A Geographic Information System-Based Integrated Multi-Criteria Decision-Support System for the Selection of Wind Farm Sites: The Case of Djibouti

Abstract

Wind energy is a promising alternative energy source to cover large amounts of electricity demand in African countries. Djibouti’s proximity to the Red Sea and its arid and semi-arid climate generate consistent and robust winds, contributing to its potential for wind energy. Notwithstanding its considerable potential, Djibouti has not been adequately examined in earlier studies to determine suitable sites for wind farms. The objective of this study is to develop a model by integrating CRiteria Importance Through Intercriteria Correlation and Combined Compromise Solution methods into a Geographic Information System-based decision-support system to establish a comprehensive framework for the selection of wind farm sites in Djibouti. Following an in-depth review of the literature, seven main criteria were identified to assess the suitability of potential sites for wind farm construction: wind velocity, changes in wind direction, ground slope, distance to urban areas, distance to road network, distance to energy transmission networks, and land use. The CRiteria Importance Through Intercriteria Correlation method objectively determines the relative importance of the criteria, identifying wind speed and proximity to power transmission networks as the most important, and ground slope and land use as less important than the other criteria. The Combined Compromise Solution method is employed to prioritize potential sites for wind farms, considering seven specified criteria. To enhance the reliability of the results derived from the Combined Compromise Solution method, validation was conducted utilizing the Multi-Attribute Ideal–Real Comparative Analysis method. The comparative analysis revealed a robust correlation between the results of the two methods, providing convincing evidence for the accuracy and reliability of the proposed decision-support system employed to determine the most suitable sites for wind farms in Djibouti. This study is expected to assist professionals and researchers in dealing with the wind farm site selection problem on an unprecedented scale and with exact coordinates through a decision-support system that concurrently integrates the most recent multi-criteria decision-making methods and Geographic Information System tools.

1. Introduction

Africa’s population is increasing at a very high rate, leading to rising energy demand. The population of Africa is predicted to reach 2.5 billion people by 2050 [1]. Given this high population increase, meeting energy demand has become challenging in African countries [2]. Djibouti is one of the African countries struggling with high energy demand, which is affected by economic development and an increase in population. Moreover, the energy demand is increasing not only for commercial use but also for residential use. Djibouti has been importing electricity from Ethiopia, corresponding to approximately 80% of the total demand. The cost of electricity in Djibouti is among Africa’s highest, which is USD 0.30 per kWh on average. The Djibouti-Country Strategy Paper 2023–2027, prepared by the African Development Bank Group, estimates that the nation’s average electrical energy demand will reach 500 MW in 2025 and 1000 MW in 2030 [3]. Considering the need for electricity, the Government of Djibouti has launched a long-term development plan foreseeing the use of alternative energy sources, including wind, geothermal, and solar energy [4,5]. Renewable energy projects have been supported in Djibouti to provide affordable clean energy, reduce electricity prices, and increase private sector competitiveness [3].

Supporting renewable energy projects can be an effective way to meet the need for electricity because Djibouti has great potential for wind energy due to being close to the Red Sea, which generates strong and consistent winds. Indeed, according to the report prepared by Helimax (2004) [6], Djibouti is one of the top fifteen countries in Africa exhibiting the most significant potential for wind energy generation [6]. Moreover, in the renewable energy literature, there are studies highlighting Djibouti’s significant potential for wind energy in the region [5,7,8,9]. However, making use of this potential highly depends on selecting the most appropriate site for wind farms. In other words, selecting the most appropriate site for wind farms is necessary not only to maximize energy extracted from the wind but also in terms of sustainability [10,11,12,13]. In spite of the fact that Djibouti has a promising future for wind energy, the literature review reveals that none of the studies focus on developing a decision-making model that can be used as a tool in selecting the most appropriate site for a wind farm in Djibouti [14,15]. This research attempts to bridge this gap by proposing a decision-making model that can serve as a tool for selecting the most appropriate site for a wind farm in Djibouti.

Developing a decision-making model for this purpose is a challenging task since selecting the most appropriate site for a wind farm is a complex decision-making problem that includes multiple conflicting criteria [10,16]. Since the most appropriate site for wind farms is solely based on comparing the sites in a region according to multiple criteria, using a multi-criteria decision-making (MCDM) method can help researchers and professionals to decide which site is more appropriate than the others. However, researchers and professionals in the field of renewable energy have not taken full advantage of MCDM methods that can provide them with more robust solutions. In the renewable energy literature, MCDM methods are mostly used to determine the weights of the criteria used in selecting the most appropriate site for wind farms. A decision-making model is a systematic approach for guiding individuals with complex decisions consisting of a set of steps including the definition of the problem, the identification of criteria, the evaluation of alternatives, and the selection of the most appropriate alternative [17]. The MCDM is a set of processes to compare and assess a set of conflicting criteria in decision making in the existence of several alternatives and criteria to take into account [18]. MDCM methods help individuals for determining and prioritizing the weights of criteria to arrive at the most appropriate solution. The Analytical Hierarchy Process (AHP) is an MCDM technique that structures a decision problem into a hierarchy [19]. The technique investigates the problem in terms of a set of criteria comparing the relative importance through pairwise comparisons. On the other hand, AHP relies on subjective judgments, where most data come from questionnaires with individuals [20].

Previous studies have already implemented a wide array of site selection methods for wind farms. A major portion of studies focused on site selection using an MCDM method or a combination of MCDM and GIS. For example, Demir et al. (2024) [21] used GIS technology to identify suitable wind farm sites, utilizing Fuzzy Stepwise Weight Evaluation Ratio Analysis (F-SWARA) and Fuzzy Measurement Alternatives and Ranking by Compromise Solution (F-MARCOS) for ranking and prioritizing the criteria of wind farm installation and deciding on the most suitable site in Sivas province in Türkiye. In another study, Van Haaren and Fthenakis (2011) [22] developed a GIS-based spatial multi-criteria methodology for selecting the most suitable wind farm site in New York State. Similarly, Sánchez-Lozano et al. (2016) [23] used a GIS-based site selection for Southeastern Spain using Fuzzy Analytic Hierarchy Process (FAHP) to determine the weights of criteria and Fuzzy TOPSIS (FTOPSIS) method to evaluate alternatives, which are among the most widely implemented MCDM methods. Yildiz (2024) [24] presented a spatial MCDM approach based on GIS to determine the most suitable wind farm installation site in Balıkesir province in Türkiye. In Yildiz’s (2024) [24] study, the weights of criteria were determined using the AHP method based on a questionnaire, and scored criteria maps were used to generate the wind farm suitability map. Konstantinos et al. (2019) [25] proposed a decision-support system methodology for wind farm site selection in Eastern Macedonia and Thrace region in Greece. In their study, they utilized GIS and AHP to spatially present the results and determine the most suitable site for wind farm installation. To rank the sites, they utilized the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) method with respect to installation suitability. In another study, Szurek et al. (2014) [26] presented a GIS-based method for determining the most suitable site of a wind farm in the Lower Silesia region in Poland. The study utilized AHP and Weighted Linear Combination (WLC) methods to determine the weights of the criteria and develop a suitability map for the related site. Despite the intensity of research for wind farm site selection, a major part of the research relies on rather older and relatively subjective methods such as AHP, which mostly relies on subjective assessments for the weights of criteria. Moreover, previous studies fail to provide a specific coordinate or set of coordinates, which could show the exact location for wind farm installation. Given these constraints, this study utilizes newer methods such as the CRiteria Importance Through Intercriteria Correlation (CRITIC) method, which can overcome the shortcomings of subjective assessment for the weights of criteria, as well as the Combined Compromise Solution (CoCoSo) and Multi-Attribute Ideal–Real Comparative Analysis (MAIRCA) methods for ranking the alternative sites.

The research presented in this paper proposes a model that uses MCDM methods not only for determining the weights of the criteria but also for ranking the locations in a region that allows prioritizing them to select the most appropriate location for wind farms. The methods used in this model involve the determination of the weights of criteria through objective assessments rather than relying on subjective judgments compared to the traditional MCDM methods such as AHP. In this respect, the CRITIC method was applied to objectively determine the relative importance of criteria, and the CoCoSo method for prioritizing potential sites for wind farms. Moreover, the proposed method validates the results through MAIRCA methods. All three methods provide a more objective assessment through unbiased data assessments and help to avoid complex pairwise comparisons unlike the traditional MCDM methods. In this respect, the proposed model is expected to help decision makers objectively evaluate and prioritize potential sites for wind farms, considering various criteria and their relative importance through unbiased data assessments. The proposed model not only meets the need for site selection models that make use of MCDM methods to prioritize the locations for wind farms but also overcomes a shortcoming in determining the weights of the criteria considered in selecting the most appropriate location. Abdi et al. (2024) [9] claim that determining the weights of criteria in AHP relies on the subjective judgments of professionals. In Abdi et al.’s study [9], it is also stated that methods exist that can objectively determine the relative importance of criteria. Therefore, in this study, the proposed model utilizes the CRITIC method to overcome the shortcomings of using methods based on the subjective judgments of professionals.

