Environmental Management Framework for Wind Farm Siting: Methodology and Case Study

Journal of Environmental Management 91 (2010) 2134e2147

Contents lists available at ScienceDirect

Journal of Environmental Management

journal homepage: www.elsevier.com/locate/jenvman

Environmental management framework for wind farm siting: Methodology and case study

Leda-Ioanna Tegou*, Heracles Polatidis, Dias A. Haralambopoulos

Energy Management Laboratory, Department of Environment, University of the Aegean, Lesvos 81100, Greece article info abstract

Article history: This paper develops an integrated framework to evaluate land suitability for wind farm siting that Received 21 September 2009 combines multi-criteria analysis (MCA) with geographical information systems (GIS); an application of Received in revised form the proposed framework for the island of Lesvos, Greece, is further illustrated. A set of environmental, 14 April 2010 economic, social, and technical constraints, based on recent Greek legislation, identiﬁes the potential Accepted 14 May 2010 sites for wind power installation. Furthermore, the area under consideration is evaluated by a variety of Available online 11 June 2010 criteria, such as wind power potential, land cover type, electricity demand, visual impact, land value, and distance from the electricity grid. The pair-wise comparison method in the context of the analytic Keywords: Wind farm siting hierarchy process (AHP) is applied to estimate the criteria weights in order to establish their relative Geographical information systems importance in site evaluation. The overall suitability of the study region for wind farm siting is appraised Multi-criteria analysis through the weighted summation rule. Results showed that only a very small percentage of the total area Analytic hierarchy process of Lesvos could be suitable for wind farm installation, although favourable wind potential exists in many Greece more areas of the island. Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction 2006). The term siting difficulty is defined as any combination of obstacles in wind turbines siting process, including environmental, Wind farm siting often constitutes a process of locating topographic, and geographic constraints; public opposition; local, a sometimes undesirable facility; nevertheless it is an important state, and federal regulatory barriers to permitting, investment, first step in wind power development. The environmental impacts and/or construction. of wind farms include generally effects on visual impact (Bishop In recent years geographical information systems (GIS) have and Miller, 2007), bird collision (Fielding et al., 2006; Yue and become increasingly popular as a tool for the selection of optimal Wang, 2006), noise generation (Cavallaro and Ciraolo, 2005), sites for different types of activities and installations. GIS are not electromagnetic interference (Baban and Parry, 2001), and safety only computer systems designed to produce maps, but also issues (Aydin et al., 2010). Greece has recently established a special powerful tools of geographical analysis. A GIS is a system of hard- regulatory framework regarding the siting of renewable energy ware, software, and procedures to facilitate the acquisition, facilities in general based on proper land use planning and management, manipulation, analysis, modelling, representation, sustainable development (HMEPPPW, 2008). However, apart from and output of spatially referenced data to solve complex planning the constraints provided by national legislation, any site selection and management problems (Carrion et al., 2008). The appropri- and assessment procedure must address the technical, economic, ateness of a GIS for locating RES facilities is portrayed in Baban and social and environmental aspects of the project to determine Parry (2001). whether it is suitable for wind energy development (Polatidis and Applications of GISeRES planning include wind farm siting, Haralambopoulos, 2007). As a result, energy and electricity photovoltaic electrification, biomass evaluation, visual impact industry professionals and policy groups have developed a variety assessment of wind parks, etc. (Amador and Dominguez, 2006; approaches to mitigate siting difficulty of new wind power plants Gadsden et al., 2003; Ma et al., 2005; Masera et al., 2006; (Lejeune and Feltz, 2008; Ramirez-Rosado et al., 2008; Rodman and Miranda, 2006; Ramachandra and Shruthi, 2007). Meentemeyer, 2006; Vajjhala and Fischbeck, 2007; Yue and Wang, One of the most common GIS based strategies that have been designed to facilitate decision making in site selection, land suitability analysis, and resource evaluation is Multi-criteria Analysis * Corresponding author. Tel.: þ30 22510 36216; fax: þ30 22510 36209. (MCA) (Malczewski, 1999). The Analytic Hierarchy Process (AHP) E-mail address: [email protected] (L.-I. Tegou). method, originally developed by Saaty (1980),isaflexible and

