Renewable Energy 181 (2022) 616e629

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Wave energy flux in the Caribbean Sea: Trends and variability

Andre
n a, *, Juan C. Restrepo a, Alex Correa-Metrio b,

s F. Orejarena-Rondo
c
Alejandro Orfila
a
Grupo de Investigacio
n en Geociencias GEO4, Departamento de Física y Geociencias, Universidad Del Norte, km 5 Vía Puerto Colombia, Barranquilla,
Colombia
b
noma de M

Instituto de Geología, Universidad Nacional Auto exico, Ciudad de M
exico, Mexico
c
neo de Estudios Avanzados, IMEDEA (CSIC-UIB), 07190 Esporles, Spain
Instituto Mediterra

a r t i c l e i n f o a b s t r a c t

Article history: Wave Energy Flux (WEF) in the Colombian Caribbean Sea is assessed using a 60 years wave reanalysis
Received 20 November 2020 from 1958 to 2017. Significant wave height (Hs) and mean wave energy period (Tm_1,0) are validated
Received in revised form against in situ buoy data from the National Oceanic and Atmospheric Administration and from the
14 September 2021
Direccio
n General Marítima in the Central Caribbean Sea and in front of the Colombian shelf. WEF is
Accepted 20 September 2021
calculated at virtual buoys allocated in the central Caribbean, at the continental shelf and around insular
Available online 23 September 2021
territories revealing a long-term decrease in magnitude with an annual rate between 0.01 and 0.2% with
different regional behavior. In addition, WEF variations at interannual scale are found to be associated
Keywords:
Wave energy flux
with ENSO phases. Wave energy power decreased from 1989 to 2017 in all analyzed locations.
Interannual variability © 2021 Elsevier Ltd. All rights reserved.
ENSO
Caribbean sea

1. Introduction temperature gradients, tidal amplitude and variability, wave po-

tential, wind, etc.). In this sense, several studies have been devel-
Waves propagating across the ocean transport accumulate en- oped worldwide trying to characterize coastal areas as potential
ergy gained from winds blowing over the surface [1,2]. Some of the wave effective energy supply zones [7e10].
energy is lost in the upper ocean by turbulent mechanisms but Since 1950 wave hindcast has improved notably thanks to ad-
most of it is dissipated in a narrow zone near the coast known as vances in the representation of the different physical processes
surf zone. Wave breaking in the surf zone induces coastal dynamics involved in wind wave interaction and wave propagation using
where intense currents transport sediments towards the coast or phase averaged models as well as the increase of observations
seawards as well as along the shore inducing continuous (either in situ or remote) [1,7,10,11]. Nowadays it is relatively
morphological changes [3]. Several studies highlighted the ex- affordable to obtain hindcast and forecast of waves during large
pected significant morphological changes in the forthcoming years periods using energy balance models with emphasis at local scales
due to the combined effects of waves and sea level rise, which will or in particularly sensible coastal areas.
have large impacts worldwide [4e6]. Variations in significant wave height (Hs), mean (T02) and peak
In the last decades large efforts have been devoted in obtaining (Tp) periods and therefore wave energy flux (WEF) have been re-
new and efficient methods of energy extraction from renewable ported around the world (e.g. Refs. [12e18]. Young et al. [12]
resources and waves are an important component of them that take studied trends in wind velocity and Hs, using 23 years of global
more relevance in developing countries, insular systems and in scale satellite altimetry data. These authors found a general yearly
areas with a large vulnerability. Marine renewable energy re- increase in average wind velocity of 0.75% and in Hs of 0.50 in all the
sources will play a crucial role during the next years for the de world oceans within the 99% percentile, excepting for tropical and
carbonization of electricity which implies increasing the knowl- equatorial regions where neutral tendencies were observed.
edge of the availability of the resources (e.g., salinity and Through a reanalysis study of waves, Wu et al. [13] found positive
trends in Hs, the mean energy period (TE ¼ Tm-1,0) and WEF at mid
latitudes in the southern hemisphere (45 S and 60 S), in particular,
* Corresponding author.
along the Australian coasts and east of the South American coasts.

n).
E-mail address: orejarenaa@uninorte.edu.co (A.F. Orejarena-Rondo These authors also found negative trends of these parameters in the

https://doi.org/10.1016/j.renene.2021.09.081
0960-1481/© 2021 Elsevier Ltd. All rights reserved.

n, J.C. Restrepo, A. Correa-Metrio et al.
A.F. Orejarena-Rondo Renewable Energy 181 (2022) 616e629

