Spatial Analysis and Modelling of Wind Farm Site Suitability in Nasarawa State, North-Central Nigeria

There has been an increasing global and local interest in developing renewable, clean, and cheap energy towards achieving Goal number 7 of the Sustainable Development Goals (SDG). However, decisions involving suitable and sustainable locations for renewable energy projects remain an important task. This study employed Geographic Information System (GIS) and Mul-ti-Criteria Decision Analysis (MCDA) to spatially analyze and model wind farm site suitability in Nasarawa State. The aim is to integrate the environmental, social, and economic aspects of decision-making for identifying sustainable wind farm sites. The study distinguished between two sets of decision criteria: decision constraints and decision factors. The former defined the exclusion zones while the latter were standardized based on fuzzy logic to depict varying degrees of suitability across the State. The MCDA applied the weighted linear combination method, with relative weights generated through pairwise comparisons of the analytic hierarchy process to analyze three policy scenarios: equal weights, environmental/social priority, and economic priority scenario. A combination of resulting fied and delineated. The composite decision constraint revealed that wind farm projects would not be viable in more than half (57.58%) of the State. Wind speed was the major constraint and accounted for the exclusion of 46.25%, with a mean fuzzy membership value of 0.2008 indicating low suitability across the State. Also, the average acceptable wind farm location for the three-policy scenario was 33.33% of the entire study area. Lafia, Obi, Keana, Awe, Nasarawa-Eggon, Wamba and Kokona LGAs were the identified priority Local Government Areas (LGAs). However, only Lafia, Obi, and Nasara-wa-Eggon were consistent with changes in the policy objectives. All the priority LGAs have one or more of the most suitable parcels within their administrative boundaries except for Wamba. Despite the severe limitations of wind speed, substantial parts of Nasarawa State still provide great development potentials for wind energy. The “most suitable” locations in Lafia, Nasara-wa-Eggon, and Obi LGAs should have first consideration for the development of wind energy in the State.


Introduction
Nowadays, the increasing human population across the globe is exerting undue pressure on available non-renewable sources of energy known for their fast depletion and environmental degradation. Hence a shift of attention towards renewable and clean sources of energy by many national governments. Such sources include the wind, which has begun to take a high position in the global dialogue about energy production. A big question to profiting from the wind as a source of energy borders on identifying the most suitable location/s for wind farm construction, with respect to achieving the highest possible rates of sustainable electricity production.
Wind energy is considered one of the eco-friendliest and economically viable forms of renewable energy [1]. [2] stated that the expansion of wind farm developments could serve as a fundamental aspect in climate change mitigation and help reduce greenhouse gas emissions. According to [3], wind energy is among the lowest impact forms of electricity generation in terms of its benefits outlined at the regional and global level; it is neither associated with air nor water pollution, and operational costs are practically zero after building a turbine.
Although Africa is top-ranked as having the world's highest reserves of renewable energy sources [4], there have been reports of underutilization of wind as a renewable energy source in West Africa, most especially in Nigeria [5]. Research findings and wind data from Nigerian Meteorological Agency have shown that wind speed of about 8.07 meters per second (m/s) could be harnessed in the northern parts of Nigeria to provide enough energy for daily needs C. E. Ozim et al. Journal of Geographic Information System [6]. This gap between potential and extent of exploitation poses questions about barriers to wind energy development in Africa and Nigeria, in particular.
However, a laudable interest in wind energy is currently on the rise across the continent and several countries including Nigeria have planned installations of wind power. The Nigerian Renewable Energy Master Plan (REMP) seeks to increase the supply of renewable electricity from 13% of total electricity generation in 2015 to 23% in 2025 and 36% by 2030 [7]. In view of achieving this, it becomes imperative to spatially analyse the potential in renewable energy sources such as Wind, Waves, Solar, Tides, and Geothermal.
To make wind farm installations sustainable, it must possess characteristics that make its operation technically and economically feasible, and at the same time guarantee preservation of environmental and social values [8]. There has been a dearth of information in this regard for most parts of Nigeria. Thus, this study incorporated Multi-Criteria Decision Analysis into Geographical Information System (GIS) to spatially analyse and model site suitability for wind farm development in Nasarawa State, Nigeria.