In sum, this study’s main objective is to develop an advanced decision-making model that can be used to select the most appropriate site for a wind farm in Djibouti. The model proposed in this study utilizes the CRITIC method to determine the weights of criteria and CoCoSo and MAIRCA methods to rank the alternative sites. The proposed GIS-based decision-support system is unique in that it (1) employs the CRITIC method to objectively determine the relative importance of criteria by analyzing their correlation and contrast intensity, ensuring an unbiased assessment of their impact on wind farm site selection; (2) integrates GIS tools to provide exact spatial analysis, offering professionals useful guidance with specific coordinates for wind farm construction; and (3) addresses a significant research gap by focusing on Djibouti, a region with substantial wind energy potential but is underexplored in the literature. In the developed model, several criteria are considered, including the wind energy potential, hourly wind speed, wind direction standard deviation, slope, distance to urban areas, distance to the road network, distance from energy transmission networks, and land use to provide the most suitable location for wind farm construction. Moreover, the proposed model provides an exact coordinate on the suitability map that can be used by professionals when constructing wind farms. Armed with such an advanced tool, it may be easier for researchers and professionals in the field of renewable energy to select the most appropriate location for wind farms. Indeed, the results of this research reveal that the proposed model can be used as an effective tool in selecting the most appropriate location for wind farms.

2. Literature Review

Wind energy is one of the significant and promising renewable energy sources providing a clean and cost-competitive alternative to fossil fuels. The performance and energy potential of wind energy systems are heavily influenced by the selected site [27]. Numerous studies have examined wind farm location selection in the renewable energy literature, which is a challenging decision-making problem. A majority of studies determined exclusion criteria, evaluation criteria, the relative importance of evaluation criteria, developed site selection methodologies, and conducted wind resource and sensitivity analyses [28].

2.1. On-Land Wind Farm Site Selection Criteria

To select the most appropriate site for wind farms, multiple criteria should be considered under various factors, including technical, topographical, meteorological, economic, infrastructural, administrative, environmental, ecological, social, and visual aspects [16,29,30,31,32,33,34,35]. There are two groups of criteria determining renewable energy sites: exclusion criteria and evaluation criteria. Exclusion criteria prevent the installation of wind farms in unsuitable locations [27]. The exclusion criteria differ according to unique characteristics and climatic conditions of the region, laws, regulations, and policies of the country, and available geographic information data. The number of exclusion and evaluation criteria was found to be higher for on-land wind farms compared to offshore wind farms due to various factors affecting land-based sites [28]. The most commonly used exclusion criteria for on-land wind farm site selection were slope of the terrain, protected areas, proximity to urban and residential areas, rural areas, proximity to airports, proximity to road networks, proximity to transmission lines and electricity grid, water bodies, unsuitable land-use areas, bird habitats, and migration corridors [28,36]. On the other hand, evaluation criteria help identify locations that optimize efficiency while balancing economic, environmental, and other considerations [27]. In previous studies, evaluation criteria were mostly related to technical, economic, environmental, and social aspects. The most frequently used evaluation criteria for on-land wind farm site selection in terms of technical aspect were wind speed, slope, wind power density, elevation, turbulence intensity, and geographical direction. In terms of economic aspect, they were the distance from roads, distance from transmission lines, distance from residential areas, distance from railroads, distance from electricity substations, costs, and population density. In terms of environmental aspect, they were land use, pollutant emission reduction benefits, distance to airports, distance to natural reserves, distance from protected areas, and noise. In terms of social aspect, they were public support, impact on local economy, and policy support [36].

Some criteria can be applied relatively consistently worldwide; however, others may differ significantly based on national laws, regulations, or specific characteristics of the region [16]. In previous studies, some criteria specific to the region were also considered. For example, distance to fault lines was included as an evaluation criterion for wind farm selection considering the seismic activity in Türkiye [11,24,37,38]. In the studies of Atici et al. (2015) [37] and Asadi and Pourhossein (2019) [39] investigating regions that include mining sites, the distance to mining sites criterion was added. In Al-Yahyai et al.’s (2012) study [40], they considered sand dunes as a criterion affecting the performance of wind turbines, considering that Oman has sand dunes that may cause sand and dust storms on windy days.

Determinant criteria are significant factors used, weighted, and deployed in models by considering specific characteristics of the area [27]. In Rediske et al.’s (2021) study, criteria are determined based on the literature review and characteristics of the studied region. On-land wind farm site studies that integrate GIS and MCDM methods are reviewed, and the most relevant criteria are presented in

Table 1. Overview of on-land wind farm site selection studies utilizing MCDM methods and GIS.

Researcher(s)	Year	Country	MCDM Method(s)	Wind Velocity	Changes in Wind Direction	Ground Slope	Distance to Urban Areas	Distance to Road Network	Distance to Energy Transmission Network	Land Use
Tegou et al. [45]	2010	Greece	AHP	X		X		X	X	X
Aydin et al. [46]	2010	Türkiye	OWA	X			X
Georgiou et al. [47]	2012	Cyprus	AHP	X		X	X	X	X	X
Al-Yahyai et al. [40]	2012	Oman	AHP, OWA	X		X	X	X
Gorsevski et al. [41]	2013	USA	WLC	X					X	X
Sanchez-Lozano et al. [48]	2014	Spain	ELECTRE-TRI	X		X	X	X	X	X
Latinopoulos and Kechagia [49]	2015	Greece	AHP, WLC	X		X	X	X		X
Atici et al. [37]	2015	Türkiye	ELECTRE III, ELECTRE-TRI, SMAA-TRI			X	X	X	X
Watson and Hudson [29]	2015	UK	AHP	X		X	X	X	X	X
Sanchez-Lozano et al. [23]	2016	Spain	AHP, TOPSIS	X		X	X	X	X
Höfer et al. [16]	2016	Germany	AHP	X		X	X	X	X	X
Baseer et al. [43]	2017	Saudi Arabia	AHP	X		X	X	X	X
Gigović et al. [30]	2017	Serbia	DEMATEL, ANP, MABAC	X	X	X	X	X	X	X
Villacreses et al. [31]	2017	Ecuador	AHP, OWA, OCRA, VIKOR	X		X	X	X	X	X
Ali et al. [50]	2017	South Korea	FTN, AHP	X		X	X	X	X	X
Pamučar et al. [51]	2018	Serbia	BWM, MAIRCA	X	X	X	X		X	X
Ayodele et al. [32]	2018	Nigeria	AHP	X			X	X	X	X
Değirmenci et al. [38]	2018	Türkiye	AHP	X			X	X
Asadi and Pourhossein [39]	2019	Azerbaijan	AHP, VIKOR, TOPSIS	X		X	X	X		X
Xu et al. [52]	2020	China	AHP, VIKOR	X	X	X	X	X	X
Moradi et al. [53]	2020	Iran	AHP	X		X	X	X	X
Nasery et al. [54]	2021	Afghanistan	AHP	X		X	X	X	X	X
Feloni and Karandinaki [55]	2021	Greece	AHP, WLC	X		X	X	X		X
Saraswat et al. [56]	2021	India	AHP	X		X	X	X	X	X
Yousefi et al. [34]	2022	Iran	AHP	X	X	X	X	X	X
Ifkirne et al. [33]	2022	France	AHP	X		X		X	X
Zalhaf et al. [35]	2022	Sudan	AHP	X		X	X	X	X
Karamountzou and Vagiona [44]	2023	Greece	AHP, TOPSIS	X		X	X	X	X	X
Yegizaw and Mengistu [57]	2023	Ethiopia	AHP	X		X	X	X	X	X
Josimović et al. [42]	2023	Serbia	PROMETHEE			X	X	X	X	X
Demir et al. [58]	2024	Türkiye	SWARA, MARCOS	X	X	X	X	X
Yildiz [24]	2024	Türkiye	AHP	X		X	X	X	X	X
Placide and Lollchund [10]	2024	Burundi	AHP	X	X	X		X	X	X
Yaman [11]	2024	Türkiye	AHP	X		X		X	X	X

Note: AHP: Analytic Hierarchy Process, BWM: Best–Worst method, ELECTRE: Elimination and Choice Translating Reality, FTN: Fuzzy Trial Numbers, MABAC: Multi-Attributive Border Approximation Area Comparison, Multi Attributive Ideal–Real Comparative Analysis (MAIRCA), MARCOS: Measurement Alternatives and Ranking by Compromise Solution, OCRA: Occupational Repetitive Actions, OWA: Ordered Weight Averaging, PROMETHEE: Preference Ranking Organization Method for Enrichment Evaluations, SMAA: Stochastic Multiobjective Acceptability Analysis, SWARA: Stepwise Weight Evaluation Ratio Analysis, TOPSIS: Technique for Order of Preference by Similarity to Ideal Solution, VIKOR: VIekriterijumsko KOmpromisno Rangiranje, WLC: Weighted Linear Combination.

