1. Introduction
Conventional energy resources have been largely exploited worldwide [
1]. The environmental impact due to carbon emission is one of the major concerns when utilizing such resources [
2]. In addition, fuel/oil resources are gradually depleting on account of exploitation/over-exploitation [
3]. An emerging trend is the utilization of nonconventional energy resources such as solar, wind, tides, and waves, which are clean and renewable [
4]. Electricity production in the United States from renewable sources surpassed coal in April 2019 (
https://www.eia.gov/todayinenergy/detail.php?id=39992). Wind power is the second-most widely used renewable energy source in the world after hydropower, accounting for 24% of the world’s electricity generation capacity (
https://www.power-technology.com/features/). Wind power can be exploited both onshore and offshore. The utilization of wind power is subject to its potentiality in addition to the demand, economy, and policies. The wind energy potential has been analyzed and implemented at a few locations around the globe. A few examples are presented here. China holds the biggest wind energy generation capacity in the world with a production of more than 211 GW in 2018 (
https://www.nsenergybusiness.com/features/china-wind-power-asia-pacific/). The total installed wind power capacity in the United States is around 105 GW (
https://windexchange.energy.gov/maps-data/321). The onshore and offshore wind power capacities in the UK are around 13.6 and 8.4 GW, respectively (
https://www.renewableuk.com). The wind power capacity from installations in India is around 4.6 GW [
5].
Accurate assessment of wind energy potential is a pre-requisite for planning and execution of wind power generation. Measured, satellite, or re-analysis winds are commonly used for such an assessment [
6,
7]. Long-term hindcast/forecast winds from global re-analysis products are considered reliable sources for wind power assessment [
8], and adequate enough when the measured or satellite observations are sparse in a particular region. A few examples of such winds are: Climate Forecast System Reanalysis (CFSR) [
9,
10], Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2) [
11], and European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis (ERA) products such as ERA40 [
12], ERA-Interim [
13], and ERA5 [
14]. These re-analysis wind data are subject to statistical corrections by assimilating extensive sets of in situ and scatterometer winds into their model outputs [
15]. In addition, site-specific validations are commonly carried out prior to wind power assessments. The ERA5 is the latest update of the ERA series, which has been used in wind climate and wind power assessments [
16,
17,
18,
19]; ERA5 winds have lower uncertainty and minimal errors than the other reanalysis winds when compared with measurements [
20,
21].
A better understanding of wind climate in a particular region is essential in prejudicing the feasibility of a resource assessment. In this study, our focus is on the onshore and offshore regions of Qatar (
Figure 1), situated on the central Arabian/Persian Gulf (hereafter called “Gulf”). The predominant winds over the Gulf are northwesterlies throughout the year [
22], and often dominated by shamal winds [
23,
24]. The shamal winds are the stronger northwesterly or northerly wind events, which occur during summer and winter [
25]. Southerly or southeasterly winds are often found in the Gulf, especially during winter [
22]. The shamal wind speeds in the northern Gulf are higher during summer than winter, while the central and southern Gulf experience the opposite [
26]. The winter mean wind speed is higher in the central Gulf, of the order of 5.9 m/s, whereas the summer mean wind speed is higher in the northern Gulf, of the order of 4.9 m/s. The variability in wind speed is high in the southern/eastern Gulf followed by the central Gulf and the northern Gulf. The wind variability in Qatar is composed of regional wind systems, as well as the sea/land-breezes [
27]. The sea/land-breezes prevail only when the regional wind systems weaken, and it is more pronounced during summer than winter [
28]. The wind direction is predominantly northwest and north-northwest in Qatar. The annual mean wind speed varies between 3.0 and 5.0 m/s, whereas the 95th percentile wind speed is between 8.0 and 12.0 m/s [
29].
