3.2. GIS-Based Spatial Database Creation
Elevation is an important criterion; however, in numerous studies, high altitudes have not been indicated for wind projects [
36,
37,
38]. The researchers mentioned have proposed that locations below 1000 m be considered extremely appropriate for wind projects.
This is due to access issues and a lack of basic infrastructure in higher places. Thus, to reduce the high expenses connected with construction, regions lower than 1000 m in height appear to be the most efficient and appropriate. As shown in
Figure 6a, elevations above 1000 m account for less than 30% of the research region’s total area.
Wind speed: Average wind speed has been the most essential and weighted parameter in wind farm location evaluation studies, as reported in most previous studies [
24,
28,
39,
40]. This parameter is directly related to the project’s profitability [
41]. In our wind farm siting analysis, areas, sites having an average annual wind speed of less than 5 m/s at a height of 100 m above mean sea level were considered inappropriate for wind farm sitings, as recommended by [
28,
42]. Nonetheless, several studies suggest that an annual average wind speed of more than 6 m/s is required for a functional wind farm installation [
43,
44]. Conversely, extremely high wind speeds can damage the wind turbines and the project execution in general.
Slope: A slope map in degrees is produced by combining two SRTM rasters acquired from the USGS website. This criterion can be applied to exclude areas with steep slopes of greater than 15 degrees and high relief. These are typically inaccessible and so unsuitable for wind turbines. The highest slopes, as illustrated in
Figure 6c, are in the research’s area northeast, towards the province of Alpes de Haute, and also surround the shoreline in the south.
Indeed, our choice of 15% (maximum limit of suitable slopes) has already been defined by research works [
45,
46]; others have adopted a constraint of 25% [
47], while some [
48,
49] have raised the constraint threshold to 30%. In addition, some researchers have considered areas with slopes greater than 10% as infeasible areas for wind turbine installation [
50,
51]. Selection of land having a slope of less than 15% is planned to facilitate crane and truck accessibility to sites and to reduce installation and maintenance costs due to turbulence.
Protected area: Wind farm construction is controlled by various laws, most notably the French energy code, the urban planning code, and the environmental code. Any prospective wind project needs to evaluate its environmental impact by including parameters such as landscape impact, biodiversity, noise, and dangers to nearby inhabitants. The protected areas in this study, which include urban areas, wetlands, biodiversity parks, and water surfaces (
Figure 7), were gathered from the IGN’s BD 2021.
The study area is characterized by the presence of forests, pastures, beautiful landscapes, biodiversity parks (fauna and flora), NATURA 2000 protected sites [
52,
53], and the most important large urban agglomerations. These areas of environmental interest were not absolutely excluded according to the literature but also according to the national legislation (the minimum distances were determined after the decision approving the environmental conditions (“DAEC”)). To avoid the destruction of these spaces and the negative impact of wind farms on the nature of these areas, a minimum distance of 2000 m is required [
54,
55].
Road network: This was generated using data from the IGN 2021 database (
Figure 6d). The road’s proximity is a critical parameter in various studies. It is particularly relevant for studies related to the search for suitable sites for a large project implementation requiring massive equipment to keep transportation costs, as well as construction and maintenance costs, low [
24,
39,
40,
56].
Electrical substations: Close proximity to electrical substations minimizes wire costs, prevents power losses, and simplifies installation and maintenance processes [
57].
Figure 6e represents electrical substations in the research area.
3.3. MCDM Using an AHP Approach
Suitable site selection for implementation of a sensitive project such as a wind farm is always difficult since it requires a combination of various parameters and criteria defining the project location. Therefore, decision-making solutions to overcome these obstacles have been developed by integrating all of these determining criteria. Generally, the multi-criteria decision-making (MCDM) approach is always applied to address problems with many stakeholders, criteria, and objectives [
58]. Moreover, this approach has been widely applied in various fields, including the energy sector to plan renewable energy projects [
59,
60,
61,
62]. The analytic hierarchy process (AHP) is a well-known MCDA approach that was initially proposed by [
63] and has subsequently been greatly improved.
The methods for weighting the criteria in the MCDA are diverse. Some of these methods include AHP, fuzzy measures [
64], Analytic Network Process (ANP) [
65], Swara [
66], entropy [
67], Dematel [
68], and standard deviation [
69]. Although these methods are quite limited, AHP is one of the most essential and widely used methods in MCDA. The AHP method is similar to Swara’s in that the expert’s opinion specifies the importance and prioritization of alternatives. As for the entropy method, there are two different views of this method. According to some studies, entropy is reliable and effective [
70]. However, from another point of view, entropy results do not always take into account the importance of the indices [
71]. Dematel is similar to the Swara method, except that the Dematel approach is used to solve extremely complex problems. In the Dematel decision process, the expert opinion is used to develop the pairwise comparison matrix, and it has three main characteristics. The attributes are compensatory and independent of each other. Qualitative attributes are transformed into quantitative attributes [
66]. The Swara and Dematel methods have been widely used in MCDA problems, especially in the renewable energy sector [
72,
73,
74]. In this study, the AHP method was employed to address site selection problems for several reasons:
It is commonly used for its ease of design and implementation. It is highly compatible with GIS, which is widely used for planning and spatial analysis of site selection problems. The consideration of the consistency and inconsistency of alternatives is one of the main advantages of this method [
60].