Eastern Tropical Pacific and the west coast of North America. During the wet season, the ITCZ migrates toward 10 e12 N,
Extreme wave climate variations along Central and South America allowing the less intense south trade winds to reach the Colombian
exhibit positive trends in Hs for the Pacific coastal zone from Cen- Caribbean basin, interacting with the northern trade winds. This
tral America to Patagonia, and for the Atlantic coastal zone in the interaction, together with a high content of water vapor, generates
Southern Hemisphere. At the same time, coastal zones of the North atmospheric instability and fosters the formation of hurricanes
Atlantic and Caribbean Sea exhibited negative trends in wave within the western Caribbean [33]. On average, eight hurricanes
height [24]. pass across or near the Caribbean between June and November,
Reguero et al. [14], analyzed wave climate variability in the with a peak of maximum activity in September [35]. Along the
coastal zone of Central and South America, using the Wavewatch III Colombian coast, wave climate is modulated by seasonal variability
model with a global reanalysis extended along 60 years. Their re- of wind regimes, associated with cold front passing, hurricane
sults showed variations in energy flux direction and positive trends incidence, and coastline configuration [33,34,36]. In general terms,
in Hs for almost all the Pacific and Atlantic coastal zones, whereas in wave climate exhibits a bimodal distribution associated with cli-
the Caribbean Sea the coastal zones displayed positive trends for matic seasons, with low and high Hs during the wet and dry sea-
wave Hs in the Greater Antilles and negative trends for northern sons, respectively [37,38]. In addition, extreme waves do not exhibit
South America, Central America, and part of the Gulf of Mexico. the same magnitude and patterns along the Colombian coast due to
Reguero et al. [15] and Reguero et al. [19] also characterized WEF at the effect of hurricanes and cold fronts over specific zones of the
a global scale to determine its monthly and seasonal variability, as coast.
well as its long term variations. These works found an 0.4% yearly Extreme wave heights are observed in the Guajira Peninsula,
increase of WEF at the global scale. In the long term, this increase mainly associated with the passage of hurricanes, with values be-
was related to variations in ocean surface temperature caused by tween 3.0 and 5.5 m for return periods of 5e100 years [36]. In the
global warming. western zone (Gulf of Morrosquillo and Gulf of Urab
a) extreme
Long term variations in both extreme and mean Hs, and to a wave heights are associated with cold fronts, but they exhibit a
lesser extent in TE, and Tp, have been also reported at the local and lower magnitude due to coastal configuration, with values between
regional scale, with contrasting results (e.g. Refs. [1,20e26]. 4.0 and 5.0 m for return periods between 5 and 100 years. Lastly, in
Nevertheless, at the local scale, WEF has not been satisfactorily the Archipelago of San Andre
s, Providencia and Santa Catalina the
analyzed in the long term, and since it is an integrated magnitude it dynamics are different from those of the continental zone. In this
may exhibit variations not reflected by parameters such as Hs or TE region, extreme wave heights are similar in magnitude to those at
[16]. In addition, some questions remain in relation to synchrony the Guajira Peninsula, mainly associated with cold front passing,
and scale of these variations, as well as on the effect of local scale with values between 4.0 and 5.0 m for return periods of 5e100
processes in energy flux variations. In this sense, to date, in the years [36].
Colombian Caribbean Sea, there is a lack of studies regarding the Climate modes of variability, such as ENSO, also influence the
long term (i.e. > 40 years) variability of wave fields. Previous work variability of waves in the Caribbean. Enfield and Mayer [39]
in the region include WEF characterizations in front of the coasts of determined increases in ocean surface temperature in the tropical
Barranquilla and La Guajira [27], in the continental portion of the Atlantic, related to the warm phase of ENSO, leading to NE trade
Colombian Caribbean Sea [28], in the Gulf of Mexico and the wind intensity reduction. According to Osorio et al. [35], this
Caribbean [11], and in the development of a methodology to decrease in wind intensity is reflected in wave height decrease from
calculate WEF in shallow waters [29]. All these works however have December to late February, as well as during March, April, and May,
been focused on the analysis of the wave energy but not in its when the greatest wave heights are observed during the cold phase
variability at the long term. (La Nin ~ a). In June, July and August, this pattern is reversed with
In this work, we analyze the long term WEF variability in the larger and smaller wave height magnitudes during El Nin ~ o and La
Colombian Caribbean coast during the last 60 years (1958e2017). In Nin~ a, respectively.
particular, we seek to identify the magnitude, scale and time extent
of these changes analyzing the existence of regional patterns along 3. Data and methods
the Colombian Caribbean coast, including its insular territories and
the role of the ENSO phenomenon in the inter-annual variability of Given that data from in situ buoys, or satellite altimetry for the
the WEF. study area are missing information critical for the proposed ana-
lyses, simulated time series are used instead. A 60-year wave
2. Study area hindcast (from 1958 to 2017) was performed in the Caribbean Sea
from which 47 virtual buoys on the continental and insular coasts of
The Caribbean basin is bordered by South America, the Lesser Colombian and Central Caribbean Sea were extracted to determine
Antilles, the Greater Antilles, and the eastern portion of Central long term changes in WEF. The simulation was performed using
America (Fig. 1). Colombia has a 2445-km-long coastline along the SWAN model [40], forced with the JRA-55 wind reanalysis [41].
Caribbean, with a predominant SW-NE orientation [30]. The Simulated wave data were validated with one NOAA buoy and three
regional climate is mainly dominated by low pressure belt dy- buoys from DIMAR. Time series were used to determine wave en-
namics formed by the Intertropical Convergence Zone (ITCZ) with ergy flux for each virtual buoy.
two seasons: the dry, or windy, season extending from December to
April and the wet season from August to November [31]. The 3.1. Wind data
remainder of the year from May to June is a transition season [32].
The ITCZ migrates towards 0 e5 S during the dry season, allowing Wind data are part of the JRA-55 database, developed to pro-
northern trade winds to prevail in the area, with velocities up to duce a suitable global set of atmospheric data to study multidecadal
15 m s1 [33]. January and February are characterized by frequent variability and climate change [68]. This database extends from
influence of polar and cold fronts that reach the Caribbean Sea, 1958 to the present day. According to Kobayashi et al. [41], this is
affecting the regional climate [34]. Trade winds over the Colombian the first complete reanalysis that covers the last half-century and
Basin become weaker in the transition season, increasing again the first to apply a four-dimensional variational analysis. This
from July to August, a time locally known as Veranillo de San Juan. database provides the best spatial and temporal resolution (every
617

n, J.C. Restrepo, A. Correa-Metrio et al.
A.F. Orejarena-Rondo Renewable Energy 181 (2022) 616e629

Fig. 1. SWAN domain used for simulating 60 years of waves in the Caribbean basin. Red triangles indicate the location of NOAA (A) and DIMAR (B, C, D) buoys used for model
validation. Dots show the location of virtual buoys where 60 years of waves were simulated for the study.

3 h) among all available wind reanalysis databases. Wind data were PN

obtained from the global grid, at 10 m in height and every 3 h, from n¼1 ðPn  On Þ 2
D ¼ PN
January 1 of 1958 to December 31 of 2017. The global grid has
n¼1 ðjPn  Oj þ jOn  OjÞ2
640  320 grid points and 0.5625  0.5616 resolution. This global
grid was cropped to 23  43 grid points that cover our study area, Whereas the bias index was calculated as:
preserving the original resolution. Coordinates for the study area
PN
are 84.3746 W, 7.5819 N (lower left corner) and 60.1871W, n¼1 ðPn  On Þ
19.9371N (upper right corner). biasðPÞ ¼ PN
n¼1 On

where N is the number of evaluated values, Pn is the predicted

3.2. SWAN model values, On is the observed data and O is the mean values of the data.
Willmott's index equals 0 indicates complete disagreement and
The model domain covers 2600  1175 km2 in a 229  101 mesh equals 1 indicates a perfect agreement. The bias index equals to
(Fig. 1). Bathymetry was obtained from Colombian Hydrographic 0 indicates perfect agreement. A bias of 0.05 indicates that the
Service nautical charts as well as from the general ocean bathy- predicted data underestimated the observed data by 5% (on
metric chart of GEBCO in the study area. Waves were simulated in a average).
non-stationary mode using the JRA-55. Wave growth due to wind The Pearson correlation was calculated as:
was configured as exponential, following the formulation of Komen
et al. [42], and nonlinear deep water interactions following the covðX; YÞ
rx;y ¼
Webb-Resio-Tracy method. JONSWAP spectrum was configured sX sy
with a gamma of 1.37. For the spectral directional resolution, the
number of the circle subdivisions was 120, whereas the range of Where cov is the covariance sX is the standard deviation of X and sy
frequencies was from 0.05 to 0.585 Hz. Other relevant processes, is the standard deviation of Y.
such as wave breaking, whitecapping energy dissipation, and bot-
tom friction were included in the simulations. For the validation, 3.3. Energy flux calculation and database analysis
the parameters of whitecapping were adjusted; the rate of white-
capping dissipation (Cds) was set as 3.18 $105 and the value of the For each of the 47 virtual buoys (Fig. 1) WEF are computed as:
wave steepness for the Pierson-Moskowitz spectrum (S^2 p.m.) as
5.02 $103. We use Collins bottom friction coefficient (cfw) as 0.01 WEF ¼ E:cg ¼
1
rg2 Tm1;0 Hs2
and the breaking constant with the default values. Time step was 64p
set as 30 min and outputs for Hs, Tm-1,0, and mean direction (qm)
recorded every 3 h. where E is energy density in mJ 2 , r the density of water (taken
Simulated waves were validated through comparison with in constant as 1023.6 kg/m3), g the acceleration of gravity. Tm1;0 (in
situ wave data taken by NOAA and DIMAR buoys, located in the s), and Hs ðin m) are obtained in all the virtual points from the wave
Central Caribbean and in front of the continental portion of the hindcast following Ponce de Leon et al. [8], and Mentaschi et al., [5].
Colombian Caribbean. For buoy NOAA-42058 H and Tm-1,0 vali- Principal component analysis (PCA) [44] was applied in order to
dation is made for 2012. For buoys -DIMAR TAS1450 and 41193 determine the behavior and temporal variation of WEF along the
comparison is made from January to March and from October to Colombian Caribbean coast. Long-term variations of WEF were
December of 2009, respectively. Lastly, for buoy TAS1980 compar- determined for each buoy, using detrended and moving window
ison is made for November and December of 2010. Error parame- methods. The detrended method extracts the natural trend of the
ters between buoys and SWAN results were calculated using series with a smoothing spline, with a 0.5 frequency response at a
Willmott's index D, the average bias(P) index and the Pearson 0.67 wavelength per series length in years [45]. Moving windows
correlation [43]. Willmott's index was calculated as: allow us to determine trends by using shape and scale parameters