. Wind speed is a fundamental criterion that determines the wind energy to be exploited by a wind farm [27,41]. Wind speed was used in nearly all studies, as can be seen in Table 1, which makes it one of the most determinant criteria [16,27]. The slope is a significant criterion, as wind turbines should be installed on the horizontal ground; otherwise, the terrain should be leveled, adding complexity and cost. Distance to urban areas controls maintaining a safe distance to residential areas, avoiding noise pollution, visual impact, effect of flickering shadows, and potential safety risks while also ensuring proximity to population centers and meeting local energy demand efficiently [11,42]. Distance to the road network evaluates the accessibility of the wind farm location to optimize transportation and logistics, affecting construction and maintenance costs. Distance to energy transmission networks measures the ease of transporting energy from the wind farm to the grid with shorter distances, reducing energy loss and network connection costs [11,16,41,43,44]. Many studies consider criteria including slope, distance to urban areas, distance to the road network, and distance from energy transmission networks. Table 1 presents these studies in detail in terms of criteria. Land use assesses the suitability of an area for wind farm installation, considering restrictions and environmental impact [27]. These criteria are among the most used criteria according to previous review studies [16,27,28,36]. Even though wind direction was considered in a low number of studies, it plays a critical role in wind turbine loads, wake dynamics, and overall wind farm performance, influencing wind resource evaluation, micro-siting optimization, operational cost reduction, and turbine efficiency enhancement [9]. Hence, it is also included in this study to fill this gap in previous studies.

2.2. On-Land Wind Farm Site Selection Research Methods

Wind farm site selection studies utilized statistical methods, MCDM methods, and GIS [28,42]. Methodological stages in wind farm site selection studies can be categorized as the exclusion stage, the evaluation criteria stage, and the assessment of suitable sites stage [28]. Different advancements were observed in the late 1970s to the late 1990s in terms of utilizing meteorological data [59] including mathematical modeling [60,61], climatology [62,63], and simulations [64]. GIS is mostly utilized in the exclusion stage, assessment stage of suitable sites, and generation of suitability maps [28]. Being one of the earliest studies that integrate GIS into suitable wind farm site selection, Baban and Parry (2001) [65] identified 14 criteria under physical, resource, economic, planning, and environmental categories and constraints of these criteria based on a questionnaire conducted with public and private organizations. Using GIS, constraint maps were generated, and two methods were used to combine different layers, namely, equal weighting and allocated weighting based on perceived importance. Suitability maps were then created. The allocated weighting method was found to be more favorable, allowing users to assign varying levels of importance to each criterion.

MCDM methods provide a structured framework to solve complex decision-making problems by evaluating and integrating multiple, often conflicting criteria [16]. MCDM methods over the last decade reveal a consistent increase in the number of publications across all methods, indicating a growing interest. When combined with GIS, MCDM enhances the capability of spatial analysis, visualization, and ranking suitability of sites. Previous studies showed the relevancy and importance of GIS-based methodologies [28]. A large portion of research focused on wind farm site selection, identifying suitable locations through the integration of GIS and MCDM methods. Many studies integrated GIS with AHP, as can be seen in Table 1, which made a significant advancement in the wind farm site selection [11,16,24,29,33,34,38,43,45,47,53,57]. AHP is most frequently used to assign weights to the criteria for wind farm site and less frequently used to prioritize decision alternatives [28].

Tegou et al. (2010) [45] utilized AHP to assign evaluation criteria weights and integrated GIS to establish spatial dimensions and generate a suitability map. However, they did not explicitly mention which method was used to determine the weights of the criteria. They analyzed Lesvos Island in Greece and found that 1.4% of the island’s surface is most suitable for wind farm projects. Georgiou et al. (2012) [47] determined exclusion and evaluation criteria under environmental, social, administrative, technical, economic, and technical factors to evaluate wind farm sites in Cyprus. They used AHP for determining the weights of the criteria, Simple Additive Weighting (SAW) for aggregation, and GIS for presentation. A limitation of this study was the lack of differentiation within criteria, assigning the same score to areas with varying values [16]. Höfer et al. (2016) [16] applied a GIS-based AHP to provide a comprehensive approach. They determined a comprehensive list of exclusion and evaluation criteria based on a thorough literature review. A survey was conducted with experts representing different stakeholders to be used in AHP. They excluded areas based on restrictions, created maps, and determined suitable areas. They found that 1.7% of the study area exhibited high suitability. The results verified that existing wind farm locations align with the identified suitable areas. However, a few existing wind farms are located in highly suitable areas, showing a focus on excluding unsuitable areas rather than finding the best possible locations. Yaman (2024) [11] developed a GIS-based AHP model to evaluate the land suitability of wind power plants in Adana, Türkiye, by using 15 criteria under technical, economic, and environmental dimensions. Of the study area, 9.9% was found suitable, and 51.7% was found moderately suitable. Furthermore, many studies combined several MCDM methods to leverage the strengths of each approach and enhance the diversity of available methods. Al-Yahyai et al. (2012) [40] developed a methodology for wind farm land suitability index, incorporating AHP and Ordered Weighted Averaging (OWA) that allows for evaluating trade-offs between criteria and adjusting decision-making strategies. They found that 0.2% of the study area in Oman was most suitable. Feloni and Karandinaki (2021) [55] utilized a GIS-based MCDM methodology by combining AHP and WLC. They considered three scenarios: first focusing only on technical criteria, then incorporating technical and economic criteria, and lastly incorporating technical, economic, and environmental criteria. The results showed that site selection is influenced not only by spatial constraints but also by which criteria are selected. Karamountzou and Vagiona (2023) [44] made the first study that evaluated the suitability and sustainability of existing wind farms in Greece at the national scale. They identified the existing wind farms, determined exclusion and evaluation criteria, proposed a suitability assessment framework, and determined priority ranking based on AHP, TOPSIS, and GIS. A high percentage of existing wind farm sites were located within suitable areas.

In the extensive array of methodologies applied to wind farm site selection, AHP and TOPSIS stand out as the most commonly utilized techniques within the literature [66]. Despite their prevalence, these methods have faced criticism, where scholars have pointed out the subjective nature of these methods, particularly for AHP in the context of pairwise comparisons and the assignment of weights to criteria, which can significantly influence the decision-making process [23,53,67]. These criticisms led to a substantial shift in mindset towards employing more objective methodologies. Hence, novel MCDM methods that address the limitations of traditional approaches were utilized in recent studies. Atici et al. (2015) [37] proposed a structured decision-making tool that consists of the elimination of infeasible sites using GIS and evaluating remaining sites with Elimination and Choice Translating Reality (ELECTRE III) for ranking, ELECTRE-TRI for sorting, and Stochastic Multiobjective Acceptability Analysis (SMAA-TRI) for sorting under uncertainty. They applied the methodology in Western Türkiye and the results indicated consistency between the MCDM methods. Gigović et al. (2017) [30] proposed a novel model for identifying suitable wind farm locations, combining GIS and MCDM methods Decision-Making Trial and Evaluation Laboratory (DEMATEL), the Analytic Network Process (ANP), and Multi-Attributive Border Approximation area Comparison (MABAC). They determined constraints and evaluation criteria across economic, social, and environmental factors. The methodology was applied in Vojvodina, Serbia, and its results were compared with those obtained from TOPSIS, VIKOR, and COPRAS methods, demonstrating a high level of consistency across the methods. Villacreses et al. [31] conducted the first study in Ecuador integrating GIS and multiple MCDM methods to evaluate wind farm suitability. They used AHP to calculate the weights of the criteria and OWA, OCRA, VIKOR, and TOPSIS to rank the alternatives. The analysis of Pearson correlation among the MCDM methods revealed strong agreement and consistency between OWA, OCRA, and VIKOR methods, while TOPSIS demonstrated a lower correlation. Josimović et al. (2023) [42] developed a simple and applicable procedure by integrating GIS to eliminate sites based on constraints and the Preference Ranking Organization Method for Enrichment Evaluations (PROMETHEE) to evaluate locations based on weight categories. They identified nine exclusion criteria as well as 18 evaluation criteria under spatial and socio-economic factors based on previous articles, practical experience, and regulations. This study differs by considering wind potential as a prerequisite rather than as an evaluation criterion.