AHP can be combined with other methods of multicriteria analysis, genetic algorithms, neural networks, etc. [
60]. It also takes into account quantitative and qualitative criteria to interpret the problem [
75].
The AHP method can apply various sensitivity analyses to the criteria. AHP facilitates the decision-making process, using pairwise comparison between criteria [
60]. For site selection problems, in which the main objective is to select the best locations, simple approaches such as AHP are satisfactory, and more complex approaches such as Fuzzy-AHP do not necessarily lead to distinct results [
76].
AHP is a structured decision support method that is primarily focused on sophisticated computations with matrix algebra [
77,
78]. Through this approach, a decomposition of a complicated decision-making issue into a top-down hierarchical structure can be carried out in most cases. In recent years, as geographic information system technologies have improved, GIS integration with MCDA approaches has become increasingly popular. This integration is adaptable and suited to the qualitative and quantitative investigation of multi-criteria issues with a geographical component.
In this research, we developed the decision process required for the usage of the AHP approach. This approach is provided in four steps, each of which requires clear problem identification or study’s objective.
Step 1: Deconstruct the decision-making problem and explain its main characteristics or components (criteria, sub-criteria, options, etc.). Then, using a limited number of levels, create a linear hierarchy of concerns (
Figure 8). Each level has a set number of selection criteria. The aim is expressed at the most fundamental level. Subsequently, the second and third layers comprise the criterion and sub-criteria. The bottom of the hierarchy is allotted to alternatives.
Step 2: Design the judgment matrix and pairwise comparison matrices for each criterion. Based on the Saaty scale (
Table 2), the pairwise comparisons are grouped into a matrix using the following criteria:
In relation to the comparisons of two criteria
C1 and
C2, we designate an important value of the evaluation element “
a”. We place the “
a” value in the cell column “
i” and line “
j” of an important criterion. Then, we need to place the value ratio “1/
a” in the cell considered less important of the comparison.
C1,
C2, and
Cn are the comparison criteria in row “
i” and column “
j”, which correspond to the comparison values
Ci and
Cj. The entries
aij are often taken from the ratio scale (1/9-9) [
79]. The matrix’s element semantic description is provided in
Table 3.
Using the evaluations provided in the previous step, each hierarchy element’s relative relevance was determined. Furthermore, the eigenvector problem is addressed to establish each matrix’s element priority.
First, compute the sum of each
jth column value as follows:
Subsequently, a normalized comparison matrix
n ×
n aij* is generated, in which each
aij in the matrix is divided by the sum of its
jth column, as expressed in Equation (3):
The weights’
ith criterion is then computed as follows:
Step 3: The individual criteria weights are calculated using the eigenvalue procedure’s pairwise comparison matrices. The eigenvalue
λmax is calculated by multiplying each column value by the criteria weight as follows:
Then, using the following equation, we determine the weighted sum value
Sw by adding the sum of each preceding matrix’s rows
ai:
Eventually, for each row, the ratio between the weighted value sum
Sw and the weighting criterion is calculated as follows:
By averaging the ratio i we obtain the highest eigenvalue max.
Step 4: Calculate the consistency ratio
CR (Equation (8)). The final criteria weights are validated using this ratio. Discrepancies in the comparison matrix are identified at this stage:
where the consistency index
CI is calculated as follows:
The value of the
RI varies with the size of the matrix.
Table 4 shows the
RI values according to the number of criteria chosen.
The RC should be lower than 10% to determine that the pairwise comparison evaluations are consistent. If this is not the case, the matrix should be updated, and the element values re-evaluated.
The weighted findings (
Table 5) indicate the most important parameters in wind farm development. The wind’s existence is the greatest driver of wind energy, with a relative weight of 38% (
Table 4), followed in second place by protected regions with a relative weight of 26%. However, distances to power plants and road networks have a significant influence, with relative weights of 13% each, while slope and elevation have the lowest relative weights of 5% and 3%, respectively. Despite their low weights, these criteria should be considered in all wind projects to avoid or minimize potential negative impacts. This weighting choice is based on our study area’s good knowledge.
After completing all of the AHP calculation processes, the following step is to normalize the criteria (
Table 6). The vector data are then converted to a raster format, and the matrices are reclassified into two groups (adequate: code 1; and inadequate: code 0).