0301-4797/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.jenvman.2010.05.010 L.-I. Tegou et al. / Journal of Environmental Management 91 (2010) 2134e2147 2135 easily implemented MCA technique and its use has been largely further used to evaluate the study area in order to bring out the explored in the literature with many examples in locating facilities most economic viable, environmental friendly, technically feasible (Dey and Ramcharan, 2008; Kontos et al., 2005; Tuzkaya et al., and social acceptable sites. The weights of these criteria are esti- 2008; Wang et al., 2009) and in land suitability analysis (Yang mated through the pair-wise comparison method in the context of et al., 2008). A comprehensive literature review of studies that AHP. An overall suitability map for the island of Lesvos is created have applied this method in different fields can be found in Vaidya through the weighted summation of the criteria. and Kumar (2006). The popularity of the AHP method in giving The paper continues as follows: Section 2 describes the area of solutions to a multi-criteria problem is attributed to the fact that it study; section 3 presents the tools and data used for the application takes into consideration both tangible and intangible criteria (Aras of the methodological framework developed below. The results of et al., 2004). There are two specific characteristics that distinguish the application are analyzed and a sensitivity analysis is performed this method from other multi-criteria methods: the construction of in section 4. Finally, concluding remarks are included in section 5. the hierarchy structure and the pair-wise comparisons between different criteria, in order to weight them with respect to the 2. Area of study overall objective (Nekhay et al., 2009). Another benefit of using the AHP method is that it employs a consistency test that can screen Lesvos island is located in the northeast of Greece, in the Aegean out inconsistent judgments (Kablan, 2004). Sea. Its total area is 1636 km2, with 90 436 inhabitants (census of The recently enacted Greek legislation for renewable energy 2001). Agriculture and tourismare themain economicactivities on the facilities siting (HMEPPPW, 2008) was the main motivation for the island. Approximately 41.3% of the island’s surface is olive plantations, present work. The objective is to give a reasonable answer to 34% maquis and garrigue, 17% forest, 4.3% other crops, while the a number of questions such as: remaining land has various uses such as constructions, wetlands, etc. Electricity production in Lesvos island is based on an autono- - How complicated is the process of siting a wind park in an mous grid, which is powered by a conventional oil station owned large autonomous island with sufficient wind potential, variety by the Public Power Corporation (PPC). The power plant is fuelled of land cover, scattered archaeological sites, and areas char- by fuel and diesel oil and is located in the outskirts of Mytilene acterized as “NATURA 2000”. (Fig. 1). Wind potential on the island is significant (Fig. 2) and three - What kind of criteria have to be taken into consideration for wind parks have been installed during the past decades for elec- the evaluation of the study area and which is their relative tricity generation, these projects however have so far managed to importance. exploit only a small fraction of the island’s full wind capacity - How sensitive is the developed framework when criteria are (Koroneos et al., 2004; Ntziachristos et al., 2005). Other RES, i.e. modified and/or weights are changed. geothermal and solar, have also been developed for hot water - How restrictive is the relevant enacted national legislation. production, albeit on a rather limited scale. The island of Lesvos was selected for numerous reasons. First, as As mentioned earlier, the present study develops an integrated most of the Aegean islands, it exhibits an excellent wind potential framework for the selection of optimal sites for wind farm instal- (Kaldellis, 2005). Second, it displays an annual increase of electricity lation, in an electricity autonomous island in Greece, considering consumption at a rate of 6% (primary data of PPC), almost 50% more the restrictions enforced by current national legislation. A combi- than the mainland rate (4.2%) (Hatziargyriou et al., 2006). Moreover, nation of environmental, technical, economic and social criteria is due to geographic, technical, economic, and social reasons, its

Fig. 1. Lesvos island: power plants, road network, main settlements. 2136 L.-I. Tegou et al. / Journal of Environmental Management 91 (2010) 2134e2147

Fig. 2. Wind potential on Lesvos island (Laboratory of Natural Disasters, Department of Geography, University of the Aegean, Greece.). interconnection to the mainland grid still remains under discussion implemented in both raster and vector GIS environments; in our (Kaldellis and Zaﬁrakis, 2007). As a result, it bases its electricity case a raster GIS was selected considering its wider mathematical supply on an autonomous power station, which cannot cover sufﬁ- capabilities (Pereira and Duckstein, 1993). The GIS software used is ciently the peak demands, especially during the summer period. ArcGIS 9.2. GIS data sets of land use, settlements, archaeological sites, 3. Material and methods monasteries, road network, electricity grid, wind potential, and digital elevation models (DEMs) were collected for the Lesvos 3.1. Software and data island from the database of the Laboratory of Natural Disasters, Department of Geography, University of the Aegean. Data about The developed framework can be operationalized using any GIS population, electricity consumption, hotels, and overnight stays system with overlay capabilities. The overlay techniques (Nath were derived from the National Statistical Service of Greece. et al., 2000) allow the ELs to be aggregated to determine the For the implementation of the current study, the Vestas composite map layer (output maps). The method can be V52e850 kW wind turbine (W/T) was selected. It has the following

Stage 1 Stage 2 Stage 3 Stage 4 Stage 5

CONSTRAINT EVALUATION AHP WEIGHTED SUITABILITY FACTORS CRITERIA SUMMATION MAP (Binary Maps) (Standardized Maps) (Weights) (Evaluation map = (CM * EM) w1* E1+…+wn*En))

C1 E1 * w1 WE1 * + C2 E2 * w2 WE2 * . . + ...... + Cm En * wn WEn

CONSTRAINT * EVALUATION SUITABILITY MAP (CM) MAP (EM) MAP

Fig. 3. The proposed methodological framework for wind farm siting. L.-I. Tegou et al. / Journal of Environmental Management 91 (2010) 2134e2147 2137

Table 1 Table 3 Constraints of the case study of the island of Lesvos. Permitted installed W/Ts in each zone on Lesvos island.

Constraints ZONE Radius of zones (m)