Wind energy resources were assessed worldwide. A few examples from the adjacent regions are: Mediterranean Sea (100–500 W/m
2) [
30,
31,
32], terrestrial area of Turkey (50–100 W/m
2) [
33], Arabian Peninsula (50–200 W/m
2) [
34], and Kuwait (100–300 W/m
2) [
35]. Wind energy potential was found to be moderate (180 W/m
2) to high (>400 W/m
2) along the Egyptian coast at different stations, calculated at a height of 50–70 m, with high wind power during summer [
36]. The estimated annual mean wind power in the Red Sea was high (up to 500 W/m
2) in the central area due to Tokar gap winds, while it was moderate (100–300 W/m
2) along the coastal belt of Saudi Arabia, at a hub height of 80 m [
37]. The western mountains of Saudi Arabia experience abundant wind resource potential compared to the Red Sea coastal areas [
34]. In Kuwait, the peak wind power density is found along the coastal region during summer [
35]. The calculated annual mean wind power in the Gulf was high in the central region, north of Qatar, of the order of 300 W/m
2 at a hub height of 50 m [
38]. The winds in the Gulf are predominantly northwesterlies [
39], with the highest wind speeds associated with shamal events, having a typical periodicity of 2–5 days [
40]. The winter shamal winds are relatively stronger than the summer shamal winds [
41]. The wind resource is more variable over the Gulf coast, compared to the Red Sea. Assessment of wind resource in the Gulf as a whole indicates that it is possible to produce a power of over 3000 GWh per year [
42].
There are a few studies highlighting the wind energy resources and their feasibility in Qatar, and energy assessment has been carried out for a few locations based on the available measured data. The first attempt was made in 1990 based on the monthly averaged winds measured at Doha at a height of 10 m above ground level, and extrapolated to a height of 25 m [
43]. Subsequent assessments were carried out using measured winds at Doha International Airport extrapolated to 20 m height, as well as at Halul Island [
44]. These assessments covered only the central eastern coast of Qatar, and the data were limited to short-term measurements (25 years at the onshore location and just 1 year at the offshore location). The above onshore estimate at Doha is adequate for a useful assessment, but installation of a wind farm in a highly urbanized region is uncertain. Furthermore, the wind power estimated at Doha offshore needs to be reconsidered as that assessment was based on short-term measurements, and likely to be underestimated. Subsequently, a preliminary resource assessment study has been conducted recently in the context of a proposed wind farm near a natural gas processing plant on the northeast coast of Qatar based on two years of measurements (extrapolated to 130 m hub height) [
45]. Although qualitatively agreeable, these assessments lack the treatment of uncertainties arising due to long-term variability as the duration of measured data is very short. Within these limitations, they identified that a plant can produce a wind power of 17 MW in this region, and that would save 6.813 tons of CO
2 and reduce the natural oil and gas consumption to a good extent.
The literature review shows the following gaps in the existing studies related to Qatar: (i) Earlier assessments were made only at a few locations based on short-term datasets, and hence, long-term datasets at a number of onshore and offshore locations are required to efficiently evaluate the wind resources, (ii) the spatial distribution of wind power within the exclusive economic zone (EEZ) of Qatar was not analyzed, and (iii) the effect of temporal variability and long-term trends in wind resource characterization were not treated, which is important to understand the role of climatic indices in the wind power variability and also to test the reliability of the wind power resource estimations. Hence, a proper assessment of wind power densities for Qatar, including onshore and offshore locations, will enable the identification of potential resource regions for efficiently executing the wind turbine generators (WTG) for power production. The rapid changes in the urbanization of Qatar and the associated sustainable developments demand implementation of such alternative energy productions. In this context, the present study was taken up with the aim to investigate the wind power resources of Qatar at eight select onshore and offshore locations using 40 years of re-analyzed ERA5 winds, and to resolve the existing gaps in the assessment.
The paper is organized as follows: The area of study and geographical features are detailed in
Section 2; the data used and the methods adopted are given in
Section 3;
Section 4 describes the main results and the discussions, and it is divided into five subsections, describing wind climate, annual and decadal wind power, inter-annual variability in wind power, seasonal wind power, and monthly wind power;
Section 5 summarizes the main interpretations.