618

n, J.C. Restrepo, A. Correa-Metrio et al.
A.F. Orejarena-Rondo Renewable Energy 181 (2022) 616e629

obtained from fitting a probability density function (PDF) to wave between in situ data and model data is obtained for both Hs and Tm-
energy flux data (e.g. Ref. [17]; Pierini et al., 2017). A 20-year time 1.0.The bias, correlation, and Willmot coefficients for Hs are 0.064,
window (240 month) was applied month to month, to determine 0.83, and 0.9, respectively and for Tm-1.0 of 0.089, 0.74, and 0.77,
the long term variability of the WEF. This window length was respectively. These relationships are statistically significant at a 95%
chosen to incorporate macro climatic phenomena at different confidence level for a two-tailed student's t distribution (p-value
scales, from quasi-biennial to cuasi-decadal, that influence wave <105).
climate in the Caribbean Sea [19,32,46]. Different PDF functions Validation of Hs in the continental portion of the Colombian
(normal, generalized extreme values (GEV), gamma, Weibull, coastal area, using the DIMAR buoys located in front of La Guajira,
extreme value, log normal, and Rayleight) were fitted to the time
is provided in Fig. 3. In all cases, a
Barranquilla, and de Gulf of Uraba
series of WEF obtaining that the best fit was given by the Weibull good fit was found between Hs values measured in situ and Hs
function. Once the fitting function was determined, parameters estimated by the model. Correlation coefficient values for buoys in
controlling the Weibull function distribution (scale (a) and shape La Guajira and Barranquilla were 0.82 and 0.85, respectively. For the
(bÞ parameters) were stored for long term variation analysis. Scale Gulf of Uraba
buoy the value was 0.76. Willmot coefficient values
parameter a, indicating how acute or flat the function may be, were above 0.8 in all cases, whereas bias coefficients for buoys in La
shows variations of WEF in time. The shape parameter b shows Guajira, Barranquilla and Gulf of Urab
a had values of 0.041, 0.17,
variations in skewness of the PDF function. b values between 3 and and 0.019, respectively.
4 indicate that the PDF function is nearly symmetrical, decreasing b
values indicate a left-skewed function (positive asymmetry),
4.2. Wave energy flux - PCA
whereas a right-skewed function would imply increasing values
[47,67]. Lastly, the annual distribution of wave energy was deter-
The first two principal components (PC1 and PC2) of the WEF
mined in terms of Hs and Tm-1,0 with combined dispersion plots
explain the 77.63% of variability for the analysis of mean and
[7e9,48], taking into account energy flux variations, at different
standard deviation (Fig. 4). Vectors representing mean WEF are
points of the Colombian Caribbean coast. In these images the total
mainly aligned with the positive side of PC1, that explains 50.94% of
energy contribution given per each sea state is calculated by
the variance, implying this component is representative of average
grouping model outputs every 3 h in bins of Hs ¼ 0.1 m and Tm-
energy. Vectors representative of standard deviation of WEF (intra-
1.0 ¼ 0.1 s. Contour lines of constant WEF are calculated applying
monthly variability) are mostly aligned with the negative end of
Equation 3.
PC2, which explains 26.69% of variance. In Fig. 4a, the gray points
are the data associated with the mean (blue) and standard devia-
4. Results tion (red) for the 720 months.
Correlations between different locations are given by the angles
4.1. Model validation between vectors; vectors with the same direction correspond to
positive correlation; opposite directions (angles of 180 ) to nega-
Validation of Hs and Tm-1.0 for 2012 in buoy NOAA-42058, tive correlation and angles of 90 representing uncorrelated data
located in the Central Caribbean is given in Fig. 2. A good fit (Fig. 4a). The results evidence a group of locations with a similar

Fig. 2. Validation of wave Hs (a) and Tm-1,0 (b) in the Central Caribbean (see buoy A in Fig. 1), with their respective dispersion diagrams and associated statistics. Color scale of
dispersion diagrams corresponds to the associated density of data used for validation.

619

n, J.C. Restrepo, A. Correa-Metrio et al.
A.F. Orejarena-Rondo Renewable Energy 181 (2022) 616e629

(see buoy D in Fig. 1), with their respective dispersion diagrams and
Fig. 3. Wave Hs in (a) La Guajira (see buoy B in Fig. 1), (b) Barranquilla (see buoy C in Fig. 1), and (c) Gulf of Uraba
associated statistics. Color scale in dispersion diagrams indicates the associated density of data used for validation.

behavior in the mean WEF, grouped in the highest values of the PC1. Fig. 4c shows that, in the long term, PC2 decreases up to 1.52 in
These buoys located in the northern and central part of the 1998, indicating an increase and maximum in wave intra-monthly
Colombian Caribbean present larger values of WEF followed by variability. From this date, PC2 increases (and intra-monthly vari-
locations in front the Gulf of Morrosquillo, the Central Caribbean, ability decreases) up to 1.0, remaining stable until 2017. In addition,
the Gulf of Urab
a and the Northern Cays (Serrana, Serranilla, Bajo Fig. 4c also shows the influence of interannual climate oscillations
Nuevo. See Fig. 1) which presents the lower values of mean WEF. (i.e. ~3e~7 years) throughout the series, being in some cases more
Overall, PC1 scores are directly related to mean WEF and, thus, pronounced.
observation scores represent energy through time (Fig. 4b).
Fig. 4b and c shows the temporal variation of PC1 and PC2 from 4.3. Wave energy flux - detrended and moving windows
1958 to 2017 with values ranging from approximately 0.7 to 1.
WEF increased from 1958 to a maximum value of 0.7 in 1980 and The detrended analysis for all 47 buoys show that WEF
decreased from this date until reaching the negative range of PC1 decreased between 1958 and 2017 with different geographical
until 2017 (Fig. 4b). In addition, the mean WEF presents large scale behavior (Fig. 5). Examples of WEF at different locations on the
oscillations (between 3 and 7 years) with a varying amplitude. Caribbean Sea, are shown in plots a, b, c, d, and e in the upper panel
Regarding the WEF standard deviation, the correlation among of Fig. 5. Even though these figures show that the decrease in WEF
vectors indicates that most of the buoys in the Colombian Carib- starts approximately in 1980 for all the buoys, the magnitude varies
bean coast present a high intra-monthly variability (large vectors), for the different zones being larger for virtual buoys located in front
whereas buoys in the Central Caribbean (opposite quadrant) show a of the insular zone (San Andres and Providencia) and lower for
lower intra-monthly variability (Fig. 4a). In summary, wave intra- virtual buoys in front of Santa Marta, Barranquilla and La Guajira.
monthly variability is negatively associated with PC2, with posi- Fig. 6 shows the behavior of mean WEF according to the Weibull
tive scores pointing to low variability and vice versa. The value distribution, adjusted to 20 year windows for buoys #5 (a), #19 (b),
range of Fig. 4c exhibits values from approximately 1.2 to 1.6. #29 (c) and #33 (d), located in front of La Guajira, Barranquilla, the
620

n, J.C. Restrepo, A. Correa-Metrio et al.
A.F. Orejarena-Rondo Renewable Energy 181 (2022) 616e629

Fig. 4. (a) Principal Component Analysis (PCA) for the mean and standard deviation of wave energy flux in time. Vectors with names in blue represent the behavior of mean wave
energy flux for each buoy. Vectors with names in red represent wave energy flux standard deviation for each buoy. (b) Principal component 1 through time, representing mean wave
energy flux for each analyzed time step (c) Principal component 2 through time, representing wave energy flux standard deviation for each analyzed time step.