C1 Petrified forest Site of interest A B C C2 Wetlands Traditional settlements 1500 3000 6000 C3 NATURA 2000 Significant settlements 1000 3000 d C4 Land of high productivity Other settlements 500 1000 2000 C5 Slope angles < 30% Archaeological sites 500 3000 6000 C6a Settlements Wind turbines/km2 047 C6b Distance from settlementsa Traditional <1500 m Significant <1000 m Other <500 m C7a Archaeological sites C7b Distance from archaeological sites <500 m C8 Distance from monasteries <500 m C9 Distance from road network >10 000 m C10 Airport C11 Wind potential <4m/s a) Sum the values in each column of the pair-wise comparison a Due to the unique cultural features of most Greek settlements, Greek legislation matrix; has characterized them as “traditional”, “significant” and “other” according to their b) Divide each element in the matrix by its column total cultural importance. (normalized comparison matrix); characteristics: height (H) ¼ 44m, rotor diameter (D) ¼ 52m. Larger c) Compute the average of the elements in each row of the size W/Ts would have been even more effective, but the restriction normalized matrix. for installations and the transportation difficulties on the narrow These averages constitute the relative weights (w) of the roads of the island limits the size and height of the W/T that can be compared criteria. transported relatively easily to the appropriate site. The pair-wise comparison procedure can be employed only for According to current legislation (HMEPPPW, 2008) the a relatively small number of elements at each decision hierarchy. minimum distance between two W/Ts must be three times the Therefore, when a large number of alternatives is considered, the rotor’s diameter, so the selected cell size in raster maps is AHP procedure is applied only at the criteria level, and then the 3*D ¼ 156 m, which means that each map cell comprises a potential criteria weights are assigned to the criteria map layers and pro- location for a W/T installation. cessed in the GIS environment. This approach is referred to as spatial-AHP method (Siddiqui et al., 1996). 3.2. The AHP method The AHP also provides mathematical measures to determine inconsistency of judgment. Based on the properties of reciprocal matrices, a consistency ratio (CR) can be calculated. In a reciprocal The AHP falls into the broader category of pair-wise comparison l techniques in which, attributes are ranked against each other to matrix, the largest eigenvalue ( max) is always greater than or equal assess their relative importance. Pohekar and Ramachandran to the number of rows or columns (n). If a pair-wise comparison l ¼ (2004) presented a literature review on multi-criteria decision does not include any inconsistencies, max n. A consistency index making on sustainable energy planning, and observed that AHP is (CI) that measures the inconsistencies of pair-wise comparisons can the most popular technique. be written as (Saaty, 1980): The first step of this method is to structure the decision problem l n in a hierarchy. A typical AHP hierarchy represents the overall CI ¼ max (1) n 1 objective of the decision-making process (goal) at the top level, criteria affecting the decision at the intermediate level, and the Table 4 decision options (alternatives) at the lower level. The second step is Grading values of the criteria associated with land value. the pair-wise comparison of criteria. By comparing a pair of criteria Type/distance Grading value at a time using a verbal scale, decision makers can quantify their Frontage on the road opinions about the criteria importance. The AHP uses a funda- Highway 3 mental nine point’s scale measurement to express individual Secondary road 2 Rural road 1 preferences or judgments (Saaty, 1980), creating a reciprocal ratio matrix, in which the number of rows and columns is defined by the Shoreline distance (m) number of criteria. The sequential steps are as follows: (Carrion 100 5 200 4 et al., 2008; Malczewski, 1999): 500 3 800 2 Table 2 1200 1 Evaluation criteria of the case study of Lesvos island. Land use Evaluation criteria Type of criterion Salt marsh 0 E1 Visual impact e VI Environmental/Social Marshes 0 E1a Visibility from settlements e VFS Environmental/Social Bare land 5 E1b Visibility from archaeological sites e VFAS Environmental/Social Urban land 10 E2 Land value e LV Economic Garrique 6 E3 Slope e S Technical Maquis 6 E4 Land cover e LC Environmental Olive trees 8 E5 Wind potential e WP Technical Other crops 7 E6 Distance from electricity grid e DFEG Economic Chestnut forest 3 E7 Distance from road network e DFRN Economic Pine forest 3 E8 Electricity demand e ED Environmental Other 3 2138 L.-I. Tegou et al. / Journal of Environmental Management 91 (2010) 2134e2147

Table 5 Table 8 Pair-wise comparison matrix and relative importance weights of the criteria Grading values of the wind potential criterion. associated with land value. Wind potential (m/s) Grading value criteria Land Dist. from Dist. from Municipal Weights 0e4 0.00 use road network shoreline District coef. 4e5 0.50 Land use 1.000 0.333 0.200 0.111 0.051 5e6 0.55 Dist. from road 3.000 1.000 0.500 0.200 0.129 6e7 0.60 network 7e8 0.65 Dist. from 5.000 2.000 1.000 0.333 0.233 8e9 0.70 shoreline 9e10 0.75 Municipal 9.000 5.000 3.000 1.000 0.587 10e11 0.80 District c. 11e12 0.85 Consistency ratio ¼ 0.014 < 0.1 12e13 0.90 13e14 0.95 14e23 1.00