In another comprehensive study, Badi et al. (2023) [68] implemented recent MDCM tools for wind farm site selection in the city of Derna, Libya. They used the Best–Worst method (BWM) and AHP to calculate the weights of the criteria. They made a comparison of five different sites using six criteria. They ranked the alternatives using Measurement of Alternatives and Ranking According to the Compromise Solution (MARCOS) method. Finally, they performed a sensitivity analysis to check the stability of results changing the weights of the criteria. The results were compared with other MCDM methods, which are Combinative Distance-based Assessment (CODAS), Evaluation based on Distance from Average Solution (EDAS), TOPSIS, Ranking of Alternatives through Functional mapping of criterion sub-intervals into a Single Interval (RAFSI), Simple Additive Weighting (SAW), Additive Ratio Assessment (ARAS), Multi-Attributive Border Approximation area Comparison (MABAC), Weighted Aggregated Sum Product Assessment (WASPAS), and Combined Compromise Solution (CoCoSo). MARCOS results were found to be consistent with CODAS, EDAS, SAW, ARAS, MABAC, and WASPAS methods. These studies highlighted the novel methods and ongoing development of MCDM models in wind farm site selection.

Some studies integrated fuzzy logic into MCDM methods to handle qualitative criteria along with quantitative criteria. Gorsevski et al. (2013) [41] developed a spatial decision-support system framework for suitability analysis of wind farm sites in northwest Ohio, USA. They determined exclusion criteria and evaluation criteria under environmental and economic aspects and then weighted based on the study group composed of students. They combined the weighted linear combination (WLC) method with fuzzy logic and also incorporated GIS into their framework. In contrast to most studies, their primary focus was on developing and testing the methodology while providing a user-friendly interface to help non-experienced GIS users. Ali et al. (2017) [50] studied the on-shore wind farm site selection in South Korea using GIS and AHP and fuzzy triangular numbers (FTN) methods. The most important criteria were wind speed and slope. They found that 0.7% of South Korea was rated as the most suitable area. Sánchez-Lozano et al. (2016) [23] proposed using GIS to determine available locations, fuzzy AHP to obtain the weights of the criteria, and fuzzy TOPSIS to rank the alternatives. They highlighted that the best alternatives are independent of the weights of the criteria according to sensitivity analysis. The results were then compared with fuzzy weighted sum model (WSM), and fuzzy revised AHP methods, and a similar priority order was obtained. Latinopoulos and Kechagia (2015) [49] developed a model to evaluate optimal wind farm sites, including licensed but not yet constructed project sites in Kozani, Greece. They excluded infeasible sites, selected evaluation criteria under technical, economic, and social factors, represented evaluation criteria as fuzzy sets, used MCDM methods, namely, AHP and WLC, and calculated the suitability index. The study revealed that more than 12% of the study area has an acceptable suitability index, showing that the number of current projects is below the wind energy potential of the region. Solangi et al. (2018) [69] integrated AHP for comparison of criteria and fuzzy TOPSIS for the selection and prioritization of alternatives. Rehman et al. (2019) [70] incorporated entropy with PROMETHEE method. Considering the geographical location of previous studies, the majority of wind farm location research mostly focused on Europe, along with many studies conducted in Asia. However, countries in America and Africa were underrepresented, except studies from the United States, West African countries, and Ecuador. On the other hand, no studies could be found for Australia or Antarctica, as mentioned by Spyridonidou and Vagiona (2020) [28]. Along with scientific interest, wind energy development and installation levels are higher in Europe and Asia, which shows a relation between development in wind energy and scientific research [28]. Most studies investigated large-scale areas, which are national and regional scales. There were also some studies investigating a regional unit, local area, or site-specific area [28].

The extensive literature review reveals that the shift towards novel objective methodologies directly addresses the limitations and criticisms of traditional MCDM methods, offering more reliable and comprehensive alternatives for decision making. These methods enable a detailed evaluation and ranking of alternatives, considering multiple criteria and their corresponding importance, making them an essential tool for decision making in complex scenarios [71]. CRITIC provides an objective method to determine the weights of criteria based on the variability and conflict among criteria, potentially reducing bias and improving decision quality, unlike AHP, which relies on subjective judgments for determining the weights of criteria [72]. Among the novel methods, MAIRCA is particularly valuable for its ability to effectively handle both qualitative and quantitative objectives, making it a widely preferred method in various research fields and showing its versatility and dependability in complicated decision-making situations [51,73].

Another novel MCDM method is CoCoSo. The versatility and resilience of this approach in tackling diverse problems have been extensively documented, with its dependability in producing desired results being specifically highlighted in multiple studies [74,75]. There is a limited but noteworthy body of research on CoCoSo for wind farm site selection. This includes studies conducted by Deveci et al. [76] in Norway, as well as research by Rong and Yu [77] in China for offshore wind farm site selection, which demonstrate the potential of the method [76,77].

In summary, the reviewed on-land wind farm selection studies made contributions to the literature by utilizing several MCDM methods and GIS; however, this study proposes a rather different approach than provided in the existing literature. The criteria weights are calculated using the CRITIC method and novel tools of MCDM are implemented, using CoCoSo and MAIRCA to develop and validate the model. Then, GIS tools are used to generate the map that provides the exact location of the suitable areas with a specific coordinate, which has not been assessed yet in previous studies. Another contribution of the study is to analyze Djibouti, which is one of the underrepresented countries in the literature. The primary objective of this study is to develop a comprehensive GIS-based MCDM framework to evaluate the wind farm site selection problem in Djibouti.

3. Research Methodology

This study implements a GIS-based integrated multi-criteria decision-support system to determine the most suitable sites for the installation of wind farms in Djibouti. The methodology of this study integrates spatial analysis and decision-making methods to evaluate potential wind farm sites in accordance with a variety of criteria. In order to accomplish the objectives of this study, the subsequent tasks were carried out:

(i)

Analysis of existing literature regarding the selection of suitable sites for wind farms: A thorough examination of current research was performed to collect data on the criteria influencing the selection of wind farm sites and the methodologies employed during the selection process.

(ii)

Determination of criteria for selecting suitable sites for wind farms: In order to determine the most suitable sites for wind farm installation, seven essential criteria were identified: wind velocity, changes in wind direction, ground slope, distance to urban areas, distance to road network, distance to energy transmission networks, and land use. The seven criteria offer a thorough framework for evaluating the suitability of sites for wind farm installation.

(iii)

Acquisition of data from the study area in accordance with the specified criteria: A variety of sources, including meteorological databases, GIS, and national databases, were utilized to acquire pertinent data from the study area for the identified criteria. The accuracy and credibility of the data are essential to the quality of the results of the study. Consequently, all relevant data were diligently acquired and verified for accuracy prior to their integration into the GIS.

(iv)

Determining the relative importance of the criteria by the CRITIC method: The CRITIC method was implemented to objectively ascertain the relative importance of each of the identified criteria. This method allows decision makers to comprehend the importance of the criteria employed in wind farm site selection by considering both “contrast intensity” and “correlation between criteria”, eliminating the necessity for any subjective judgment.

(v)

Application of the CoCoSo method to rank alternative sites for wind farms within the designated study area: The CoCoSo method was utilized on the data matrix, incorporating the relative importance of the criteria derived from the CRITIC method to prioritize the alternative sites for the wind farm. This method enables decision makers to pinpoint the most suitable sites for wind farms by providing a balanced compromise among all pertinent criteria.

(vi)

Application of the MAIRCA method to validate the results of the CoCoSo method through comparative analysis: To enhance the reliability of the results derived from the CoCoSo method, validation was conducted using the MAIRCA method. This comparative analysis provides a more nuanced comprehension of the decision-making process and identifies any inconsistencies, if present.

(vii)

Examination of the limitations of the study and provision of suggestions for future research: The limitations of this study are addressed, and suggestions for future research are provided to enhance transparency, guide further research, and augment the real-world importance of research findings.

The CRITIC, CoCoSo, and MAIRCA methods are summarized in the subsequent subsections, which also provide an overview of the steps involved in their computations.