Study Area
Nasarawa State occupies the space between latitudes 7˚45'00" and 9˚35'00" of the District; Akwanga, Nasarawa-Eggon and Wamba in the Northern District and Lafia, Keana, Doma, Awe and Obi in the Southern District [9]. The climate of Nassarawa State is typical of a tropical sub-humid climate having two distinctive seasons. The rainy season sets in from about the beginning of May and lasts until October. The dry season spans between November and April. The average hourly wind speed in Nasarawa experiences mild seasonal variation over the year. The windier part of the year lasts for about nine months, usually from 2 nd September to 30 th November. Figure 1 presents a map of Nasarawa State depicting its thirteen Local Government Areas (LGA), and Figure 2 shows the wind speed characteristics.

Sources of Datasets and Collection Methods
This study integrated Multi-Criteria Decision Analysis (MCDA) into Geographic Information System to develop decision support models for identifying potential sites for wind energy farm development in Nasarawa State. The MCDA applied Weighted Linear Combination (WLC) method using the criteria weights Journal of Geographic Information System   aided the criteria selection. The spatial formats (such as: shapefiles, raster and NetCDF) of the identified criteria were acquired and used as datasets for the study.
The datasets are publicly available and accessible from various secondary sources of data collection. The number of features included in the set of assessment criteria was based on data availability and regarded as suitable according to [13]. Data on forest reserves, wetlands, and inland waterbodies constituted the ecologically significant areas. Road and rail networks formed the safety restriction. Table 1 shows a summary of the datasets used in this study with their respective method of acquisition.

Analysis Procedure
Data analysis was based on MCDA and performed in ArcGIS 2. The study investigated three policy priority scenarios (equal weights, environmental/social priority, and economic priority scenario). In scenario 1, all the decision factors were assigned equal weights of 0.1667 (1/6), simulating equal importance. Conversely, in scenarios 2 and 3, the criteria weights were obtained by applying the analytic hierarchy process (AHP) based on the pairwise comparison of the relative importance of decision factors.
According to [19], the AHP is a commonly applied method in multi-criteria decision making, which uses a pairwise comparison approach (PWCA) to de- termine the weightage of the decision factors. Weights are derived based on the input by the user in terms of the relative importance of criteria, for example, how important is criteria "A" in comparison to criteria "B", "C", "D" etc. Its ability to convert subjective assessments to a set of overall weights was deemed appropriate for this study. AHP is designed to numerically evaluate the relative power of alternative factor(s) to achieve the overall goal [20]. In this study, we entered literature-based values of the relative importance of the decision factors into ArcGIS AHP extension 2.0 to obtain their corresponding weights.
We applied the nine-point continuous scale (Table 2) to determine the AHP pairwise comparisons values, and the results were recorded in a matrix as introduced by [21]. The AHP also provides a mathematical measure for checking inconsistency of judgments by calculating a consistency ratio (CR), where a CR value below 0.1 is considered acceptable [20].
Matrices of pairwise comparisons and associated weights for Scenarios 2 and 3 are presented in Table 3 and Table 4. Pairwise assessments were consistent in scenarios I and 2, with CR values of 0.013 and 0.018, respectively. Also, a summary of weights associated with the respective decision factors for the three policy scenarios is presented in Table 5.

Decision Constraints
In line with the standards and procedures presented in Table 6 and Figure 3, we Values of inverse comparison: if factor z has one of the above numbers assigned to it when compared with factor y, then a reversed comparison means that y would have the reciprocal.
Source: [21].  Wind resource is often a crucial consideration in assessing optimal locations for wind farm development: the better the resource, the more promising for potential power generation and project revenue [22]. According to [2] and the C. E. Ozim et al. Developing wind farms in protected areas like national parks, nature reserves, or state conservation areas endanger the natural environment [10]. In the study of [8], they pointed out that a 1000 m buffer around areas of ecological value should be applied, in addition to a 400 m buffer from water bodies. Therefore, our study created a 400 m buffer around the inland waters (constituting significant rivers and lakes) and a 1000 m buffer around other ecologically essential areas (forest reserves and wetlands). All the regions within the buffer were deemed unfeasible for wind farm development and hence excluded.
Furthermore, it is unsafe to situate wind farms too close to infrastructures like road and rail networks and airports. Hence, several safety distances have been applied in previous studies ranging from 150 m as the minimum for roads and rail networks [1] [24] to 500 m [10] [25]. Roads (single/dual carriageway: federal or state) and rail network, together with a buffer of 500 m, were considered not unfit for wind farm development and thus excluded.