Table 6 Grading values of the slope criterion. Table 9 Slope (%) Grading value Grading values of the distance from electricity grid criterion. 0 1.0 0e10 0.9 Distance from electricity grid (m) Grading value 10e15 0.6 0 1.00 15e20 0.2 0e100 0.95 20e25 0.1 100e200 0.90 25e74 0.0 200e300 0.85 300e400 0.80 e and a measure of coherence of the pair-wise comparisons can be 400 500 0.75 500e600 0.70 calculated in the form of consistency ratio (CR): 600e700 0.65 700e800 0.60 CI e CR ¼ (2) 800 900 0.55 RI 900e1000 0.50 1000e1100 0.45 where RI is the average CI of the randomly generated comparisons 1100e1200 0.40 (Saaty, 1980). As a rule of thumb, a CR value of 10% or less is 1200e1300 0.35 considered as acceptable. Otherwise, the decision maker has to 1300e1400 0.30 e revise his judgments (Kablan, 2004). 1400 1500 0.25 1500e1600 0.20 1600e1700 0.15 3.3. Description of the methodology 1700e1800 0.10 1800e1900 0.05 e 3.3.1. Introduction 1900 2000 0.01 >2000 0.00 The siting process is a MCA problem requiring consideration of a comprehensive set of attributes in determining the suitability of a particular area for a deﬁned land use (Gamboa and Munda, 2007; Table 10 Tsoutsos et al., 2007). These attributes involve bounding Grading values of the distance from road network criterion. constraints that comprise of physical, technical, economic, environmental, and cultural issues, and criteria that represent a yard- Distance from road network (m) Grading value stick or means by which a particular alternative can be evaluated as 0 1.00 e more desirable than another. These constraints and evaluation 0 100 0.96 100e200 0.92 criteria determine the selection of potential sites. The constraints 200e300 0.88 are based on the Boolean relation (true/false) and limit the study 300e400 0.84 area to particular sites. The evaluation criteria deﬁne a degree of 400e500 0.80 continuous measure of suitability for all feasible alternatives. 500e600 0.76 600e700 0.72 The siting methodology presented here is a combination of MCA 700e800 0.68 with GIS. The AHP method is used to assign weights of relative 800e900 0.64 900e1000 0.60 1000e1100 0.56 Table 7 1100e1200 0.52 Grading values of the land cover criterion. 1200e1300 0.48 1300e1400 0.44 Type of land cover Grading value 1400e1500 0.40 Bare land 1.0 1500e1600 0.36 Garrique 1.0 1600e1700 0.32 Maquis 0.9 1700e1800 0.28 Olive trees 0.5 1800e1900 0.24 Chestnut forest 0.1 1900e2000 0.20 Pine forest 0.1 2000e2100 0.16 Other crops 0.0 2100e2200 0.12 Marshes 0.0 2200e2300 0.08 Salt marsh 0.0 2300e2400 0.04 Urban land 0.0 2400e2500 0.01 Other 0.0 >2500 0.00 L.-I. Tegou et al. / Journal of Environmental Management 91 (2010) 2134e2147 2139

Table 11 Estimated electricity demand per sector in Lesvos island for year 2025 (based on data from PPC and NSSG).

Sector Domestic Commercial Industrial Agricultural Public and Municipal Street lighting Total % 49.1 28.3 5.8 4.5 9.4 2.9 100 KWh 281 579 846 162 139 655 33 142 141 25 909 383 53 865 190 16 471 767 573 107 982

Sectors Sub-sectors Data DOMESTIC Population

50% TOURISM Accommodations Overnights COMMERCIAL 50% GENERAL Commercial stores COMMERCE ELECTRICITY INDUSTRIAL Industries DEMAND

AGRICULTURAL Crops

PUBLIC & MUNICIPAL Population

STREET LIGHTING Settlements’ area

Fig. 4. Modelling of electricity demand on Lesvos island. importance to each evaluation criterion. An overall suitability index (evaluation map) using the add overlay operation on the for each potential site of the study area is calculated using the weighted standardized ELs. weighted overlay technique (Nath et al., 2000). Particularly, the Stage 5 Calculate the ﬁnal suitability index map by multiplying the presented methodological framework involves the following stages evaluation map with the constraint map. (Fig. 3):

Stage 1 Define the constraints that bound the study area and create 3.3.2. Determination of constraints and computation of the the relative binary constraint layers (CL); calculate the constraint map constraint map through the multiplication of all CLs. The first stage of the methodology addresses the issue of Stage 2 Identify the set of criteria and alternatives, generate the defining the bounding constraints; they could be environmental, relative evaluation layers (ELs) and standardize each EL. In technical, social, economic, cultural, etc., and they depend on our case, the feasible alternatives are the map cells. standing legislation and on the characteristics of the study area. In Stage 3 Estimate the criteria weights using the pair-wise compar- this study, they were selected in accordance with recent Greek ison method in the context of the AHP and check the legislation for RES siting (HMEPPPW, 2008). In particular, eleven consistency of the process. (11) constraint layers (CLs) were produced according to the Stage 4 Create the weighted standardized ELs by multiplying the constraints presented in Table 1. A binary GIS grid is created for standardized ELs with the corresponding weights; each constraint, with cells falling within a constrained area generate the overall suitability index for each cell assigned “0” and the rest of them assigned “1”.

Table 12 Computation of the electricity demand map, Lesvos island.

Sector Sub-sector Data Calculations Domestic Population Percentage of population (PP) of each settlement to the total population Multiplication of PP with the total domestic demand (kWh/settlement) Division with settlement’s area (kWh/m2) and multiplication with 24 336a (kWh/cell) Commercial Tourism Overnights Percentage of overnights (PO) of each settlement to the total overnights Multiplication of PO with the 50% of the total commercial demand (in kWh/settlement) Division with settlement’s area (kWh/m2) and multiplication with 24 336 (kWh/cell) General Commerce Commercial stores Percentage of commercial stores (PCS) of each settlement to the total commercial stores Multiplication of PCS with the 50% of the total commercial demand (in kWh/settlement) Division with settlement’s area (kWh/m2) and multiplication with 24 336 (kWh/cell) Industrial Industries Percentage of industries (PI) of each settlement to the total industries Multiplication of PI with the total industrial demand (in kWh/settlement) Division with settlement’s area (kWh/m2) and multiplication with 24 336 (kWh/cell) Agricultural Crops Percentage of each crop’s area (PCA) to the total crops’ area Multiplication of PCA with the total agricultural demand (in kWh/crop) Division with crop’s area (kWh/m2) and multiplication with 24 336 (kWh/cell) Public and Municipal Population Multiplication of PP by the total public and municipal demand (in kWh/settlement) Division with settlement’s area (kWh/m2) and multiplication by 24 336 (kWh/cell) Street lighting Settlements’ area Percentage of each settlement’s area (PSA) to the total settlements’ area Multiplication of PSA with the total street lighting demand Division with settlement’s area (kWh/m2) and multiplication with 24 336 (kWh/cell)

a Since cell size is 156m the area of each cell is equal to 156 m 156 m ¼ 24 336 m2. 2140 L.-I. Tegou et al. / Journal of Environmental Management 91 (2010) 2134e2147

Fig. 5. The standardized evaluation layers. (a) visual impact, (b) land value, (c) slope, (d) land cover, (e) wind potential, (f) distance from electricity grid, (g) distance from road network, (h) electricity demand.