3.1. CRITIC Method

The CRITIC method, a multi-criteria decision-making technique formulated by Diakoulaki et al. (1995) [78], focuses on determining the importance of criteria according to their interdependencies and significance. The principle behind the method is that the importance of a criterion can be evaluated through examining its correlation with other criteria and its variability (i.e., standard deviation). This dual consideration enables a more nuanced comprehension of the interaction and influence of criteria within the decision-making context. The CRITIC method integrates variability (standard deviation) and interdependence (correlation coefficients) to prevent highly correlated criteria from dominating the weighting process, thus eliminating redundancy and enhancing the objectivity of the evaluation. Through its emphasis on variance and correlation, the CRITIC method ensures that each criterion provides unique, non-redundant information, thereby enhancing the objectivity and robustness of the decision-making process. Traditional methods may be unable to accurately represent the intricate relationships between criteria in certain circumstances; however, the CRITIC method is distinguished by its capacity to effectively manage complex correlations between criteria [79]. One of the notable advantages of the CRITIC method is its objectivity in determining the importance of criteria [80]. Through the utilization of statistical measures of correlation and variability, the method mitigates subjective biases that may result from expert judgment alone, thereby enhancing the reliability of the decision-making process. The method allocates weights based on statistical measures instead of subjective judgments, thereby reducing potential biases and ensuring that criteria with greater differentiation influence the decision-making process more substantially. The computation steps of the method are simply described below:

Step 1: Establishment of the decision matrix X with n number of alternatives (i = 1, …, n) and m number of criteria (j = 1, …, m).

$$X={\left[x_{ij}\right]}_{n*m}=\left[\begin{array}{ccc}{x}_{11}& \cdots & x_{1m}\\ ⋮& \ddots & ⋮\\ x_{n1}& \cdots & x_{nm}\end{array}\right]$$

where $x_{ij}$ is the performance value of alternative i on criterion j.

Step 2: Normalization of the components in the decision matrix based on the type of criteria (i.e., beneficial and cost).

If the type of criterion j is beneficial:

$$n_{ij}={\displaystyle \frac{x_{ij}-{x_i}^{min}}{{x_i}^{max}-{x_i}^{min}}}$$

If the type of criterion j is cost:

$$n_{ij}={\displaystyle \frac{{{x_i}^{max}-x}_{ij}}{{x_i}^{max}-{x_i}^{min}}}$$

where $n_{ij}$ is the normalized performance value of the alternative i with respect to the criterion j.

Step 3: Computation of the standard deviation (σ_j) for each criterion.

Step 4: Determination of the amount of information (C_j) in each criterion j.

$$C_j={\sigma }_j\sum _{k=1}^m\left(1-r_{jk}\right)$$

Step 5: Calculation of the importance of criteria (w_j).

$$w_j={\displaystyle \frac{C_j}{\sum _{j=1}^m{C}_j}}$$

3.2. CoCoSo Method

The CoCoSo method, introduced by Yazdani et al. (2019) [81], is a multi-criteria decision-making technique that integrates a variety of aggregation strategies to evaluate and rank alternatives based on multiple criteria. The method is based on the concept of compromise solutions, which are designed to balance conflicting criteria through a structured framework in order to identify the most appropriate alternative. It integrates components from the Weighted Aggregated Sum Product Assessment (WASPAS), Simple Additive Weighting (SAW), and exponentially weighted product (EWP) methods, enabling an adaptable decision-making process that considers both qualitative and quantitative data [82,83]. One of the notable advantages of the CoCoSo method is its effectiveness in managing incomplete and uncertain data. Its adaptable framework enables the incorporation of both objective and subjective data, providing a reliable option for decision makers who are confronted with complex problems. The procedural steps of the method are clearly outlined below [81]:

Step 1: Development of the decision matrix X with n number of alternatives (i = 1, …, n) and m number of criteria (j = 1, …, m).

$$X={\left[x_{ij}\right]}_{n*m}=\left[\begin{array}{ccc}{x}_{11}& \cdots & x_{1m}\\ ⋮& \ddots & ⋮\\ x_{n1}& \cdots & x_{nm}\end{array}\right]$$

where $x_{ij}$ is the performance value of alternative i on criterion j.

Step 2: Normalization of the components in the decision matrix based on the type of criteria (i.e., beneficial and cost).

If the type of criterion j is beneficial:

$$n_{ij}={\displaystyle \frac{x_{ij}-{x_i}^{min}}{{x_i}^{max}-{x_i}^{min}}}$$

If the type of criterion j is cost:

$$n_{ij}={\displaystyle \frac{{{x_i}^{max}-x}_{ij}}{{x_i}^{max}-{x_i}^{min}}}$$

where $n_{ij}$ is the normalized performance value of the alternative i with respect to the criterion j.

Step 3: Computation of the power of weighted comparability (P_i) and sum of weighted comparability (S_i) sequence scores for each alternative i, respectively.

$$P_i=\sum _{j=1}^n({n_{ij})}^{w_j}$$

$$S_i=\sum _{j=1}^n\left(w_j{n}_{ij}\right)$$

where w_j denotes the weight of the criterion j.

Step 4: Calculation of the appraisal score for each alternative i employing three combination strategies.

$$a_{ia}={\displaystyle \frac{P_i+S_i}{\sum _{i=1}^m\left(P_i+S_i\right)}}$$

$$a_{ib}={\displaystyle \frac{S_i}{min\left(S_i\right)}}+{\displaystyle \frac{P_i}{min{(P}_i)}}$$

$$a_{ic}={\displaystyle \frac{\lambda P_i+{(1-\lambda )P}_i}{\lambda max\left(S_i\right)+(1-\lambda )P_i}}$$

where the parameter λ is determined by decision makers within the range of 0 to 1 (λ = 0.5 by default).

Step 5: Determination of the final appraisal score (A_i) for each alternative i.

$$A_i=({a_{ia}{a}_{ib}{a}_{ic})}^{{\displaystyle \frac{1}{3}}}+{\displaystyle \frac{1}{3}}{(a}_{ia}{+a}_{ib}+a_{ic})$$

Step 6: Rank alternatives in descending order based on A_i values. As a result, the alternative with the highest appraisal score is determined as the most appropriate for the final decision.

3.3. MAIRCA Method

The MAIRCA method, introduced by Pamučar et al. (2014) [73], is a multi-criteria decision-making technique designed to evaluate and rank alternatives according to multiple criteria. The primary objective of the method is to assess alternatives based on their closeness to an ideal solution [51]. The method relies on the comparison of theoretical (ideal) and empirical (real) ratings of alternatives, enabling decision makers to determine the most appropriate alternative by measuring the distance from the ideal solution [84]. A notable characteristic of the MAIRCA method is its dependence on a distinctive normalization procedure and an aggregating function that improves computational efficiency relative to traditional MCDM methods [85]. MAIRCA distinguishes itself from other MCDM methods by its particular emphasis on the gap between ideal and empirical priorities. Although older MCDM methods such as TOPSIS and VIKOR attempt to rank alternatives based on their closeness to an ideal solution, MAIRCA’s approach offers a more sophisticated assessment of the distance to both ideal and real scenarios, potentially facilitating more informed decision making [86]. The steps for the application of the method are explicitly specified as follows [30]:

Step 1: Formation of the decision matrix X with n number of alternatives (i = 1, …, n) and m number of criteria (j = 1, …, m).

$$X={\left[x_{ij}\right]}_{n*m}=\left[\begin{array}{ccc}{x}_{11}& \cdots & x_{1m}\\ ⋮& \ddots & ⋮\\ x_{n1}& \cdots & x_{nm}\end{array}\right]$$

where $x_{ij}$ is the performance value of alternative i on criterion j.

Step 2: Determination of the preferences ($P_{A_i}$) of decision makers for alternatives. The method assumes that the decision maker (DM) is neutral when evaluating alternatives. Simply put, the decision maker is equally distant from each alternative, rendering the presented alternatives of equal importance.

$$P_{A_i}={\displaystyle \frac{1}{n}}$$

Step 3: Calculation of the elements ${{(t}_p}_{ij}$) of theoretical matrix (T_p).

$${T_p={\left[{t_p}_{ij}\right]}_{n*m}=t}_{p_{ij}}=P_{A_i}\times w_j$$

where w_j represents the weight of the criterion j.

Step 4: Calculation of the elements ${{(t}_r}_{ij}$) of real matrix (T_r) based on the type of criteria (i.e., beneficial and cost).

If the type of criterion j is beneficial:

$$t_{r_{ij}}=t_{p_{ij}}\times \left({\displaystyle \frac{x_{ij}-{x_i}^{min}}{{x_i}^{max}-{x_i}^{min}}}\right)$$

If the type of criterion j is cost:

$$t_{r_{ij}}=t_{p_{ij}}\times \left({\displaystyle \frac{x_{ij}-{x_i}^{max}}{{x_i}^{min}-{x_i}^{max}}}\right)$$

Step 5: Calculation of the elements ${(g}_{ij}$) of the total gap matrix (G).

$$G={\left[g_{ij}\right]}_{n*m}=t_{p_{ij}}-t_{r_{ij}}$$

Step 6: Calculation of the final values of criteria functions (Q_i) with respect to the alternatives.

$$Q_i=\sum _{j=1}^m{g}_{ij}$$

Step 7: Rank alternatives in ascending order based on Q_i values. As a result, the alternative with the highest appraisal score is determined as the most appropriate for the final decision.