621

n, J.C. Restrepo, A. Correa-Metrio et al.
A.F. Orejarena-Rondo Renewable Energy 181 (2022) 616e629

Fig. 5. WEF trends (in W/m) as determined by detrended analysis for each location on the Colombian Caribbean coast. Plots a, b, c, d, and e are representative examples of WEF
behavior at different locations in the Colombian Caribbean Sea.

Fig. 6. Weibull distribution for 480 different WEF time bins (20 years) in buoys (a) 5, (b) 19, (c) 29, and (d) 33. Each color represents sixty different 20-year periods that cover a total
of 5 years starting in January of 1978.

622

n, J.C. Restrepo, A. Correa-Metrio et al.
A.F. Orejarena-Rondo Renewable Energy 181 (2022) 616e629

Gulf of Urab
a, and San Andres, respectively. To visualize the func- 43e44, located in front of the Gulf of Urab
a and Cayos de Bajo
tion behavior, we used different colors for every 60 displacements Nuevo island, respectively, with decreases between 0.8 and 1.8%.
of the 20-year window, corresponding to the total coverage of 5 The evolution of dimensionless shape parameter b of the Wei-
years, starting in January of 1978 (see caption in Fig. 6). Results in bull function for the 47 virtual buoys is shown in Fig. 8 where
front of La Guajira, Barranquilla, and San Andres show that the panels a, b, c, d, and e, correspond to buoys #5, #19, #29 and #33,
Weibull distribution had an initial displacement towards the left respectively. These examples show that the shape parameter b
(i.e. negative asymmetry), evidencing that WEF increased until the increases until 1990, stabilizing during a 5-year period and finally
period between 1988 and 1992. From that period, the function decreasing until 2017. In addition, the magnitude of this parameter
starts displacing to the left (i.e. positive asymmetry), which in- varies in space, ranging from 1.7 to 3.4.
dicates that WEF decreases up to 2017. Results in front of the Gulf of
Uraba
show that the Weibull function starts to move slightly to the 4.4. Variations in annual total energy contribution in the Caribbean
left from the 1993e1997 period until 2017, indicating a slight WEF
decrease in this zone. Virtual buoys located in front of the continental zone (Fig. 9)
From the analysis of moving windows, the a scale parameters show, for both time periods assessed, energy values concentrated
were obtained for each PDF function fitted to all 480 different 20- between 4.0 and 5.5, with wave heights between 1.5 and 3.0 m.
year periods in each of the virtual buoys in the Caribbean. Based Buoys located in the insular zone of the Caribbean (Fig. 10) show
on these results we estimate the percentage of change in WEF (W/ that, in both time periods assessed, energy values are concentrated
m) for the 60 analyzed (Fig. 7a and b). The highest decrease in WEF approximately between 4.5 and 6.5 s, with wave heights between
were found for buoys 1 to 21, located in front of La Guajira, Santa 2.0 m and 3.5 m. Observed energy peaks evidence changes in the
Marta, and Barranquilla, as well as in buoys 31 to 34, in front of San period of WEF decrease (1989e2017) in all locations (Fig. 9 a’, b’, c’,
Andre
s island. Both buoys showed decrease values ranging from 9.0 d’, e’, and f’). For example: during the initial period (1958e1987), in
to 11.7%. This decrease is followed by changes experienced in buoys the continental zone, Fig. 9a and b shows that energy peaks
22e24, 30, 35e40, and buoys 46e47, located in front of Cartagena, concentrated between 4.9 and 5.3 s and wave heights between 2.1
Gulf of Morrosquillo, Gulf of Uraba
, Providencia island, Cayos de and 2.5 m. During the WEF decrease period (1989e2017), Fig. 9a’
Serrana island, and the Central Caribbean, respectively. This group and 9c0 energy peaks were concentrated at periods between 4.9
showed a decrease between 4.8 and 8.7%. Buoys 25e26, 29, 4e42 and 5.0 s and wave heights between 2.1 and 2.2 m. The buoys
and 45, located in front of the Gulf of Morrosquillo, the Gulf of located in the insular zone show the same behavior. During the
Uraba
, Cayos de Serranilla Island, and the Central Caribbean, initial period (1958e1987), Fig. 10d and e shows that the energy
respectively, showed WEF decreases between 2.1 and 3.3%. Lastly, peaks concentrated between 5.5 y 5.9 s and wave heights between
the lowest WEF changes were found in buoys 27e28, and buoys 2.4 and 2.7 m. At the same time, during the 1989e2017 period,

Fig. 7. (a) Decrease in WEF along the Colombian and Central Caribbean, obtained from the scale parameter (a). (b) Stem plot of wave energy flux indicating the percentage decrease
for each point of interest.

623

n, J.C. Restrepo, A. Correa-Metrio et al.
A.F. Orejarena-Rondo Renewable Energy 181 (2022) 616e629

Fig. 8. Evolution of the dimensionless shape parameter for all buoys located in the Caribbean coast and Central Caribbean. Upper panels a, b, c, d, and e represent different examples
of the behavior of this shape parameter in time corresponding to buoys #5, #19, #29 and #33, respectively.

