The hourly wind speed and direction data used in this study were obtained from satellite records available on the NASA website. These data cover an uninterrupted period of ten years (2013-2023), providing a robust dataset for analysis. This extended observation period allows variations in meteorological parameters to be considered and facilitates projections for long-term wind potential forecasting. The exclusive use of NASA satellite data is justified by the lack of ground-based meteorological data at the exact location of the study. This is the first limitation of this study, as the lack of these ground-based measurements may influence the wind resources.

The data processing and analysis tools used in this study are both accessible and highly effective for handling meteorological data:

There are several methods for determining the scale (c) and shape (k) parameters from a given wind distribution [3] :

$c=\frac{1.125V_{moy}}{\left(1-B\right)}$ (4)

$k=1+0,483{\left(v_{moy}-2\right)}^{0,51}$ (5)

$B=1-0.81{\left(v_{moy}-1\right)}^{0.089}$ (6)

The graphical method used in this article involves the linearization of cumulative distributions. This is done by applying a logarithmic transformation to the distribution function F(v):

$\ln \left[-\ln \left(1-F\left(v\right)\right)\right]=k\ln v-k\ln c$ (7)

And by posing:

$Y=\ln \left[-\ln \left(1-F\left(v\right)\right)\right]\text{et}X=\ln v$ (8)

We obtain a linear function of X:

$Y=aX+b,k=a\text{et}b=-k\ln c$ (9)

The scale parameter c is worth:

$c={\text{e}}^{-\frac{b}{a}}$ (10)

If the mean velocity and variance of a site are known (otherwise they can be determined from the statistical distribution), the shape parameter k is determined using the following approximation [4] :

$k={\left(\frac{\sigma }{v}\right)}^{-1.086}$ (11)

We can deduce the scale parameter:

$c=\frac{v}{\Gamma \left(1+\frac{1}{k}\right)}$ (12)

Estimating the wind potential at the height of the wind turbine tower, which is considered meters above the ground, requires the transformation of the Weibull parameters. The following equation is used to estimate the coefficient at a different height:

$\frac{C_2}{C_1}={\left(\frac{z_2}{z_1}\right)}^{m^{″}}$ (13)

For any heights, Mikhail gave the expression for the exponent $m^{″}$ considering the roughness of the site as follows [8] :

$m^{″}=\frac{1}{\ln \left(\frac{z_g}{z_0}\right)}+\frac{0.0881\left(1-\ln C_1\right)}{1-0.0881\ln \left(\frac{z_1}{10}\right)}$ (14)

With z₀, the roughness, defined as the height from which the friction speed slows down the wind speed.

The geometric mean of z₀ is z_g and determined by the following relation:

$z_g=\sqrt{z_1{z}_2}$ (15)

Thus, the expression for the exponent $m^{″}$ becomes:

$m^{″}=\frac{1}{\ln \left(\frac{\sqrt{z_1{z}_2}}{z_0}\right)}+\frac{0.0881\left(1-\ln C_1\right)}{1-0.0881\ln \left(\frac{z_1}{10}\right)}$ (16)

Wind speed varies with height above the ground. This phenomenon is called vertical wind shear. Over flat terrain and with neutral atmospheric stratification, the logarithmic wind profile provides a good approximation of the vertical shear. The logarithmic profile is derived from the ratio of the vertical wind profile and is given by the following expression:

$v_2=v_1\frac{\ln \frac{h_2}{z_0}}{\ln \frac{h_1}{z_0}}$ (17)

where h₁ is the height of the mast, v₁ is the speed measured at the level of the mast and v₂ is the speed we want to determine.

The roughness length z₀ is defined relative to the type of terrain. For the present work, we considered a very uneven terrain whose roughness length z₀ is theoretically defined to be equal to 0.4 m [8] .