4. Implementation of the Proposed Decision-Support System: The Case of Djibouti

A case study demonstrates the application of the proposed decision-support system by end users, its practicality, and its effectiveness. The case study focuses on the determination of the most appropriate site for the construction of wind farms in Djibouti. In East Africa, Djibouti encompasses an area of 23,200 km² and features a sub-tropical desert climate marked by elevated temperatures and drought (Figure 1). Djibouti is an ideal candidate for renewable energy initiatives, particularly wind energy, in order to respond to the increasing demand for electricity and diminish its reliance on imported energy due to its strategic proximity to the Red Sea and plentiful wind resources.

The following subsections provide a detailed elucidation of the application of the proposed decision-support system in determining the most appropriate site for the development of wind farms in Djibouti:

4.1. Determination of Criteria for the Selection of Appropriate Sites for Wind Farms

A comprehensive literature review was conducted to determine the criteria that influence the selection of sites for wind farm construction. Following a review of the literature, seven main criteria—wind velocity, changes in wind direction, ground slope, distance to urban areas, distance to road network, distance to energy transmission networks, and land use—were identified to evaluate the suitability of potential sites for the construction of wind farms. A brief discussion of the impact of each criterion on the selection of a site for a wind farm is provided below.

Wind velocity (C₁): The wind velocity criterion plays a significant role in the selection of sites for wind farms since it has a direct impact on the power output of these facilities. The feasibility of generating wind energy at a specific site is frequently determined by the average long-term wind speed, which is an essential variable in evaluating the suitability of a wind farm site. As wind velocities increase, electricity generation increases. Consequently, sites with elevated wind velocities are more appropriate than those with diminished wind velocities [41,45,58,87].

Changes in wind direction (C₂): Wind direction refers to the angle of the wind velocity vector relative to the geographical north, quantified in degrees in a clockwise manner. The standard deviation of wind velocity direction defines the changes in wind direction over time. This metric is essential for measuring levels of directional turbulence in the wind velocity. Furthermore, the factors that are detrimental to the assessment of wind farm sites are the sustained low wind velocity and frequent changes in wind direction. As a result, sites with infrequent wind direction changes or more consistent winds are more appropriate to those with frequent wind direction changes [88].

Ground slope (C₃): The efficiency, cost-effectiveness, and safety of wind power plant installations are all significantly influenced by the slope of the site. While steep slopes make wind turbine construction and maintenance challenging, on the other hand, slight slopes make it unchallenging [89]. To optimize operational efficiency and reduce installation challenges, wind turbines are frequently located on sites with gentle slopes. Thus, sites with gentler slopes are more appropriate than those with steeper slopes [90,91].

Distance to urban areas (C₄): In the process of selecting wind farm sites, the proximity to urban areas is a critical factor that affects considerations for the environment, efficiency in operations, and community acceptance. To mitigate possible disagreements with local populations and to adhere to regulatory requirements, such as noise limits, it is imperative to maintain an appropriate distance from urban centers [92,93]. In addition, it is imperative to prioritize the well-being of inhabitants and the possible interruptions to day-to-day life caused by wind turbine operations. Therefore, locations that are farther from urban areas are more appropriate than those closer to them [16,94].

Distance to road network (C₅): The proximity of wind farm sites to road networks substantially affects construction and maintenance expenses, in addition to operational efficiency [41,45]. Sites close to road infrastructure facilitate equipment transportation, diminish maintenance and emergency response expenses, and improve overall economic feasibility. Moreover, proximity to current roads reduces habitat fragmentation, mitigates wildlife disturbances, and lessens the necessity to construct new roads, thereby promoting more sustainable land use. Accordingly, sites that are located near road networks are more appropriate than those that are situated at a greater distance [95,96,97].

Distance to energy transmission networks (C₆): The distance to energy transmission networks refers to the proximity of the site to the nearest energy transmission line, irrespective of the size of the wind farms or the type of grid (high voltage or medium voltage). The proximity to energy transmission networks substantially influences the operational efficiency, cost-effectiveness, and grid integration of wind farms. Wind farms located near current transmission infrastructure experience diminished transmission losses, reduced cabling costs, and enhanced grid connectivity, thereby improving their economic feasibility [22]. Closeness to transmission networks enhances the reliability of wind power generation and promotes the effective integration of it into the grid, advancing the transition in the direction of a more sustainable energy composition. Consequently, locations that are near transmission grids are more appropriate than those that are located at a distance [54,98].

Land use (C₇): The availability of land is a crucial determinant influencing the feasibility of wind farm projects, requiring meticulous evaluation to prevent conflicts with current land uses, including agriculture and residential developments. Analyzing land use is crucial for the identification of appropriate locations for wind farms that mitigate the detrimental impacts on land functionality and local ecosystems. This also allows developers to achieve energy generation objectives with environmental sustainability [99]. Furthermore, a comprehensive evaluation of land use promotes an understanding of the social and economic impacts of wind farms on cultural landscapes, local communities, and wildlife habitats, thereby guaranteeing adherence to regulatory standards and conservation objectives. Wind farms are more suitable on sites characterized by short vegetation rather than those characterized by long vegetation [9,100].

4.2. Acquisition of Data from the Study Area for Seven Criteria

Upon determining the seven criteria that could potentially affect the decision to select appropriate sites for the construction of wind farms, the next step involves the acquisition of the actual data related to the criteria. The quality of the research outcomes greatly depends on the reliability and accuracy of the data. Hence, all necessary data were carefully gathered and double-checked for correctness before being analyzed in the GIS. The preliminary phase in the data acquisition procedure is the removal of inappropriate alternatives. In this context, the analysis intentionally removed all locations designated as biologically, ecologically, spiritually, or culturally significant. In the case of Djibouti, there are 23,299,806 potential sites to be evaluated for wind farm construction. Data acquisition for 23,299,806 potential sites was performed using a diverse array of measurement units (i.e., m/s, °, %, km) in seven criteria. The ECMWF Reanalysis v5 (ERA5) datasets were used to acquire the wind velocity data. The Global Land Data Assimilation System (GLDAS) datasets were utilized to acquire land-use data, whereas NextGIS datasets were employed to acquire ground slope, distance to power transmission networks, distance to road networks, and distance to urban areas. All criteria utilized actual data except for the land-use criterion, which employed a nominal scale (i.e., (1) wetland/settlement, (2) forest, (3) shrub land, (4) agricultural land, (5) bare land) during the data acquisition process. Following the completion of the data acquisition process related to the criteria from the study area, an initial decision matrix was generated, including the performance values of 23,299,806 potential sites within seven criteria.

Table 2. Symbolic representation of the initial decision matrix.

Alternative ID	Longitude	Latitude	C1 (m/s)	C2 (°)	C3 (%)	C4 (km)	C5 (km)	C6 (km)	C7 (Scale)
1	43.297225020	12.791963050	4.12	90.91	0.00	0.19	20,542.33	132,062.52	4
2	43.297502790	12.791963050	4.12	90.92	0.00	0.19	20,567.20	132,072.94	4
3	43.296391720	12.791685280	4.12	90.90	0.00	0.18	20,449.32	132,008.08	4
4	43.296669490	12.791685280	4.12	90.90	0.00	0.18	20,473.95	132,015.86	4
5	43.296947250	12.791685280	4.12	90.91	0.00	0.19	20,499.27	132,024.13	4
.	.	.	.	.	.	.	.	.	.
.	.	.	.	.	.	.	.	.	.
.	.	.	.	.	.	.	.	.	.
23,299,802	42.855853840	11.316466670	6.93	81.53	3.75	62.72	1655.17	1994.42	5
23,299,803	42.856131600	11.316466670	6.93	81.53	2.05	62.46	1648.22	1995.09	5
23,299,804	42.856409370	11.316466670	6.94	81.53	0.78	61.70	1641.35	1995.87	5
23,299,805	42.856687130	11.316466670	6.94	81.52	0.26	61.54	1634.53	1996.79	5
23,299,806	42.856964900	11.316466670	6.94	81.52	1.10	61.60	1627.94	1994.17	5

Note = C₁: Wind velocity, C₂: Changes in wind direction, C₃: Ground slope, C₄: Distance to urban areas, C₅: Distance to road network, C₆: Distance to energy transmission networks, C₇: Land use.

symbolically presents the performance values of a subset of 23,299,806 potential sites within seven criteria, corresponding to their longitude and latitude data.

4.3. Application of the CRITIC Method to Determine the Relative Importance of the Seven Criteria

After concluding the data acquisition process, the CRITIC method was employed on the decision matrix comprising the performance values of the suitable locations among 23,299,806 potential sites within seven criteria to objectively ascertain the relative importance of the criteria. MATLAB 9.13 (R2022b) was utilized to code and implement the CRITIC method. Following the execution of the CRITIC method calculation steps to the decision matrix, the relative importance of the seven criteria was obtained without the use of any subjective judgment.