For the constraints C1, C2, C3, C4, C5, C6a, C7a, C10 and C11, the specific SoI and for an observatory height of 1.5 m. The “visible” corresponding binary CLs are produced by the procedure of cells are scored “0” and the rest of them “1”. reclassification embedded in the ArcMap. For constraint C9 a 10 000 m buffer zone is created and the According to national legislation constraints C6b, C7b and C8 areas out of this zone are valued as “0”. In Lesvos island, this apply only to visible wind turbines (HMEPPPW, 2008). Therefore, criterion does not make any difference since the 10 000 m buffer our first step is to calculate the visibility maps for every Site of zone is enough to cover the whole island, so there are no areas Interest e SoI (settlements, archaeological sites and monasteries) valued “0”. through the following procedure: Initially, a buffer zone for each SoI By multiplying all constraint layers (CL), the final Constraint is created with width according to data presented in Table 1; Map is calculated. Only the cells with value “1” in each input layer subsequently, each map cell is evaluated in terms of visibility from will have non-zero value in the Constraint Map, meaning these cells a set of given SoI by the use of viewshed function of ArcMap meet all constraints and are eligible for further consideration. (OFFSETA ¼ 1.5, OFFSETB ¼ 44, RADIUS2 ¼ radius of buffer zone) In any case, it should be ensured that the level of noise to the and the relevant visibility maps are generated. The production of inhabitants of the above residential areas is less than 45 db, which the visibility maps is based on the Digital Elevation Model (DEM) of could be checked by field inspection. Appropriate siting and Lesvos island. These maps illustrate the cells in which an installed planning conditions are essential to minimise this impact, but as (into the buffer zone) W/T with H ¼ 44 m is visible or not from the noted, due to the wide variation in individual tolerance to noise, L.-I. Tegou et al. / Journal of Environmental Management 91 (2010) 2134e2147 2141

Wind turbine site selection

VI LV S LC WP DFRN DFEG ED

site 1 site 2 site 3 site 4 site 5 ……………….. site n

VI: visual impact site 1,2,3,…,n: map cells LV: land value S: slope LC: land cover WP: wind potential DFRN: distance from road network DFEG: distance from electricity grid ED: electricity demand

Fig. 6. Hierarchy structure of the wind farm siting problem. there is no completely satisfactory way to predict unfavourable giving the following values: 0: into settlement; 2: into Zone A; 4: into reactions. Zone B; 8: into Zone C e) Summation of reclassified map with visibility map (calculated 3.3.3. Evaluation criteria description and modelling in step (a)) / reclassification, giving the following values: 0: In the next phase, evaluation criteria that score the potential sites restricted area; 5: into Buffer 3000 and visible; 7: into Buffer were defined and represented as map layers (Evaluation Layers e 6000 and visible; 9: beyond Buffer 6000 and visible; 10: non- ELs) in a GIS database. Eight criteria were set up based on literature visible. review, experts’ judgment, and personal experience; they are presented in Table 2. The procedure followed for the establishment of The “archaeological sites’ visibility map” is calculated in the the grading values for the criteria used is also adopted by other same lines. The two visibility maps have the same weight factor researchers (e.g. Baban and Parry, 2001; Bennui et al., 2007; Rodman (50% each). Thus, visual impact map (E1) ¼ (0.5*settlements’ visi- and Meentemeyer, 2006). A description of these criteria follows. bility map) þ (0.5*archaeological sites’ visibility map)

