy by months and seasons are transferred to the variability coefficients MV and SV, as shown in Figures 13 and 14, in which their present and future values for the 14 points are plotted. Figure 13 shows how MV values for the RCP4.5 scenario are greater than those for the present situation, in particular in the western part of the study area (increases between 10% and 20% at points P2 to P8), while in the eastern area the increases are smaller, with even a slight reduction of the MV value at the easternmost point (P14) for this scenario. Table 3. Comparison of the average monthly wave power (obtained from the multimodel ensemble and considering all the study area) between present and future conditions. The percentages represent the relative variation of future with respect to present conditions.    On the contrary, for scenario RCP8.5, the increase of MV values is greater than for scenario RCP4.5 (except at points P2, P3 and P5) and, in particular, in the eastern part of the study area. At points P3 to P14, the increase of MV values with respect to the present ones ranges between 15% and 31%. Nevertheless, although this means a significant increase of the monthly variability of wave power, the maximum values of MV are smaller than 1.15, indicating that the monthly variability is moderate, even in the worst scenario.

Winter
Spring Summer Autumn     On the contrary, for scenario RCP8.5, the increase of MV values is greater than for scenario RCP4.5 (except at points P2, P3 and P5) and, in particular, in the eastern part of the study area. At points P3 to P14, the increase of MV values with respect to the present ones ranges between 15% and 31%. Nevertheless, although this means a significant increase of the monthly variability of wave power, the maximum values of MV are smaller than 1.15, indicating that the monthly variability is moderate, even in the worst scenario.

Winter
Spring Summer Autumn On the contrary, for scenario RCP8.5, the increase of MV values is greater than for scenario RCP4.5 (except at points P2, P3 and P5) and, in particular, in the eastern part of the study area. At points P3 to P14, the increase of MV values with respect to the present ones ranges between 15% and 31%. Nevertheless, although this means a significant increase of the monthly variability of wave power, the maximum values of MV are smaller than 1.15, indicating that the monthly variability is moderate, even in the worst scenario. Concerning the SV index ( Figure 14), in the RCP4.5 scenario it increases at all points except at both ends, which show a slight decrease (around 3%). The increases are particularly significant in the central part of the studied stretch (points P5 to P9) and in P12, with variations between 14% and 28%, although the maximum value of SV for this scenario is 0.6, indicating a low seasonal variability.
As for MV, the changes in the SV values are larger for the RCP8.5 scenario revealing great variations with respect to present values. The increases range between 19% and 63%, indicating a larger seasonal variability. In any case, the maximum value of SV is 0.71, which denotes a reduced variability of wave power throughout the seasons.
A test of hypothesis was carried out for both parameters MV and SV with a level of significance α = 0.05. In both cases and for both scenarios the null hypothesis was rejected, indicating that these changes are statistically significant.
Therefore, the intra-annual variability of wave power will increase significantly in the future, in particular for the RCP8.5 scenario. Nevertheless, since in the present situation the distribution of wave power throughout the year is fairly regular, such variability will be limited and considerably smaller than in other areas of the Mediterranean Sea, which present a strong seasonal character.