Table 3. The relative importance of the criteria for the selection of wind farm sites.

Criterion ID	Criterion Name	Weight
C1	Wind velocity	0.174
C2	Changes in wind direction	0.161
C3	Ground slope	0.065
C4	Distance to urban areas	0.164
C5	Distance to road network	0.159
C6	Distance to energy transmission network	0.170
C7	Land use	0.107

illustrates the objective relative importance of the seven criteria. It should be noted that the relative importance of the seven criteria obtained from the CRITIC method is an essential input utilized in the computational steps of the CoCoSo and MAIRCA methods.

The findings shown in Table 3 reveal that the most important criteria are the wind velocity and the distance to energy transmission networks [101]. In contrast, ground slope and land use were identified as the least important criteria relative to the others. The findings are consistent with the findings of earlier studies (e.g., [10,102,103]). It is important to acknowledge that the CRITIC method determines the relative importance of each criterion based on the decision matrix acquired from the study area [104]. Consequently, the relative importance of the criteria may differ in different application areas. Moreover, this case demonstrates that the CRITIC method is applicable to any study area where actual data can be acquired for the specified criteria.

4.4. Application of the CoCoSo Method to Rank Potential Alternative Sites for Wind Farms

The ranking of the suitable locations out of 23,299,806 potential sites was determined using the CoCoSo method after the CRITIC method was used to determine the relative importance of the criteria in the wind farm site selection problem. Similar to the CRITIC method, MATLAB 9.13 (R2022b) was utilized to code and implement the CoCoSo method for the current problem. In the context of this selection problem, changes in wind direction (C₂), ground slope (C₃), distance to road network (C₅), and distance to energy transmission networks (C₆) serve as cost criteria, with lower values being desirable. On the other hand, wind velocity (C1), distance to urban areas (C₄), and land use (C₇) are beneficial criteria for which higher values are desirable. Consequently, wind velocity (C₁), distance to urban areas (C₄), and land use (C₇) are maximized, whereas the other criteria are minimized. Upon executing the CoCoSo method computation steps on the decision matrix, the ranking of the suitable locations out of 23,299,806 potential sites was determined.

Table 4. The top ten most appropriate sites for wind farm construction determined through the CoCoSo method.

Longitude	Latitude	Si	Pi	${\mathit{a}}_{\mathit{i}\mathit{a}}$	${\mathit{a}}_{\mathit{i}\mathit{b}}$	${\mathit{a}}_{\mathit{i}\mathit{c}}$	${\mathit{A}}_{\mathit{i}}$	Rank
43.084455780	11.438683990	0.80	6.58	1.71 × 10−7	5.19	0.99	2.07	1
43.084733540	11.438683990	0.80	6.58	1.71 × 10−7	5.19	0.99	2.07	2
43.085011310	11.439239530	0.80	6.58	1.71 × 10−7	5.19	0.99	2.07	3
43.085289080	11.439239530	0.80	6.58	1.71 × 10−7	5.19	0.99	2.07	4
43.084733540	11.438961760	0.80	6.58	1.71 × 10−7	5.19	0.99	2.07	5
43.084178010	11.438683990	0.80	6.58	1.71 × 10−7	5.19	0.99	2.07	6
43.085011310	11.438683990	0.80	6.58	1.71 × 10−7	5.19	0.99	2.07	7
43.084455780	11.438961760	0.80	6.58	1.71 × 10−7	5.19	0.99	2.07	8
43.085011310	11.438961760	0.80	6.58	1.71 × 10−7	5.19	0.99	2.07	9
43.083344710	11.438683990	0.80	6.58	1.71 × 10−7	5.19	0.99	2.07	10

Note = S_i: Sum of weighted comparability sequence scores, P_i: Power of weighted comparability, $a_{ia},a_{ib},a_{ic}$: Combination strategies, A_i: Final appraisal score.

presents the top ten most appropriate locations for a wind farm among 23,299,806 prospective sites, assessed according to seven criteria with their relative importance. Additionally, Figure 2 precisely illustrates the locations of the top ten most appropriate sites.

According to Table 4 and Figure 2, the ten most appropriate sites, evaluated based on the performance of each alternative across the seven criteria, are situated in Holhol. In other words, the top ten most appropriate sites represent a set of coordinates that provide an area to focus on for constructing a wind farm. The proximity of the top ten most appropriate sites supports the validity of the results derived from the proposed decision-support system. In addition, the findings align with the research conducted by Abdi et al. (2024) [9]. A final suitability map for the construction of a wind farm in Djibouti was obtained by Abdi et al. (2024) [9] by taking into account seven criteria. The final suitability map by Abdi et al. (2014) [9] indicates that the Holhol region is among the most suitable locations for wind farms. In this study, the primary goal differs from that of the research by Abdi et al. (2024). The primary goal is the comprehensive ranking of all potential alternative sites for wind farm construction based on their performance across seven criteria, followed by the exact determination of the most appropriate site that comprises a set of coordinates.

Upon meticulous analysis of the performance of the top ten most appropriate sites across the seven criteria, their common characteristics can be identified as follows:

• Favorable wind speed (7.80 m/s);
• Low wind direction change (80.89°);
• Slightly sloped the ground (2.03%);
• Distance from urban areas (61.93 km);
• Proximity to road networks (83.44 km);
• Proximity to energy transmission networks (126.35 km);
• Barren vegetation in land use (5: bare land).

When examining the performance of the top ten most appropriate sites on the seven criteria, it is important to note that they do not have the most favorable performance on each criterion. However, when the seven criteria are evaluated simultaneously, along with their relative importance, they are determined as the most appropriate among 23,299,806 potential sites.

4.5. Application of the MAIRCA Method to Validate the Results of the CoCoSo Method

Following the ranking of the suitable locations among 23,299,806 potential sites, for wind farm construction via the CoCoSo method, validation was conducted using the MAIRCA method to improve the reliability of the results derived from the CoCoSo method. This study selected the MAIRCA method for comparative analysis due to its mathematically stable framework, which effectively accommodates variations in criteria attributes, rendering it a robust tool for complex decision-making scenarios where reliable and consistent rankings are essential [30,51]. This comparative analysis enables decision makers to acquire a more nuanced comprehension of the decision-making process and to identify any inconsistencies in the results. The MAIRCA method was coded and executed for the present problem using MATLAB 9.13 (R2022b), just like the CRITIC and CoCoSo methods. When applying the MAIRCA method to the identical site selection problem, the types of the seven criteria and their relative importance derived from the CRITIC method are consistent with those utilized in the CoCoSo method. Following the execution of the MAIRCA method calculation steps on the decision matrix, the ranking of the suitable locations among 23,299,806 potential sites was determined.

Table 5. The top ten most appropriate sites for wind farm construction determined through the MAIRCA method.

Longitude	Latitude	Qi	Rank
43.084455780	11.438683990	3.05 × 10−8	1
43.085289080	11.439239530	3.05 × 10−8	2
43.085011310	11.439239530	3.05 × 10−8	3
43.084733540	11.438683990	3.05 × 10−8	4
43.084733540	11.438961760	3.06 × 10−8	5
43.085011310	11.438683990	3.06 × 10−8	6
43.084178010	11.438683990	3.06 × 10−8	7
43.085566840	11.439239530	3.06 × 10−8	8
43.084455780	11.438961760	3.06 × 10−8	9
43.092511010	11.439239530	3.06 × 10−8	10

Note = Q_i: Final appraisal score.

presents the top ten most appropriate locations for wind farms among 23,299,806 prospective sites that were assessed according to seven criteria with their relative importance using the MAIRCA method. Furthermore, Figure 3 accurately demonstrates the positions of the ten most appropriate sites within the study area for wind farm construction.

5. Discussion

The discussion on the findings of the study encompasses (1) the assessment of the relative importance of the criteria, (2) the accuracy of the proposed decision-support system, and (3) the implication of the proposed decision-support system.

5.1. Assessment of the Relative Importance of the Seven Criteria

The results of the CRITIC method revealed that wind velocity and distance to energy transmission networks are prioritized over other criteria, including ground slope and land use, when determining the relative importance of criteria for wind farm site selection in Djibouti. The results, as presented in Table 3, can be attributed to several primary reasons associated with Djibouti’s unique characteristics, including geography and infrastructure.

In accordance with the climatic and geographical characteristics of Djibouti, which encompass regions with substantial wind resources that can facilitate efficient energy generation, the prioritization of wind velocity as a criterion is appropriate [105]. Elevated wind velocity is a fundamental technical criterion for suitable wind farm sites, as it directly influences energy production efficiency [106]. In a country such as Djibouti, where energy accessibility and security are paramount issues, utilizing high wind velocity is crucial to optimize electricity generation capacity and satisfy increasing energy requirements [7]. To meet the increasing energy demands and optimize electricity generation potential in a country such as Djibouti, where energy access and security are critical concerns, it is imperative to harness high wind velocity [5]. As a result, this criterion was assigned the highest weight, which is indicative of its critical role in the evaluation of potential sites’ suitability.