3.3.3.1. Visual impact (E1). In order to limit the visual impact of 3.3.3.2. Land value (E2). The evaluation of land in terms of wind parks, Greek legislation confines the number of visible wind economic value is accomplished through these criteria: (i) A turbines in an area that is close to a SoI depending on the distance Municipal District coefficient, as defined by national legislation from this site. In detail, the area next to a SoI is separated into three (HMEF, 2008)(Table A1, Appendix A); (ii) Frontage on the road zones, A, B, C, with different radius each one for every type of SoI. network; (iii) Distance from shoreline; (iv) Land use. The grading Non-visible wind turbines are excluded from this limitation. Table 3 values of the above criteria (ii), (iii), and (iv) are shown in Table 4. shows the number of permitted installed wind turbines in each The pair-wise comparison method is used to determine the zone on Lesvos island, for the specific type of wind turbine. For the weights of the four criteria. Table 5 illustrates the pair-wise computation of E1, two visibility maps have to be created based on comparison matrix. the DEM of Lesvos island; (i) the settlements’ visibility map (E1a) After the standardization of above four criteria, the land value and (ii) the archaeological sites’ visibility map (E2b). map is produced as follows: The procedure for E1a calculation consists of the following Land Value Map (E2) ¼ (0.051*Land use) þ (0.129*Dist. from steps: road network) þ (0.233*Dist. from shoreline) þ (0.587*Municipal District coefficient) a) Computation of visibility from each settlement to a zone with radius 10 000 m through viewshed function / reclassification giving the values: 0: visible; 1: non-visible Table 13 b) Creation of a buffer zone for each type of settlement with Pair-wise comparison matrix and relative importance weights of the evaluation criteria. radius as Zone A of Table 3 (restricted zone) / union all of Criteria LV S VI ED DFEG DFRN LC WP Weights them / union with shoreline / conversion to raster (cell LV 1.000 0.500 0.250 0.200 0.200 0.200 0.167 0.111 0.025 size ¼ 156 m) / reclassification giving the following values: 0: S 2.000 1.000 0.500 0.333 0.250 0.250 0.200 0.167 0.039 into restricted area; 1: out of buffer zone VI 4.000 2.000 1.000 0.500 0.333 0.333 0.333 0.250 0.065 ED 5.000 3.000 2.000 1.000 0.500 0.500 0.333 0.333 0.095 c) Summation of reclassified map with visibility map (calculated DFEG 5.000 4.000 3.000 2.000 1.000 1.000 0.500 0.500 0.145 in step (a)) DFRN 5.000 4.000 3.000 2.000 1.000 1.000 0.500 0.500 0.145 d) Creation of a multiple buffer zone for each type of settlement, LC 6.000 5.000 3.000 3.000 2.000 2.000 1.000 0.500 0.210 according to Table 3 / union all of them / union with WP 9.000 6.000 4.000 3.000 2.000 2.000 2.000 1.000 0.276 ¼ < shoreline / conversion to raster (cell size ¼ 156m) / reclassification, Consistency ratio 0.024 0.1 2142 L.-I. Tegou et al. / Journal of Environmental Management 91 (2010) 2134e2147

Fig. 7. Map of suitability index for the island of Lesvos.

3.3.3.3. Slope (E3). Slope is a technical criterion since very steep 3.3.3.5. Wind potential (E5). The initial wind potential map is slopes of land are not suitable for wind farm installation. In many reclassified according to the power curve of the specific W/T studies slopes greater than 10e20% are not candidates for wind (Vestas, 2009) resulting to the grading values of the wind potential turbine installation (Baban and Parry, 2001; Hatziargyriou et al., criterion (Table 8). The grading values range from 0 to 1, where 2007). Nevertheless, in the Greek island of Crete a wind park has 0 corresponds to zero W/T output and 1 to maximum output. been installed on a land with a slope of 30% (PPC, personal commu- nication). Therefore, we select to exclude the land slopes greater than 3.3.3.6. Distance from electricity grid (E6). Multiple buffer zones 30%. The grading values (0e1) of the slope are shown in Table 6. per 100 m are created in order to evaluate the whole study area; the closer to the grid, the greater the value (Table 9). Areas farther than 3.3.3.4. Land cover (E4). The type of land cover influences the 2000 m from the electricity grid are considered as not economically evaluation of potential sites. For example, a land area covered with viable, assigned the value of “0”. garrique or maquis is preferable than a forest area. Table 7 presents the grading values of land cover types. Note that the intan- 3.3.3.7. Distance from road network (E7). Similarly, multiple buffer gible criteria are initially scored in the common measurement zones per 100 m are created; the closer to the road network, the scale of grading values 0e1, so as they have not to be greater the value (Table 10). The maximum distance that is standardized. considered as feasible is 2500 m. L.-I. Tegou et al. / Journal of Environmental Management 91 (2010) 2134e2147 2143 a ii. The electricity demand of the agricultural sector emerges in 11.0% 1.4% the map cells that correspond to cultivated areas. iii. Street lighting exists only along the boundaries of the settlements. 16.8% As a result, only the cells corresponding to settlements and crops have non-zero values for electricity demand. These areas are, however, restricted from wind parks installation; in order, there- 56.8% fore, to produce a useful map for the estimation of the suitability 10.0% index, buffer zones are created around settlements and crops. Thus, 4.0% for each settlement a multiple buffer zone with widths 3000 m and 6000 m is created, with values equal to x and x/2 respectively, ¼ 0 0-0.6 0.6-0.7 0.7-0.8 0.8-0.9 0.9-1 where x yearly electricity demand in kWh/cell for each sector of the settlement. As for the crops, a buffer zone of 2000 m is created that receives the value of the yearly electricity demand of the cor- b responding crop. 41% The next step consists of the union of all generated maps with shoreline following by the conversion to raster. Using the raster fi 25% calculator the raster maps are summed to produce the nal map, which depicts the spatial electricity demand for the year 2025.

3.3.4. Standardization of evaluation criteria In order to allow direct comparability between ELs, they are standardized and all map cells have a grading value between 0 and 19% 10% 1(Fig. 5). 1% 4% The standardization method used here is the ‘maximum score procedure’ (Malczewski, 1999). These standardized scores range 0-4 m/s 4-6 m/s 6-8m/s 8-10 m/s 10-14 m/s 14-23 m/s from 0 to 1. The advantage of this method is that the relative order of magnitude of the standardized scores remains equal.