Decision Factors
The decision factors were mainly standardized through the application of fuzzy sets theory. The fuzzy set theory acknowledges the possibility of having multiple groups simultaneously with the degree of each group described as a membership value. It plays a significant role in handling uncertainty by allowing the 'grouping of individuals into classes without sharply defined boundaries, which is suitable when describing "ambiguity, vagueness, and ambivalence in models of empirical phenomena" [24]. In the fuzzy set's theory, the degree of membership of an item p in a fuzzy subset P is defined in the range of membership value of zero (0) and one (1) through the application of a membership function (MF) [26]. Although several types of membership functions exist for describing fuzzy sets, this study adopted the linear process. The process was considered fitting for representing the varying degrees of site suitability at varying distances from features defined in the deciding factors. Furthermore, a linear function was also adopted in similar studies by [1] [10] and [24].  Table 7 presents the threshold values used to standardize each decision factor, and Figure 4 shows the suitability model. The decision factors are described as follows: Decision factor F1 describes the site suitability of the study area for wind energy farm with respect to wind speed. This factor assumed an Increasing Fuzzy   Energy generated by wind turbines is transmitted to the end users through a transmission/interconnection facility. Three types (transmission, distribution and direct connection to delivery point) of electricity network often serve this purpose depending on the voltage level of the generated electric-power [27]. We used Decision factor F3 to symbolise site suitability with respect to proximity to the national transmission network, based on a DFMF. From an economic viewpoint, [8] stated that wind farms must not be located further than 10,000 m away from the grid; yet, not closer than 100 m. In this study, the q-value and p-value for decision factor F3 were set to 10,000 m and 100 m, respectively.
Decision factor F4 was used to symbolise the site suitability with respect to should not be further than 10,000 m. This will enable easy access, and reduce the cost of construction and maintenance. Hence, we adopted 10,000 m as the q-value for decision factor F4, and 500 m for the p-value (in accordance with decision constraint C4).
The site suitability with respect to distance from ecologically significant areas was denoted decision factor F5 (IFMF). [10] suggested that in order to mitigate negative environmental impacts, it is suitable to assign a higher membership value to places further away from ecologically significant areas. In the study of [12], a p-value of 5000 m was considered to be appropriate in their fuzzy set for protection of bird habitat. Hence, in this study, a q-value of 1000m was set for decision factor F5 in line with the buffer distance defined under decision constraint C3, and the p-value was set to 5000 m.
Decision factor F6 represents the suitability of the study area with respect to the qualitative attribute of land cover/use. We indexed the nine (9) land cover/use classes of the study area using a Likert-type scale ranging from zero (0) to one (1) in a similar approach to the study of [10] and [1]. We evaluated the site suitability with reference to land cover/use based primarily on tree density. Higher density indicates greater environmental impact [28], clearing costs for developers [29], as well as risks of nearby vegetation affecting the wind speed and direction of flow [8].
Furthermore, to make statistical analysis possible, we reclassified the suitability index for each of the three policy scenarios to conform with the suitability classes described in Table 8. After the reclassification, we calculated the total area for the respective suitability class in each policy scenario. The areas deemed "satisfactory/acceptable" for wind farm development were defined based on the lower SI score of the "moderate suitability" class and hence described with the following suitability index score interval: 0.50 < SI ≤ 1.
We identified and delineated "priority areas", defined in this study as fascinating areas for further investigation by generating statistics for the suitability classes across the 13 LGA of the State. Then the six LGA that showed the highest prospect for wind farm development under each policy scenario were identified.
We achieved this by comparing the percentage of land classified as "high suitability" (SI > 0.75) with the total administrative area of each Local Government (LG) to determine the spread per LG. According to [10], the priority areas are of keen interest for further examination as they represent an increased probability of identifying suitable wind farm locations in the real-world context. Furthermore, we identified and delineated the most suitable locations for wind farm development by extracting the high suitability parcels in each priority LGA. Following the extraction, we converted the resulting raster to a polygon feature. Then the parcels with a total area of 5 km 2 or more were selected as the most suitable locations for wind farm development in the study area. The choice of the parcel size was in line with the minimum acceptable land area for wind farm development (5 km 2 ) given by [30]. For easy locating of these most suitable parcels, we calculated their centroid from which we created centre points and coordinates for individual parcels. We also determined the number of most suitable parcels per priority LGA under each policy scenario. Figure 5 and     environmental protection. Conversely, land areas excluded based on safety concerns and settlement preservation accounted for relatively small portions of the study area, representing 4.51% and 2.60%, respectively. Overall, more than half (57.58%) of the entire study area was deemed exclusion zones due to the collective influence of decision constraints. Figure 7 shows the exclusion of more