Summary and Conclusions
In this paper, the potential wave power in the Mediterranean coast of Morocco was assessed for present (period 1986-2005) and future (2081-2100) conditions, considering two climate change scenarios (RCP4.5 and RCP8.5). The aim was to assess how this resource will be modified by climate change. For this, wave datasets from four European institutes generated during the Med-CORDEX project, and fourteen grid points located along the Moroccan Mediterranean coast were used.
The multimodel ensemble projects that the total amount of available wave power in the area will be very similar in the future, with a slight decrease for scenario RCP4.5 (-1.8%) and practically unchanged for RCP8.5 (-0.1%). Although the spatial distribution of the energy will have a similar pattern, with the central part of the study area (points P7 to P11) being the most energetic, the variations will have a different sign in different sections of this coast. In the western part, the wave power will increase for both scenarios (up to 2% for RCP4.5 and up to 8% for RCP8.5), while in the eastern part, it will decrease (up to -4% for RCP4.5 and up to -6% for RCP8.5). One remarkable feature of these changes is that the most energetic points currently will experience the greatest reductions in wave power in both future scenarios.
Concerning the directional distribution of the energy, it is mainly concentrated in two sectors, the first comprising directions between NE and E and the second between WSW to WNW. The projections suggest that the directional distribution of wave power will be practically the same in the future for both scenarios. Concerning the SV index ( Figure 14), in the RCP4.5 scenario it increases at all points except at both ends, which show a slight decrease (around 3%). The increases are particularly significant in the central part of the studied stretch (points P5 to P9) and in P12, with variations between 14% and 28%, although the maximum value of SV for this scenario is 0.6, indicating a low seasonal variability.
As for MV, the changes in the SV values are larger for the RCP8.5 scenario revealing great variations with respect to present values. The increases range between 19% and 63%, indicating a larger seasonal variability. In any case, the maximum value of SV is 0.71, which denotes a reduced variability of wave power throughout the seasons.
A test of hypothesis was carried out for both parameters MV and SV with a level of significance α = 0.05. In both cases and for both scenarios the null hypothesis was rejected, indicating that these changes are statistically significant.
Therefore, the intra-annual variability of wave power will increase significantly in the future, in particular for the RCP8.5 scenario. Nevertheless, since in the present situation the distribution of wave power throughout the year is fairly regular, such variability will be limited and considerably smaller than in other areas of the Mediterranean Sea, which present a strong seasonal character.

Summary and Conclusions
In this paper, the potential wave power in the Mediterranean coast of Morocco was assessed for present (period 1986-2005) and future (2081-2100) conditions, considering two climate change scenarios (RCP4.5 and RCP8.5). The aim was to assess how this resource will be modified by climate change. For this, wave datasets from four European institutes generated during the Med-CORDEX project, and fourteen grid points located along the Moroccan Mediterranean coast were used.
The multimodel ensemble projects that the total amount of available wave power in the area will be very similar in the future, with a slight decrease for scenario RCP4.5 (−1.8%) and practically unchanged for RCP8.5 (−0.1%). Although the spatial distribution of the energy will have a similar pattern, with the central part of the study area (points P7 to P11) being the most energetic, the variations will have a different sign in different sections of this coast. In the western part, the wave power will increase for both scenarios (up to 2% for RCP4.5 and up to 8% for RCP8.5), while in the eastern part, it will decrease (up to −4% for RCP4.5 and up to −6% for RCP8.5). One remarkable feature of these changes is that the most energetic points currently will experience the greatest reductions in wave power in both future scenarios.
Concerning the directional distribution of the energy, it is mainly concentrated in two sectors, the first comprising directions between NE and E and the second between WSW to WNW. The projections suggest that the directional distribution of wave power will be practically the same in the future for both scenarios.
Concerning the seasonal distribution of wave power, the multimodel ensemble shows a future increase in spring and autumn, limited in RCP4.5 (less than 2%) but significant in RCP8.5 (14% and 4.5%, respectively). On the contrary, a decrease of wave power is projected for summer (−1.8% in RCP4.5 and −4.9% in RCP8.5), and also for winter a remarkable reduction of wave power is calculated (−9.6% in RCP4.5 and −18% in RCP8.5).
These changes in the seasonal distribution of the wave power are also reflected in a consistent increase of the temporal variability, with greater values for coefficients SV and MV due to the increase of wave power in the most energetic season (spring) and its decrease in the less energetic season (summer). Irrespective of the increase of these indices, their magnitudes indicate a fairly even distribution of the wave energy throughout the year under future conditions. Therefore, the Moroccan Mediterranean coast offers a modest wave energy potential, which will not experience large changes in magnitude in the future due to climate change, and although the intra-annual variability of the resource will increase, it will not be as large as in other areas of the Mediterranean Sea. This will allow extracting wave energy in an efficient way and without significant changes during this century, assuming that the cost-benefit ratio of WECs will be reduced in the future, making WEC deployment economically profitable.