Fig. 10d’ and 10e’ show energy peaks with periods concentrated decrease in mean WEF and increased intra-monthly variability. In
from 5.6 to 5.7 s and wave heights between 2.5 and 2.6 m. These contrast, in the 1967 La Nin ~ a event, mean WEF increased and
results indicate that the annual energy contribution decreased in standard deviation decreased. This link may be related to sea sur-
the Caribbean between 1989 and 2017. face temperature (SST) anomalies in the Equatorial Pacific and
Tropical Atlantic induced by ENSO (e.g. Ref. [39]; and [51]. Using
5. Discussion observational methods, previous works determined an SST increase
in the northern Tropical Atlantic in the boreal spring, posterior to El
Simulated Hs and Tm_1,0 values are used to determine WEF Nin~ o peaks, related to the weakening of the Northeast trade winds.
variability. The model was validated by in situ data from buoys of These works also determined an opposite behavior during La Nin ~a
the NOAA and DIMAR (Figs. 2 and 3). Correlation, bias and Weibull events [39,52,53]; and [51]. An SST increase in the northern Tropical
coefficients are all indicative of a robust fit to data, showing sta- Atlantic, forced by NE trade wind weakening over the Atlantic basin
tistically significant values for a confidence level of 95% (Figs. 2 and during the boreal spring, was also observed in the atmospheric
3). In contrast with previous wave reanalysis in the Colombian models of Huang et al. [54]. These observations agree with the re-
Caribbean (e.g. Refs. [35,49], this study is further extended in time, sults of this work and those of Osorio et al. [35]. In particular, it has
encompassing the last 60 years (1958e2017) [50]. been found that, during the warm phase of ENSO, wave heights in
The PCA analysis allowed reducing the dataset, represented by the Caribbean decrease from December to May. This situation re-
the number of buoys, mean and standard deviation of WEF for the verses with the increase in wave heights during the cold phase of
60-year study period. PCA variable reduction allowed assessing the ENSO in the same period. Similar results were obtained by Vega
monthly behavior of WEF mean and standard deviation for each et al. [55], who established that average extreme waves were
buoy located in the Caribbean Sea, using vector directions relative greater during La Nin ~ a, whereas the lowest average extreme waves
to Principal Components and their correlation according to the occurred during El Nin ~ o. Between 1968 and 1980 the effect of ENSO
angles among them (Fig. 4a). PCA also provided a simplified way to on the signal of WEF mean and standard variation was not clearly
observe WEF (mean and standard deviation) long-term behavior distinguished (Fig. 11a and b). The lack of a clear incidence in this
for all buoys (Fig. 4b and c). WEF mean and standard deviation period may be related to the percentage (77.63%) of data explained
values show 3-to-7-year periodic oscillations along the series by the first two Principal Components. PCA has been widely applied
(Fig. 4b and c). Comparison between Oceanic Nin ~ o Index (ONI) to studies of coastal morphology (e.g Refs. [56e59], and studies of
monthly series, and WEF mean and standard deviation values the interaction between fluvial and oceanographic parameters
(Fig. 11), shows that WEF mean decreases whereas intra-monthly [60e62], among others. Nevertheless, to our knowledge, this is the
variability gradually increased as temperatures in the Tropical Pa- first time that PCA is used to analyze a wave database, allowing a
cific became higher (ENSO warm phase). On the other side, as the simplified observation of the intra-annual and long-term patterns
Tropical Pacific temperature lowered (ENSO cold phase), mean WEF of WEF.
increases and WEF standard variation gradually decreases. This Long-term WEF decrease in the Colombian Caribbean (Fig. 4 b)
influence is present along the time series and can be observed, for may be related to global scale variations in SST and wind temper-
example, in the 1966 El Nin ~ o event. This event coincided with a ature associated to global warming [16]. Following Zheng et al. [63],

624

n, J.C. Restrepo, A. Correa-Metrio et al.
A.F. Orejarena-Rondo Renewable Energy 181 (2022) 616e629

Fig. 9. Comparisons between wave annual energy contributions in front of the continental zone of the Colombian Caribbean in the period from 1958 to 1987 (a, b, c) and from 1989

). Colors indicate wave annual energy distributions per meter of wave front in (kWh/(m*year)).
to 2017 (a’, b’, c’) in buoys 5 (La Guajira), 19 (Barranquilla), and 29 (Gulf of Uraba
Dashed lines correspond to constant energy flux. Numbers represent average occurrences in hours per year. Locations of each virtual point and their respective wave rose diagrams
are represented in the upper left and center parts of the Figure.

wind speed in most of the ocean started to show significant positive regional differences in their trend. Furthermore, these studies could
trends from 1988. Only some seas showed significant positive not identify the starting point of long term changes in WEF, or the
trends from that date. The PCA analysis results (Fig. 4 b) showed a influence of ENSO on inter-annual WEF variations in the Colombian
turning point in 1980, after which mean WEF decreased almost Caribbean (Fig. 11).
constantly until 1998, when it stabilized and maintained almost Based on the results, detrended analysis of WEF variation shows
constant until 2017. These long-term variations were also reported that WEF did not vary in the same way, or with the same magni-
by Reguero et al. [14] and Reguero et al. [19]. Global scale results of tude, in all locations analyzed in the Caribbean Sea (Fig. 5). This
these authors showed negative trends for the Caribbean basin in pattern may be reflecting the diverse locations of virtual buoys
annual mean Hs and WEF. Nevertheless, these studies did not show along continental and insular coastal zones in the Caribbean. The
in a detailed way the variations in energy flux along the continental buoys more influenced by NE trade winds exhibited a greater en-
and insular coastal zones of the Colombian Caribbean, as well as ergy flux as shown in Appendini et al. [11]. We also observed that

625

n, J.C. Restrepo, A. Correa-Metrio et al.
A.F. Orejarena-Rondo Renewable Energy 181 (2022) 616e629

Fig. 10. Comparisons between wave annual energy contributions in front of the insular zone of the Colombian Caribbean in the periods 1958e1987 (d, e, f) and 1989e2017 (d’, e’, f’)
in buoys 33 (San Andres island), 37 (Providencia island), 40 (Cayos de Serrana island). Colors indicate wave annual energy distributions per meter of wave front in (kWh/(m*year)).
Dashed lines correspond to constant energy flux. Numbers represent average occurrences in hours per year. Locations of each virtual point and their respective wave rose diagrams
are represented in the upper left and center parts of the Figure.

the turning point where energy flux decreased for most buoys in behavior can be observed. The first one is in La Guajira, with WEF
1980, appeared in 1992 for buoys 25e29, located in the Gulf of decreasing between 9 and 10%. The second zone, between Santa
Morrosquillo and Gulf of Urab
a. It is possible that, due to their Marta and Barranquilla, had decreased values between 11 and 12%.
location, these buoys did not register the initial variations in WEF The third zone, between Cartagena and the Gulf of Morrosquillo,
until the magnitude of change was greater, in 1992 (Fig. 4b). In the showed WEF decreases between 7 and 8%. The fourth zone, be-
same way, WEF percent change varies in the different locations tween the Gulf of Morrosquillo and the Gulf of Uraba
, decreased
chosen for the -Caribbean, as indicated by the scale parameter (a) values between 1 and 3%. In a similar way, the insular zone can be
(Fig. 7b). Fig. 7a shows how the decrease in WEF magnitude divided into three zones with different levels of WEF decrease. The
differed along the coasts of the Colombian Caribbean during the first zone, in San Andres island, had WEF decrease values between 9
period of study. In the continental zone, four zones with different and 10%. The second zone is in Providencia island, with decrease

626

n, J.C. Restrepo, A. Correa-Metrio et al.
A.F. Orejarena-Rondo Renewable Energy 181 (2022) 616e629

Fig. 11. Comparison between (a) mean WEF variation in time, (b) intra-monthly variation given by standard deviation, and (c) the Oceanic Nin~ o Index (ONI). Red circles and lines
represent mean WEF decreases in figure (a) and intra-monthly variability increases in figure (b). Blue circles and lines represent mean WEF increases in figure (a) and decreases of
intra-monthly variability in figure (b).