2. Area of Study
Qatar is situated in the Gulf, between 24°00′ N, 50°30′ E and 26°00′ N, 51°31′ E. It covers an area of ~11,600 km
2 (
Figure 1). It is a relatively flat peninsula with natural topography that varies from 5 to 103 m above mean sea level (MSL). The north-to-south length of the Qatar peninsula is about 160 km. The length of coastline is about 700 km, extending from the Salwa Bay at the border of Saudi Arabia to the border of the United Arab Emirates. The exclusive economic zone (EEZ) of Qatar is approximately 35,000 km
2 with an average water depth of 35 m. The EEZ extends about 176 km seaward to the east and about 94 km to the north [
46]. Qatar experiences an extreme humid and hot climate during summer and dry and cold climate during winter. It often experiences the shamal winds and dust storms. Significant warming has been identified in Qatar in recent decades [
47]. In winter, shamal winds are dominant, which are considered as one among the extremes in the region [
48]. In summer, the wind conditions are controlled by the prevailing regional and local climate systems. Summer shamal winds blow quite frequently, which brings dust from the northern parts of the Arabian Peninsula [
23].
Four offshore and four onshore locations, representing the southeast, southwest, central east, north, and northeast regions of Qatar, were considered for site-specific resource assessment of wind energy extraction. The onshore locations are Mesaieed, Al Khor, and Dukhan. The offshore regions are off Doha, off Ras Laffan, off Al Ruwais, and off Dukhan. Doha, the capital city of Qatar, is situated on the central east coast, adjacent to the Doha Bay. The Mesaieed is an industrial area in the southeast part of Qatar, about 45 km south of Doha. Mesaieed Port located in the natural bay is the premier Port of Qatar, providing services to petrochemical industries, metallurgical plants, and construction-related industries. Al Khor is on the northeast coast, which is one of the major cities of Qatar, situated around 50 km north of Doha. Ras Laffan, located 80 km north of Doha, is an industrial city on the northeast coast of Qatar, developed and operated for the production of natural gas and its derivatives. Al Ruwais is on the northern tip of Qatar facing the central Gulf. It is about 127 km from Doha and holds Al Ruwais Port, which is the second-most important port in Qatar. Dukhan is on the west coast of Qatar, situated about 80 km west of Doha, known for its oil and natural gas reservoir.
3. Data and Methods
ERA5 is the updated reanalysis product of the European Centre for Medium-range Weather Forecasts (ECMWF), which provides hourly estimates of a large number of atmospheric, land, and oceanic climate variables (
https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era5). It covers the Earth on a 30 km grid and resolves the atmosphere using 137 levels from the surface up to a height of 80 km. It includes information about uncertainties for all variables at reduced spatial and temporal resolutions. In the ERA5 reanalysis winds, the scatterometer and in situ wind data are assimilated to improve the accuracy of predictions; thus, the long-term and short-term variabilities are reasonably well-captured [
15]. The ERA5 winds are applied for wind climate studies [
49], wind power assessments [
21], and forcing the hydrodynamic and wave models [
50]. In this study, we used the ERA5 surface winds of 40 years (1979–2018) for the assessment of wind power potential along the onshore and offshore locations of Qatar. This is a long-term dataset, useful for the evaluation of wind power resources as the long-term variability can be treated well. The ERA5 winds were validated with measurements in the Gulf, and a correlation coefficient of 0.95 and bias of 0.07 were obtained [
51]. A few recent works utilized ERA5 winds for the wind climate and energy assessment studies [
8,
16,
51].
The ERA5 winds have been compared with the measured winds (at 10 m height) at Doha Airport (
Figure 1) obtained from the Qatar Meteorology Department (QMD) website, which are available for every 1 h on a daily basis. We used the data during Dec 2019 to April 2020 for validation (
Figure 2). The wind speeds are given in integers such as 5 knots, 6 knots, etc., which has been converted to m/s. The wind direction is given with 22.5° spacing (like NNE, NE, ENE, etc.). Within the limitations of available measured data, ERA5 winds give a reasonable match with the measured winds. The correlation coefficient between the measured and ERA5 wind speeds is 0.78, bias 1.30 m/s, and RMSE 2.22 m/s (17.2% with the maximum wind speed). There is a relatively wider scatter between the measured and ERA5 wind speeds and directions, which are due to numerical limitations in the measured data availability (in the website) and also coarser spatial resolution (0.25° × 0.25°) of the ERA5 data. Moreover, the measurement location is at the land–sea boundary and the data are subject to terrain effects, whereas the ERA5 data mostly represent the sea conditions.