Fig. 8. Percentage distribution of (a) Suitability index and (b) Wind Potential in Lesvos island. 3.3.5. Estimation of weights Not all evaluation criteria are equally important; the AHP 3.3.3.8. Electricity demand (E8). The electricity demand of each method is used to assign weights to the criteria. Initially, the three- map cell of the island is modelled according to the electricity level hierarchy structure of the problem is designed (Fig. 6) consumption and the population growth of the last 15 and 20 years In our case, the potential sites include all the map cells; that is respectively. Data on population were taken from National Statis- considered a large number of candidate sites. Therefore, the AHP tical Service of Greece (NSSG) and on electricity consumption from procedure will be terminated at the criteria level. The pair-wise NSSG and PPC (NSSG, 2008; Table 11). The modelling of the elec- matrix for the estimation of the weights of the evaluation criteria is tricity demand is shown in Fig. 4, while Table 12 illustrates the created as shown in Table 13. computation of the electricity demand map. The rationale behind the particular criteria weighting used here The process is based on the following assumptions: is shortly presented in the following:

i. The electricity demand of the commercial sector is equally - The wind potential is considered to be the most important divided into two sub-sectors; tourism and other commercial criterion since it determines the output of the wind turbine; activities. - The land cover criterion comes next since the type of the land might hinder the wind farm installation and/or licensing; 25% - The distance from electricity grid and road network are restricted area/ thought to be less significant as they affect mainly the final cost total area 20% of installation and the grid losses; - Fifth in the order of priorities is the electricity demand criterion which is strongly related to the losses of electricity grid; 15% - The criterion of visual impact ties up with the public acceptance and the number of permitted installed W/Ts. Although, % public acceptance is a very critical issue in wind parks place- 10% ment (Dimitropoulos and Kontoleon, 2009; Papadopoulos et al., 2008; Tsoutsos et al., 2009; Wolsink, 2007), the wind 5% parks already placed on Lesvos island have not met strong public opposition. Besides, according to a recent Euro- barometer survey, the Greek public appears to favour strongly 0% PF LC N S Set AS M Air WP the development of wind power (EC, 2006). Thus this criterion is deemed less important than the previous ones; Constraints - The slope of the land surface is mainly a technical criterion; PF: Petrified Forest; LC: land cover; N: Natura 2000; S: slope; Set: settlements; technical difficulties, however, could be overcome at the expense AS: archaeological sites; M: monasteries; Air: airport; WP: wind potential of the project economics since very steep slopes will lead to higher Fig. 9. Percentage of restricted area to total area, for each individual constraint. installation costs; so, it ranks seventh in the order of priorities; 2144 L.-I. Tegou et al. / Journal of Environmental Management 91 (2010) 2134e2147

Fig. 10. Maps of suitability index for the four cases of the sensitivity analysis. (a) Case 1; (b) Case 2; (c) Case 3; (d) Case 4.

- Finally, the land value is placed last in order of importance Xn ¼ v since it has only to do with the cost of the land that will host OSIi wj ij (3) the wind park. j ¼ 1

3.3.6. Generation of the suitability map where OSIi is the overall suitability index for cell i, wj is the relative After the criteria weights have been estimated, a weighted sum importance weight of criterion j, vij is the score of cell i under aggregation function is used in order to compute an Overall Suit- criterion j, and n is the total number of criteria. In GIS, this tech- ability Index (OSI) for each cell of the study area. More speciﬁcally, nique results in an overall Evaluation Map: each EL is multiplied by the respective weight. The weighted ELs are ¼ð : * Þþð : * Þþð : * Þ summed in order to provide a total performance score for each cell. Evaluation Map 0 065 E1 0 025 E2 0 039 E3 The map cells are ranked from the highest to the lowest score. The þð0:210*E4Þþð0:276*E5Þþð0:145*E6Þ applied mathematic formulation (Yoon and Hwang, 1995): þð0:145*E7Þþð0:095*E8Þ (4)

4. Application of the methodology 40000 Original 4.1. Results 35000 Case 1 Case 2 The Suitability Map is derived from the multiplication of the 30000 Case 3 Constraint Map with the Evaluation Map (Fig. 7). This operation is Case 4 25000 equivalent to removing the restricted areas out of the Evaluation Map. The most appropriate areas for wind farm siting are those 20000 shown in dark blue, with suitability index 0.9e1. Sites with 0.8e0.9 15000 score are also suitable, having, however, more disadvantages, i.e.

Number of map cells higher costs, technical difﬁculties, intense but within acceptable 10000 levels visual impact, etc. It is noticeable that over half of the entire

5000 study area (56.8%) is restricted from wind farm installation, while only a considerably small percentage (1.4%) of the island achieved 0 the optimal suitability index (Fig. 8a), even though wind potential is 0 0-0.6 0.6-0.7 0.7-0.8 0.8-0.9 0.9-1 favourable in more areas (Fig. 8b). Nevertheless, the best-ranked Level of Suitability index sites coincide with those with the highest wind potential (Figs. 2 ﬁ Fig. 11. Number of map cells for each level of suitability index for the four cases of the and 7). This could be attributed to the signi cant weight of the sensitivity analysis and the original case. “wind potential” criterion. L.-I. Tegou et al. / Journal of Environmental Management 91 (2010) 2134e2147 2145