values between 7 and 8%. Located in the northern Cays, the third Results of different statistical analyses indicated a long-term
zone showed WEF decreases between 1 and 3%. According to these decrease in WEF and variations in the inter-annual scale, associ-
results, for the continental zone of the Caribbean the WEF decrease ated with ENSO phases. The results showed that WEF decreased at
percentage between La Guajira and Barranquilla ranged from 0.15 an annual rate between 0.15 and 0.2% for the zone extending be-
to 0.2% per year. For the rest of the continental zone, between tween La Guajira and Barranquilla, 0.01 and 0.014% for the zone
Cartagena and the Gulf of Urab
a, WEF decrease percentage ranged between Cartagena and the Gulf of Uraba
, and between 0.016 and
between 0.01 and 0.14% per year. In the Central Caribbean and 0.14% for the insular zone and Central Caribbean. Annual wave en-
insular zone, percent WEF decrease was between 0.016 and 0.14% ergy contribution was assessed before (1958e1987) and after
per year. These results are in the same order of magnitude as those (1989e2017) the turning point of WEF. Results of this work will be
reported by Reguero et al. [16], who found WEF decreases for the used to analyze the changes in the coastal zone as the result of the
North Atlantic between 0.5 and 1.0% per year. variation of the WEF incident to the coast and to identify potential
The results of this work are also coherent with wave predictions coastal zones suitable for deploying efficient and cost effective wave
made from low wind pattern variations under the IPCC's worst energy converters in the Colombian area. Wave climate is the major
climate change scenario [64]. Casas-Prat et al. [65] using wind fields factor affecting wave energy production. A significant fraction of the
derived from climate models, simulated wave conditions between annual energy could be obtained during extreme events but such
2081 and 2100, under the RCP 8.5 climate change scenario. Their conditions pose serious engineering challenges increasing he costs
projections showed a decrease in wave height for the North in the development and maintenance of wave energy converters. On
Atlantic, including the Caribbean Sea. These changes were consis- the other hand, in semi enclosed seas with small fetch and relatively
tent with the respective variations in wind speed patterns. Based on moderate wave conditions such as the Caribbean Sea, many tech-
the RCP 8.5 climate change scenario, Mentaschi et al. [5] indicated nical issues related to extreme sea climate could be more easily
that, for the year 2100, WEF would experiment mostly with nega- solved if interannual extreme events are delimited and possible
tive trends in the Northern Hemisphere, including the Caribbean changes in waves due to climate change assessed.
Sea.
Data availability statement
6. Conclusions
The dataset generated for this study is available on request to
Variation in WEF was assessed in the long term for the Colom- the corresponding author.
bian Caribbean coast and Central Caribbean, based on a 60-year
reanalysis (1958e2017) made with the SWAN wave model. Using Funding
Hs and Tm_1,0, both simulated and in situ validated, wave energy flux
values for analysis were obtained for different locations of the This work was supported by COLCIENCIAS (Departamento
Colombian Caribbean and Central Caribbean.
n), convocatoria
Administrativo de Ciencia, Tecnología e Innovacio
627

n, J.C. Restrepo, A. Correa-Metrio et al.
A.F. Orejarena-Rondo Renewable Energy 181 (2022) 616e629

727. A. Orfila would like to thank financial support from Ministerio [14] B.G. Reguero, F.J. Me
ndez, I.J. Losada, Variability of multivariate wave climate

n y Universidades of Spain through the in Latin America and the Caribbean, Global Planet. Change 100 (2013) 70e84,
de Ciencia, Innovacio
https://doi.org/10.1016/j.gloplacha.2012.09.005.
MOCCA project (grant no. RTI2018-095441-B-C21). [15] B.G. Reguero, I.J. Losada, F.J. Me
ndez, A global wave power resource and its
seasonal, interannual and long-term variability, Appl. Energy 148 (2015)
366e380, https://doi.org/10.1016/j.apenergy.2015.03.114.
CRediT authorship contribution statement
ndez, A recent increase in global wave power
[16] B.G. Reguero, I.J. Losada, F.J. Me
as a consequence of oceanic warming, Nat. Commun. 10 (2019) 205, https://
Andre
s F. Orejarena-Rondo
n: Conceptualization, Validation, doi.org/10.1038/s41467-018-08066-0.
[17] C. Izaguirre, F.J. Me
ndez, A. Espejo, I.J. Losada, B.G. Reguero, Extreme wave
Writing e original draft, Writing e review & editing, Formal anal-
climate changes in central-south America, Climatic Change 119 (2013)
ysis. Juan C. Restrepo: Conceptualization, Writing e review & 277e290, https://doi.org/10.1007/s10584-013-0712-9.
editing, Supervision. Alex Correa-Metrio: Conceptualization, [18] C. Izaguirre, F.J. Me
ndez, M. Mene
ndez, I.J. Losada, Global extreme wave
Writing e review & editing, Formal analysis. Alejandro Orfila: height variability based on satellite data, Geophys. Res. Lett. 38 (2011) L10607,
https://doi.org/10.1029/2011GL047302, 2011.
Conceptualization, Validation, Writing e review & editing, Formal [19] J.C. Restrepo, A. Higgins, J. Escobar, S. Ospino, N. Hoyos, Contribution of low-
analysis, Supervision. frequency climaticeoceanic oscillations to streamflow variability in small,
coastal rivers of the Sierra Nevada de Santa Marta (Colombia). Hydrology and
Earth System, Sci 23 (2019) 2379e2400, https://doi.org/10.5194/hess-23-
Declaration of competing interest 2379-2019.
[20] S. Bacon, D.J.T. Carter, Wave climate changes in the north atlantic and North
The authors declare that they have no known competing sea, Int. J. Climatol. 11 (1991) 45e558, https://doi.org/10.1002/
joc.3370110507.
financial interests or personal relationships that could have [21] J. Allan, P. Komar, Are ocean wave heights increasing in the eastern North
appeared to influence the work reported in this paper. Pacific? Eos 81 (2000) 561e567, https://doi.org/10.1029/EO081i047p00561-
01.
[22] W.C. Dragani, P.B. Martin, C.G. Simionato, M.I. Campos, Are wind wave heights
Acknowledgments increasing in south-eastern South American continental shelf between 32 S
and 40 S? Continent. Shelf Res. 30 (2010) 481e490, https://doi.org/10.1016/
The authors acknowledge financial support to Andres Orejarena j.csr.2010.01.002.
[23] E. Charles, D. Idier, P. Delecluse, M. De
que