In the same vein, the distance to energy transmission networks has become an important criterion as a result of the current energy infrastructure in Djibouti, which necessitates optimization to facilitate large-scale renewable energy projects. Closeness to transmission lines diminishes both construction and maintenance costs associated with connecting new wind farms to the grid, promoting a more cost-effective and dependable energy supply to urban and rural areas alike [43,53]. In an evolving energy market, reducing logistical and financial obstacles is crucial for attracting investment and fostering sustainable growth, underscoring the importance of this criterion [107].

Conversely, ground slope was identified as one of the least important criteria within the framework of this study. While slope can influence the construction and stability of wind turbines [58], Djibouti’s largely flat topography mitigates the importance of this criterion in site selection [108]. Due to the minimal topographical variation throughout much of the country, the impact of slope on construction feasibility and cost is comparatively minor, which accounts for its lower relative importance.

Land use was also assigned a lower importance, which is indicative of the relatively limited land-use restrictions in Djibouti. The low population density and extensive arid regions result in diminished competition for land, allowing much of the area to be developed without substantially interfering with current land uses [109]. The diminished incidence of land-use conflict, coupled with the presence of uninhabited or sparsely populated regions, results in a lower prioritization of land use as a criterion within the decision-support model.

The weights derived from CRITIC offer a robust foundation for assessing the suitability of each site, highlighting criteria that directly influence the technical and economic viability of wind farm development in Djibouti. This assessment framework aligns with the country’s energy policy objectives and addresses Djibouti’s distinct environmental and infrastructural attributes, enabling an effective, context-specific approach to sustainable energy planning.

5.2. Accuracy of the GIS-Based Integrated Multi-Criteria Decision-Support System

The strong alignment between the CoCoSo and MAIRCA methods serves as validation for the accuracy and reliability of the GIS-based integrated multi-criteria decision-support system developed in this study. Table 5 demonstrates that eight of the top ten ranked alternatives are present in the results of both methods, which is equivalent to an 80% similarity. The notable consistency between the outcomes of the two distinct MCDM methods provides compelling evidence for the model’s accuracy and reliability in determining the most appropriate locations for wind farms in Djibouti.

The observed consistency indicates that the model’s results are not significantly affected by methodological variations, despite each method employing a unique approach. This level of consistency implies that the site selection results are dependable, regardless of the MCDM method used to evaluate them, a critical factor in real-world decision making.

The significant similarity between methods highlights the model’s applicability, especially in contexts such as Djibouti, where wind farm site selection must consider diverse geographic, technical, and infrastructural limitations. The reliability of the proposed decision-support system, demonstrated by the strong consistency between the methods, enhances stakeholders’ confidence in site selection for wind energy projects. Furthermore, the agreement between the results of the methods demonstrates that the decision-support system can offer a more structured approach to the development of renewable energy in the region.

Ultimately, the GIS-based integrated multi-criteria decision-support system achieves a high level of accuracy, thus allowing for the reliable identification of appropriate sites for wind farm development. The results of the proposed system are confirmed by the consistency between CoCoSo and MAIRCA rankings, demonstrating its practicality and adaptability as a tool for use in wind energy projects.

5.3. Implication of the GIS-Based Integrated Multi-Criteria Decision-Support System

The GIS-based integrated multi-criteria decision-support system that was developed in this study has substantial implications for sustainable development and wind energy planning in Djibouti. This system employs advanced spatial analysis and structured decision-making methods to systematically determine the most appropriate sites for wind farm installations based on a comprehensive set of criteria. The integration of GIS technology provides a robust instrument for spatial data visualization and analysis, enabling the evaluation of extensive geographical regions and complex datasets. This enables a more accurate assessment of criteria such as wind velocity, changes in wind direction, distance to urban areas, distance to road network, and land use, which are essential for identifying viable wind energy sites that are in accordance with both technical requirements and community considerations.

The utilization of the CRITIC and CoCoSo methods within this GIS framework guarantees an objective, data-driven approach to the prioritization of potential sites, the reduction of subjectivity in the determination of the weights of criteria, and the enhancement of transparency in the decision-making process. The CRITIC method quantifies the importance of each criterion through contrast intensity and inter-criteria correlation, thereby reducing human bias and enhancing result reliability [79]. In a similar vein, the CoCoSo method provides a balanced compromise across all criteria [81], allowing stakeholders to pinpoint sites that optimize resource utility while avoiding problems with nearby urban areas or land use.

Moreover, the integration of the MAIRCA method for validation enhances the reliability of this decision-support system by cross-verifying the prioritization results. The dual-method validation strengthens confidence in the selected sites, rendering the results more relevant for policymakers and investors in pursuit of dependable data to facilitate wind farm development in Djibouti.

In conclusion, the GIS-based decision-support system proposed in this study not only supports Djibouti’s renewable energy objectives but also functions as a model for other regions with comparable socio-economic and geographic contexts. This decision-support system enhances the site selection process, facilitating efficient resource use and sustainable infrastructure planning, thereby supporting long-term energy security and economic growth in accordance with environmental preservation objectives.

6. Conclusions

The worldwide transition to renewable energy has highlighted the essential requirement for effective and sustainable energy production techniques. Wind energy, a highly promising renewable energy source, presents various advantages, such as sustainability, minimal operational costs, and the capacity to diminish greenhouse gas emissions. Nevertheless, the success of wind energy projects is contingent upon the selection of the most appropriate sites for wind farm construction, as determining the most suitable sites for wind farms is crucial for maximizing energy production, ensuring environmental sustainability, and achieving cost efficiency. This task necessitates the meticulous evaluation of multiple criteria, including wind velocity, distance to urban areas, and land use. Djibouti is a promising candidate for wind energy projects owing to its advantageous geographical and climatic conditions, yet it remains insufficiently explored regarding the identification of suitable sites for wind farms. In areas such as Djibouti, where significant wind energy potential aligns with the pressing demand for cost-effective and sustainable energy, the development of robust decision-support systems is essential for optimal resource utilization.

This study fulfills this need by providing a comprehensive framework for wind farm site selection in Djibouti, integrating the CRITIC and CoCoSo methods within a GIS-based decision-support system. The robustness and applicability of the proposed decision-support system for large-scale and data-intensive problems are demonstrated through the analysis of the suitable locations among 23,299,806 potential sites in accordance with seven criteria. The CRITIC method objectively determined the relative importance of the criteria, highlighting wind speed and proximity to energy transmission networks as most important, while ground slope and land use were deemed less important in comparison to other criteria. These findings, which are consistent with previous research, underline the necessity of adapting the proposed decision-support system to the unique characteristics of the study area, as the importance of criteria may differ across regions.

The CoCoSo method, utilized to evaluate alternative sites, yielded a precise prioritization of prospective sites for wind farms, maximizing beneficial criteria such as wind velocity and distance to urban areas while minimizing cost criteria such as changes in wind direction, ground slope and distance to road network. The top ten most appropriate sites for wind farm construction determined by the CoCoSo method were further validated through the MAIRCA method, thereby ensuring consistency and reliability in the ranking procedure. The comparative analysis revealed a strong consistency between the results of the two methods, illustrating the robustness of the proposed decision-support system and confirming its reliability for tackling large-scale, data-intensive multi-criteria decision-making problems.

This study contributes to the current body of knowledge on renewable energy planning by simultaneously integrating cutting-edge MCDM methods and GIS tools to solve the wind farm site selection problem on an unprecedented scale, with precise coordinates. The proposed decision-support system is designed to be flexible and applicable to a variety of contexts, as it is based on readily available real-world datasets, including wind velocity, land use, and distance to infrastructure. Providing a practical tool for renewable energy planning, this data-driven system enables the framework to be potentially applied to other regions with similar data availability. Professionals and researchers are encouraged to utilize the well-developed decision-support system that evaluates multiple criteria for identifying the most appropriate site for wind farm installation. Additionally, the proposed decision-support system provides professionals in Djibouti with actionable insights, thereby facilitating the transition to sustainable energy independence. Future research could enhance the framework by incorporating additional criteria, such as environmental and social factors, and by implementing it in other areas to evaluate the model’s adaptability and applicability. As future research, the proposed decision-support system, which was utilized for determining suitable on-land wind farm sites in the present study, could be modified to evaluate offshore wind energy potential in Djibouti, thereby exploring another promising opportunity for sustainable energy development and enhancing the applicability of the framework to various alternatives of renewable energy.