The power law of Justus and Mikhail gives a good extrapolation for higher altitudes, the wind speed v(z₂) at the desired altitude z₂ is expressed as follows [9] :

$v\left(z_2\right)=v\left(z_1\right)\times {\left(\frac{z_2}{z_1}\right)}^{\alpha }$ (18)

The formula for estimating the coefficient of friction α to plot the vertical profile and thus allow instantaneous extrapolation of wind speeds is written in the following form:

$\alpha =a+b\times \ln \left(v_1\right)$ (19)

where a and b are constants whose values depend on the height of the anemometer, given by:

$a=\frac{0.37}{1-0.088\times \ln \frac{z_1}{10}}$ (20)

$b=\frac{-0.008}{1-0.088\times \ln \frac{z_1}{10}}$ (21)

$\alpha =\frac{0.37-0.088\times \ln v_1}{1-0.088\times \ln \frac{z_1}{10}}$ (22)

Both profiles are commonly used to represent the vertical distribution of wind speed: the logarithmic law applied in this work gives a good extrapolation of wind speed for altitudes less than or equal to 100 meters. The power law is adapted to higher altitudes (≥200 m).

Wind measurements can be used to calculate the wind energy available at a given site. In practice, the annual wind energy, expressed in kWh/m², is calculated using the BETZ limit, written as [3] :

$E_B=3.56{\displaystyle \sum f_i{v}_i^3}$ (23)

where f is the frequency corresponding to the speed v of the wind of class i.

This calculation can be done using the Weibull distribution, we obtain the following expression:

$E_B=3.56C^3\Gamma \left(1+\frac{3}{k}\right)$ (24)

It provides information on the prevailing wind directions. The wind frequencies are distributed over 360˚, divided into 12 sectors of 30˚. A frequent but very weak wind will have a small share of the energy of the wind rose. This rose therefore makes it possible to identify the directions that contain the most energy, i.e. strong and frequent winds. The frequency distribution of wind speeds according to their direction is summarized in Table 1 .

The series of hourly wind speed measurements over ten years show that winds with speeds of approximately ±3 m/s are dominant and consistent throughout all years of measurement. Figure 1 shows the variations in the time series of wind speeds at the site. The data from the hourly wind speed measurements are grouped by wind class together with their frequency of occurrence. Table 1 summarizes this information and shows the percentage of time the wind is blowing within each speed interval.

The Weibull parameters of the frequency distribution are a shape factor k = 2.2 and a scale factor c = 3.76 m/s. These values of k and c were obtained by logarithmic transformation of the distribution function, as shown in Figure 2 . The relatively high value of the Weibull shape parameter k indicates that the distribution is tightly centred around the mean, suggesting that the prevailing winds are relatively constant with minimal variability. The probability density and cumulative probability density of the distribution are plotted for the obtained values of c and k. In Figure 3 , the frequency distribution derived from the time series measurement data is shown in red, while the Weibull distribution is plotted in blue. The mean wind speed calculated from the Weibull distribution is 3.33 m/s.

Wind resources are generally exploited at heights above 10 meters, which is the height of the measurement mast. Wind turbines are generally installed at a height of around 100 meters to maximize the kinetic energy of the wind. It is therefore essential to extrapolate wind speeds vertically. Table 2 shows a significant increase in the amount of energy produced as a function of height z (m). At a height of 100 meters, the average wind speed is 5.71 m/s, which corresponds to a generated energy of 1315.30 kWh/m². This represents a 59% increase in the energy produced at a height of 10 meters.

These results, obtained using the logarithmic method, are valid for altitudes below 100 meters and can be a prerequisite for more detailed studies of the wind potential of the site.

Analysis of the wind statistics shows that the predominant winds come from the north-north-east, east-north-east and south-west directions, represented by sectors 1, 2, 3 and 8 of the wind rose in Figure 4 . The other sectors correspond to areas with lower average wind speeds. A wind sensor should be oriented towards the direction of the most dominant winds, with a preference for sector 2, i.e. the north-north-east direction, as indicated in Table 3 .