, G. Le Cozannet, Climate change
under a grant from COLCIENCIAS (Convocatoria 727). We also thank impact on waves in the Bay of Biscay, France, Ocean Dynam. 62 (2012)
DIMAR (Direccio
n General Maritima) for the wave parameters 831e848, https://doi.org/10.1007/s10236-012-0534-8.
supplied for the validation of the model. [24] P. Ruggiero, Is the intensifying wave climate of the U.S. Pacific northwest
increasing flooding and erosion risk faster than sea-level rise? J. Waterw. Port,
Coast. Ocean Eng. 139 (2) (2013) 88e97, https://doi.org/10.1061/(asce)
References ww.1943-5460.0000172.
[25] S.L. McSweeney, Temporal and spatial variability of the open coast wave
[1] G. Dodet, X. Bertin, R. Taborda, Wave climate variability in the North-East climate of Victoria, Australia, Mar. Freshw. Res. 71 (3) (2019) 394e413,
Atlantic Ocean over the last six decades, Ocean Model. 31 (2010) 120e131, https://doi.org/10.1071/MF18489.
https://doi.org/10.1016/j.ocemod.2009.10.010. [26] V. Morales-M
arquez, A. Orfila, G. Simarro, M. Marcos, Extreme waves and
[2] F. Ardhuin, A. Orfila, Wind waves. In new frontiers in operational oceanog- climatic patterns of variability in the eastern North Atlantic and Mediterra-
raphy, in: E. Chassignet, A. Pascual, J. Tintore
, J. Verron (Eds.), GODAE nean basins, Ocean Sci. 16 (6) (2020) 1385e1398, https://doi.org/10.5194/os-
OceanView, Florida State University College of Medicine, 2018, pp. 393e422. 16-1385-2020.
ISBN-10: 1720549974 ISBN-13: 978-1720549970, Chap.14, https://archimer. [27] R.R. Torres, C.A. Andrade, Potential en Colombia para el Aprovechamiento de
ifremer.fr/doc/00461/57251/. la Energía no Convencional de los Oce
anos, Boletín Científico CIOH 24 (2006)
[3] V. Morales-Ma
rquez, A. Orfila, G. Simarro, L. Go
mez-Pujol, A. Alvarez-Ellacuría, 11e25, https://doi.org/10.26640/01200542.24.11_25.
D. Conti, A. Gala
n, A.F. Osorio, M. Marcos, Numerical and remote techniques [28] A.F. Osorio, S. Ortega, S. Arango-Aramburo, Assessment of the marine power
for operational beach management under storm group forcing, Nat. Hazards potential in Colombia. Renewable and Sustainable, Energy Rev. 53 (2016a)
Earth Syst. Sci. 18 (12) (2018) 3211e3223, https://doi.org/10.5194/nhess-18- 966e977, https://doi.org/10.1016/j.rser.2015.09.057.
3211-2018. [29] S. Ortega, A.F. Osorio, P. Agudelo, Estimation of the wave power resource in
[4] R.C. Winter, A. Sterl, J.W. de Vries, S.L. Weber, G. Ruessink, The effect of climate the Caribbean Sea in areas with scarce instrumentation. Case study: isla
change on extreme waves in front of the Dutch coast, Ocean Dynam. 62 (2012) Fuerte, Colombia, Renew. Energy 57 (2013) 240e248, https://doi.org/10.1016/
1139e1152, https://doi.org/10.1007/s10236-012-0551-7, 2012. j.renene.2012.11.038.
[5] L. Mentaschi, M.I. Vousdoukas, E. Voukouvalas, A. Dosio, L. Feyen, Global [30] N.G. Rangel-Buitrago, G. Anfuso, A.T. Williams, Coastal erosion along the
changes of extreme coastal wave energy fluxes triggered by intensified tele- Caribbean coast of Colombia: magnitudes, causes and management, Ocean
connection patterns, Geophys. Res. Lett. 44 (2017) 2416e2426, https:// Coast Manag. 114 (2015) 129e144, https://doi.org/10.1016/
doi.org/10.1002/2016GL072488. j.ocecoaman.2015.06.024.
[6] A.F. Orejarena-Rondo
n, J.M. Sayol, M. Marcos, L. Otero, J.C. Restrepo, [31] A. Orfila, C.P. Urbano-Latorre, J.M. Sayol, S. Gonz
alez-Montes, A. Caceres-Euse,
I. Herna
ndez-Carrasco, A. Orfila, Coastal impacts driven by sea-level rise in I. Herna
ndez-Carrasco, A.G. Mun ~ oz, On the Impact of the Caribbean counter
Cartagena de Indias, Frontiers in Marine Science 6 (2019) 614, https://doi.org/ current in the Guajira upwelling system, Frontiers in Marine Science 8 (2021)
10.3389/fmars.2019.00614. 128, https://doi.org/10.3389/fmars.2021.626823.
[7] M.F. Canals, C.G. García, On the spatial distribution of the wave energy [32] G. Poveda, La Hidroclimatología de Colombia: una síntesis desde la escala
resource in Puerto Rico and the United States Virgin Islands, Renew. Energy inter-decadal hasta la escala diurna, Rev. Acad. Colomb. Cienc. 28 (107) (2004)
136 (2019) 442e451, https://doi.org/10.1016/j.renene.2018.12.120. 201e222. ISSN:0370-3908.
[8] S. Ponce de Leon, A. Orfila, G. Simarro, Wave energy in the Balearic Sea. [33] C.A. Andrade, Oceanografía Dina
mica de la Cuenca de Colombia, Alpha Edi-
Evolution from a 29-year spectral wave hindcast, Renew. Energy 85 (2016) tores, Cartagena de Indias, Colombia, 2015.
1192e1200, https://doi.org/10.1016/j.renene.2015.07.076. [34] J.C. Ortiz-Royero, L.J. Otero, J.C. Restrepo, J. Ruiz, M. Cadena, Cold fronts in the
[9] Y. Lin, S. Dong, Z. Wang, C.G. Soares, Wave energy assessment in the China Colombian Caribbean Sea and their relationship to extreme wave events, Nat.
adjacent seas on the basis of a 20-year SWAN simulation with unstructured Hazards Earth Syst. Sci. 13 (2013) 2797e2804, https://doi.org/10.5194/nhess-
grids, Renew. Energy 136 (2019) 275e295, https://doi.org/10.1016/ 13-2797-2013.
j.renene.2019.01.011. [35] A.F. Osorio, R.D. Montoya, J.C. Ortiz, D. Pela
ez, Construction of synthetic ocean
[10] M.V.W. Cuttler, J.E. Hansen, R.J. Lowe, Seasonal and interannual variability of wave series along the Colombian Caribbean Coast: a wave climate analysis,
the wave climate at a wave energy hotspot off the southwestern coast of Appl. Ocean Res. 56 (2016b) 119e131, https://doi.org/10.1016/
Australia, Renew. Energy 146 (2020) 2337e2350, https://doi.org/10.1016/ j.apor.2016.01.004.
j.renene.2019.08.058. [36] L.J. Otero, J.C. Ortiz-Royero, J.K. Ruiz-Merchan, A.E. Higgins, S.A. Henriquez,
[11] C.M. Appendini, C.P. Urbano-Latorre, B. Figueroa, C.J. Dagua-Paz, A. Torres- Storms or cold fronts: what is really responsible for the extreme waves
Freyermuth, P. Salles, Wave energy potential assessment in the Caribbean regime in the Colombian Caribbean coastal region? Natural Hazards and Earth
Low Level Jet using wave hindcast information, Appl. Energy 137 (2015) System, Sci 16 (2016) 391e401, https://doi.org/10.5194/nhess-16-391-2016.
375e384, https://doi.org/10.1016/j.apenergy.2014.10.038. [37] Y.-F. Thomas, A. Nicolae Lerma, P. Durand, B. Posada, C. Garcia Valencia,
[12] I.R. Young, S. Zieger, A.V. Babanin, Global trends in wind speed and wave C.A. Andrade Amaya, Altura significativa del oleaje en la cuenca colombiana
height, Science 332 (2011) 451, https://doi.org/10.1126/science.1197219. del caribe, datos de altimetría radar, Boletín Científico CIOH 29 (2011) 27e45,
[13] L. Wu, J. Qin, T. Wu, X. Li, Trends in global ocean surface wave characteristics https://doi.org/10.26640/01200542.29.27_45.
as represented in the ERA-Interim wave reanalysis for 1979e2010, J. Mar. Sci. [38] A.F. Orejarena, L.J. Otero, C.J. Dagua, L. Marriaga, E. Herrera, Determinacio
n del
Technol. 23 (2018) 2e9, https://doi.org/10.1007/s00773-017-0450-1. Clima de Oleaje Medio y Extremal en el Norte del Golfo de Urab
a. Boletín

628

n, J.C. Restrepo, A. Correa-Metrio et al.
A.F. Orejarena-Rondo Renewable Energy 181 (2022) 616e629