Suitability Index (SI) for the Policy Priority Scenarios
The final suitability index result for the three-policy scenario is presented in      Furthermore, this study presents more viable options for wind farm development than was reported by [1]. They found only 12% of their study area in Greece to represent "satisfactory" locations (SI > 0.50). These variations in total land areas deemed "satisfactory" across the compared studies could be due to the variations in the degree of restrictions/constraints exerted by each decision criteria, particularly wind speed, in the various study areas. Additional reasons could be the differences in weights assigned to respective decision factors, the spatial resolution of the input data, the magnitude of the study area and the regularised input data cell size used in the analysis.  The spatial locations of the priority LGAs identified under each of the three policy scenarios were delineated by Figures 11-13. A visual assessment of the maps revealed slight variations in the geographical locations of the priority areas.

Priority Areas for Wind Farm Development
These variations mirrored the final suitability index maps and suggest that the developed decision support model is susceptible to changes in policy objectives.
Furthermore, it is visually ascertainable that only three out of the seven priority LGAs maintained consistency with changes in policy objectives. These include: Lafia, Obi and Nasarawa-Eggon. According to [13], one of the various sources of uncertainty in Multi-Criteria Decision Analysis remains the allotting of weights to individual decision factors. Therefore, they argued that any attempt to identify the priority areas, locations whose suitability are susceptible to slight alterations in the relative importance weighting of criteria do not present the best possible option. Only those areas that maintained high suitability conditions across all investigated policy scenarios are in the best position for further analysis.
Consequently, it is imperative to deem Lafia, Obi and Nasarawa-Eggon LGAs the topmost options among the priority locations for wind farm development in Nasarawa State.    2, we found a total of 80 parcels of land that met the most suitable conditions.
Out of these parcels, 20 had their centroid located in Kokona, 19 in Lafia, 14 in Awe and 13 in Keana. The rest 14 parcels of land had their centroid located in Nasarawa-Eggon (11) and Obi (3) LGAs. In scenario 3, we found only two parcels of land considered the most suitable locations for wind farm development.
They are both located in Lafia. The most suitable parcels of land had almost identical geographical positions but the numbers varied across the priority areas.
This finding indicates relatively low sensitivity of the most suitable land locations to changes in policy objectives. It is also similar to the decision support tool developed by [10] for New South Wales in Australia.

Journal of Geographic Information System
This finding implies that Lafia had the most (27) locations deemed the most suitable areas across the three policy scenarios. Hence, it presents the best option for wind farm development in the State. Also, where environmental protection is not of paramount concern, Kokona LGA shows promising prospects with the 20 most suitable locations under policy scenario 2. Other LGAs that shows wind farm development potentials under scenario 2 are Awe, Keana and Nasarawa-Eggon. Obi LGA showed an unencouraging prospect under scenario 2. This poor prospect is attributable to the substantial number of forest reserves in the LGA.

Conclusions
This study developed decision support models to identify potential wind farm locations in Nasarawa State, Nigeria. Wind speed created severe limitations to the quest for sustainable wind farm locations in the State. Despite the barrier imposed by wind speed, substantial parts of the State still show great development potentials for wind energy farms, given that 36.34%, 42.44% and 21.22% of the entire State satisfied the fundamental conditions under the equal weights, environmental/social and the economic policy scenarios, respectively. Lafia, Nasarawa-Eggon, and Obi LGAs are consistent priority areas across the three investigated policy scenarios and are thus, regarded as areas of peculiar interest for further assessment. The study recommends that the "most suitable" locations in Lafia, Nasarawa-Eggon, and Obi LGAs receive first consideration for wind energy farm development in the State. The results from this study could be resourceful to the State and the local governments in performing land management and risk assessment. The variations in the comprehensiveness of input datasets may affect the accuracy of GIS-based WLC results. Hence, it becomes imperative to note that; (1) data on critical habitat used in this study did not include the locations of some critical habitats (zoo, game reserves and parks). Disclosing such areas would expose the habitat to a significant threat. Also, the input data on wind speed was at 50 m height above ground level. Data at higher elevations may result in higher potential areas for wind energy farms in the study area. Furthermore, the specific nature of wind turbines as their: sizes, and heights, were not considered. Any proposal for wind farm projects in the State would require a more specific analysis tailored to the features of the proposed turbines.