Wind power assessments are commonly carried out at various hub heights (50, 70, 90, 120 m, etc.) and most of the wind resource characterizations are based on winds that are reconstructed from the 10 m height, assuming neutral stability conditions (e.g., [
8,
34,
37]). In the present study, we applied a hub height of 90 m as a reference for the wind power assessment, targeting the 2.1 MW WTG such as Suzlon S97 (
https://www.suzlon.com/in-en/energy-solutions/). Hence, the wind speeds at 10 m height (U
10) have been reconstructed to 90 m height (U
90) using a logarithmic wind profile [
52].
where
is the wind speed at a height (
= 90
,
is the surface friction velocity derived from U
10,
is the aerodynamic roughness length (=0.005 assuming a smooth surface), and
= 0.4 is the von Karman constant. The selected onshore regions are basically flat terrains with a negligible amount of vegetation or obstruction. Hence, we used a similar roughness length for both offshore and onshore areas.
The wind power generation capacity is generally determined by the annual mean wind speeds derived from time series data or by their probability distributions. In a few studies, the wind power potential was estimated using the probability distributions such as Weibull, Rayleigh, and log-normal distributions [
53,
54,
55,
56]. These distributions have also proved effective in the wind power generation. The typical cut-in wind speed for the generation of wind power in a small turbine is 3.5 m/s, and at least 2 m/s wind speed is required to start rotating the turbines. The wind power generators are generally designed for maximum wind speeds, of the order of 10–15 m/s [
57]. The cut-out speed for most of the turbines is 25 m/s [
58]. Recently, slow-wind-speed wind turbines have also been designed [
59,
60,
61], and their efficiency in low wind speeds (~3 m/s) is better than those designed for moderate or strong winds. Effective operation of wind turbines and the forecast of wind power outputs can be achieved using ARIMA–GARCH(-M) approach-based models, which can predict both the mean and volatility of wind speed [
62].
The wind power density is the number of watts of electrical energy produced per square meter of air space (W/m²), computed using the following equation:
where
is the air density and
is the wind speed. The air density could be variable, depending on the variations in temperature and pressure in different time spans (diurnal to seasonal). This may have certain impacts on the wind power estimations [
63]. Moreover, accurate representation of air density variation is often difficult to obtain due to insufficient data, especially when a longer historical period is taken into account. Thus, we used
= 1.225 kg/m
3, the standard air density in the estimations of wind power. The wind power density is an independent estimation, irrespective of the turbine features. For actual power production, the mechanics of the flow passing through the blades and the efficiency of the generator must be taken into account [
36].
In the present study, the time series of wind power densities of Qatar has been calculated using 40 years (1979–2018) of hourly wind data extracted from ERA5. This has been processed to derive the statistics of the wind power densities for each month, season, year, decade, and the 40 year period. The spatial distributions of annual, decadal, inter-annual, seasonal, and monthly mean and standard deviations have been computed. These have been analyzed for site-specific locations. The trends have been evaluated using the Sen’s slope estimate [
64,
65].
The resource variability and effectiveness have been analyzed by estimating the standard deviation (SD), coefficient of variation (CoV), skewness (Sk), kurtosis (K), and variability indices such as annual variability index (AVI), seasonal variability index (SVI), and monthly variability index (MVI). The SD is expressed as:
where
is the value of the variable in each time step,
is the mean of the variable, and N is the total number of timestep.
CoV is used to evaluate the dispersion of wind power distributions around their mean values [
66]. It is the ratio of the SD to the mean.
The
and
are expressed as:
The AVI/SVI/MVI is the difference of mean values between the most energetic and least energetic years/seasons/months divided by the annual mean for the years in consideration [
67].
where
is the maximum annual mean power densities,
is the minimum annual mean power densities,
is the maximum seasonal mean power densities,
is the minimum seasonal mean power densities,
is the maximum monthly mean power densities,
is the minimum monthly mean power densities, and
is the mean power density considering the entire duration.