Científico, CIO 31 (2013) 109e124, https://doi.org/10.26640/ [54] B. Huang, P.S. Schopf, Z. Pan, The ENSO effect on the tropical Atlantic vari-
01200542.31.109_124. ability: a regionally coupled model study, Geophys. Res. Lett. 29 (21) (2002)
[39] D.B. Enfield, D.A. Mayer, Tropical atlantic sea surface temperature variability 2039, https://doi.org/10.1029/2002GL014872.
and its relation to el nin ~ o-southern oscillation, Journal of Geophysical Reseach [55] M.J. Vega, O. Alvarez-Silva, J.C. Restrepo, J.C. Ortiz, L.J. Otero, Interannual
102 (1997) 929e945, https://doi.org/10.1029/96JC03296. variability of wave climate in the Caribbean Sea, Ocean Dynam. 70 (2020)
[40] N. Booij, R.C. Ris, L.H. Holthuijsen, A third-generation wave model for coastal 965e976, https://doi.org/10.1007/s00382-019-05081-9.
regions 1. Model description and validation, Journal of Geophysical Research. [56] J.J. Mun~ oz-Pe

rez, R. Medina, B. Tejedor, Evolution of longshore beach contour
Oceans 104 (1999) 7649e7666, https://doi.org/10.1029/98JC02622. lines determined by the EOF method, Sci. Mar. 65 (4) (2001) 393e402, https://
[41] S. Kobayashi, Y. Ota, Y. Harada, A. Ebita, M. Moriya, H. Onoda, et al., The JRA-55 doi.org/10.3989/scimar.2001.65n4393.
reanalysis: general specifications and basic characteristics, J. Meteorol. Soc. [57] J.K. Miller, R.G. Dean, Shoreline variability via empirical orthogonal function
Jpn. 93 (2015) 5e48, https://doi.org/10.2151/jmsj.2015-001. analysis: part I temporal and spatial characteristics, Coast Eng. 54 (2) (2007)
[42] G.J. Komen, K. Hasselmann, K. Hasselmann, On the existence of a fully 111e131, https://doi.org/10.1016/j.coastaleng.2006.08.013.
developed wind-sea spectrum, J. Phys. Oceanogr. 14 (1984) 1271e1285, [58] G. Medellín, R. Medina, A. Falque
s, M. Gonza
lez, Coastline sand waves on a
https://doi.org/10.1175/1520-0485(1984)014<1271:oteoaf>2.0.co;2. low-energy beach at “El Puntal” spit, Spain, Mar. Geol. 250 (2008) 143e156,
[43] J. Ortíz, A. Mercado, An intercomparison of swan and wavewatch III models https://doi.org/10.1029/2007JC004426.
with data from NDBC-NOAA buoys at oceanic scales, Coast Eng. J. 50 (1) [59] B.V.J. Borges, S.R. Dillenburgb, Application of multivariate statistical tech-
(2008) 47e73, https://doi.org/10.1142/S0578563408001739. niques in alongshore differentiation of coastal barriers, Mar. Geol. 419 (2020)
[44] I. Jolliffe, Principal Component Analysis, second ed., Springer, New York, 2002. 106077, https://doi.org/10.1016/j.margeo.2019.106077.
[45] E.R. Cook, L.A. Kairiukstis (Eds.), Methods of Dendrochronology: Applications [60] A. Hamouda, S. El-Gharabawy, M. Awad, M. Shata, A. Badawi, Characteristic
in the Environmental Sciences, Springer, 1990. -13: 978-0-7923-0586-6. properties of seabed fluvial-marine sediments in front of Damietta promon-
[46] J.C. Restrepo, J.C. Ortíz, J. Pierini, K. Schrottke, M. Maza, L. Otero, J. Aguirre, tory, Nile Delta, Egypt, The Egyptian Journal of Aquatic Research 40 (4) (2014)
Freshwater discharge into the Caribbean Sea from the rivers of northwestern 373e383, https://doi.org/10.1016/j.ejar.2014.11.006.
south America (Colombia): magnitude, variability and recent changes, [61] A.Y. Otsuka, F.A.N. Feitosa, M.J. Flores-Montes, A. Silva, Influence of fluvial
J. Hydrol. 13 (2014) 266e281, https://doi.org/10.1016/j.jhydrol.2013.11.045. discharge on the dynamics of Chlorophyll-a in the continental shelf adjacent
[47] E.A. Calvo, C.S. Fern
andez, Te
cnicas de Mantenimiento en instalaciones min- to the Recife Port Basin (Pernambuco-Brazil), Braz. J. Oceanogr. 66 (1) (2018),
eras, 2017. https://ocw.unican.es/course/view.php?id¼233&amp;section¼1. https://doi.org/10.1590/s1679-87592018149106601.
(Accessed February 2020). accessed. [62] A.C. Torregroza, J.C. Restrepo, A. Correa, N. Hoyos, J. Escobar, J. Pierini,
[48] R. Waters, J. Engstro €m, J. Isberg, M. Leijon, Wave climate off the Swedish west J.M. Martinez, Fluvial and oceanographic influences on suspended sediment
coast, Renew. Energy 34 (2009) 1600e1606, https://doi.org/10.1016/ dispersal in the Magdalena River Estuary, J. Mar. Syst. 204 (2020) 103e282,
j.renene.2008.11.016. https://doi.org/10.1016/j.jmarsys.2019.103282.
[49] C.J. Dagua, S. Lonin, C.P. Urbano, A. Orfila, Calibracio
n del modelo SWAN y [63] C.W. Zheng, J. Pan, C.Y. Li, Global oceanic wind speed trends, Ocean Coast
validacio
n de reana
lisis del oleaje en el Caribe, Boletín Científico CIOH 31 Manag. 129 (2016) 15e24, https://doi.org/10.1016/j.ocecoaman.2016.05.001.
(2013) 13e28, https://doi.org/10.26640/01200542.31.13_28. [64] Ipcc, in: R.K. Pachauri y, L.A. Meyer (Eds.), Cambio clima
tico (2014): Informe
[50] A.F. Orejarena-Rondo
n, A. Orfila, J.C. Retrepo, I.M. Ramos, I. Hernandez-Car- de síntesis. Contribucio
n de los Grupos de trabajo I, II y III al Quinto Informe
rasco, A 60 year wave hindcast dateset in the Caribbean Sea, Data in Brief 37 de Evaluacio
n del Grupo Intergubernamental de Expertos sobre el Cambio
(2021) 107153, https://doi.org/10.1016/j.dib.2021.107153. Clima
tico [Equipo principal de redaccio
n, IPCC, 2014, p. 157. Ginebra, Suiza.
[51] S.A. Klein, B.J. Soden, N.C. Lau, Remote sea surface variations during ENSO: [65] M. Casas-Prat, X.L. Wanga, N. Swartb, CMIP5-based global wave climate
evidence for a tropical atmospheric bridge, J. Clim. 12 (1999) 917e932, projections including the entire Arctic Ocean, Ocean Model. 123 (2018)
https://doi.org/10.1175/1520-0442(1999)012<0917:RSSTVD>2.0.CO;2. 66e85, https://doi.org/10.1016/j.ocemod.2017.12.003.
[52] D.L. Covey, S. Hastenrath, The pacific el Nin ~ o phenomenon in the atlantic [67] H. Rinne, The Weibull Distribution: A Handbook. Chapman & Hall, first ed.,
circulation, Monthly Weather. Rev. 106 (1978) 1280e1287. https://doi.org/10. CRC Press, 2008b.
1175/1520-0493(1978)106&lt;1280:TPENPA&gt;2.0.CO;2. [68] Japan Meteorological Agency/Japan, JRA-55: Japanese 55-year Reanalysis,
[53] S. Curtis, S. Hastenrath, Forcing of anomalous sea surface temperature evo- Daily 3-Hourly and 6-Hourly Data, Research Data Archive at the National
lution in the tropical Atlantic during Pacific warm events, J. Geophys. Res. 100 Center for Atmospheric Research, Computational and Information Systems
(1995) 15,835e15,847, https://doi.org/10.1029/95JC01502. Laboratory, Dataset, 2013, https://doi.org/10.5065/D6HH6H41.

629