Fig. 9 below shows the percentage of the island’s area that is elaborated them in order to produce the overall suitability map. A restricted from wind projects development due to the existence of sensitivity analysis on the weights of the evaluation criteria was each individual constraint. From this perspective the most impor- also performed showing that each criterion is influential in the tant constraint is the existence of archaeological sites that excludes evaluation of the suitability of a site; so the selection of criteria is an 20% of the island’s area, followed by the limited wind potential essential stage in a siting methodological framework. (excludes 19%), the settlements (16%), and the Petrified Forest The results identified the optimal locations for wind projects, (12%). excluding over 50% of the whole study region, where the best The last step of the siting process is to evaluate local charac- scored areas occupy only the 1.4% of the island’s surface. The teristics of the high-scored areas after field inspection, in order to proposed methodology, however, allows the analyst to consider verify their overall appropriateness. even less suitable sites, by reducing the acceptable threshold for the suitability index. This would result in designating more areas as 4.2. Sensitivity analysis appropriate for wind farm development. In any case, it would be after field inspection that potential investors would engage them- In most multi-criteria exercises a ‘what if’ sensitivity analysis is selves in developing a project. Thus, future work could include the recommended as a means of checking the stability of the results individual assessment of the emerging areas in conjunction with against the subjectivity of the expert judgments (Meszaros and field inspection in order to make the final selection of sites for wind Rapcsak, 1996). The most common method is to modify the parks installation. weightings obtained from the experts. The assumption of equal The proposed modelling and operationalization of the evalua- weightings is also used for this purpose (Nekhay et al., 2009). tion criteria, relevant for such an exercise, in addition to the way In this study, the sensitivity analysis performed considers the these criteria were used in conjunction with the legislative effect of criteria weights changes upon the overall suitability index. boundary constraints under a unified multi-criteria decision aiding To that aim, the following four cases were examined: and spatial analysis framework is main innovatory dimension of this work. Case 1 All criteria have the same weights. Energy planners may use this framework to determine locations Case 2 The weight of the criterion “visual impact” is zero (0). This is for feasible wind farm installation in the Greek autonomous to explore the actual wind potential, without taking into islands. The presented methodology, however, could be applied to account the particular social dimension of wind energy. other kinds of project siting since, due to its generic nature, it can Case 3 All environmental criteria have weights equal to zero (0). incorporate a variety of criteria and constraints. Case 4 All economic criteria have weights equal to zero (0). Acknowledgements In cases (2e4) the relative pair-wise comparisons of the non- zero criteria remain the same. This research is co-financed by National and Community Funds The results of the four cases are illustrated in Fig. 10. As it can (20% from the Greek Ministry of Development e General Secre- be seen, the presented framework is sensitive to the criteria tariat of Research and Technology and 80% from E.U. e European weights. This was expected since, as it is mentioned earlier, the Social Fund). The authors would also like to thank two anonymous evaluation criteria are selected with respect to the specific char- reviewers for their valuable comments. acteristics of the study area. The change of the final suitability map that is derived from the changing of criteria weights, implies that Appendix A each selected criterion is influential in the evaluation of the study region. It is noticeable that, although the resulting maps for the four Table A1 fi fi cases of the sensitivity analysis demonstrate considerable modifi- Municipal district coef cients de ned by national legislation. cations in the suitability index, Fig. 11 shows that the number of the Municipal District Municipality Municipal district coefficient most suitable sites (suitability index ¼ 0.9e1) for wind farm siting Afalonas Mytilene 35 remains relatively low in all cases. A noteworthy variation is Agia Marina Mytilene 30 observed in case 3, where no economic criteria are taken into Agiassos Agiasos 10 Agra Kalloni 10 consideration; the majority of the potential sites belong to the Akrasi Plomari 5 suitability index class “0e0.6”. This could be attributed to the Alifanta Mytilene 35 elimination of the criteria of “distance from electricity grid” and Ampeliko Plomari 5 “distance from road network”. These two criteria favour many areas Anemotia Kalloni 5 that coincide (Fig. 5f and g) and their simultaneous exclusion Antissa Eressos-Antissa 15 fi Argenos Mithimna 5 results to severe overall disquali cation of these areas. Asomatos Evergetoulas 15 Chidira Eressos-Antissa 5 5. Conclusions Eressos Eressos-Antissa 10 Filia Kalloni 5 Ipios Evergetoulas 20 The current work reports on the development of a methodo- Ipsolometopo Petra 10 logical framework for wind turbine site selection based on the Kalloni Kalloni 20 combination of the multi-criteria analysis and GIS techniques; Kapi Mantamados 15 the developed framework is applied in Lesvos island, Greece. The Kapi Agia Paraskevi 10 objective of the study was to find suitable sites for wind farm Kato Tritos Evergetoulas 10 Keramia Evergetoulas 15 installation taking into account a number of emerging economic, Kleio Mantamados 15 social, environmental, and technical criteria. The pair-wise Komi Loutropolis Thermis 10 comparison method in the context of the AHP was utilized to assign Lampou Myloi Evergetoulas 15 the relative weights to the evaluation criteria, while GIS established Lepetimnos Mithimna 5 the spatial dimension of constraints and evaluation criteria and (continued on next page) 2146 L.-I. Tegou et al. / Journal of Environmental Management 91 (2010) 2134e2147

Table A1 (continued ) Systems. Available from: http://ec.europa.eu/research/energy/pdf/energy_tech_ eurobarometer_en.pdf (last access date 27.08.09.). fi Municipal District Municipality Municipal district coef cient Fielding, A.H., Whitfield, D.P., McLeod, D.R.A., 2006. Spatial association as an indi- Lisvori Polichnitos 10 cator of the potential for future interactions between wind energy develop- Loutra Mytilene 35 ments and golden eagles Aquila chrysaetos in Scotland. Biological Conservation e Loutropolis Thermis Loutropolis Thermis 30 131, 359 369. Gadsden, S., Rylatt, M., Lomas, K., 2003. Putting solar energy on the urban map: Mantamados Mantamados 15 a new GIS-based approach for dwellings. Solar Energy 74, 397e407. Megalochori Plomari 5 Gamboa, G., Munda, G., 2007. The problem of wind farm location: a social multi- Mesagros Gera 10 criteria evaluation framework. Energy Policy 35, 1564e1583. Mesotopos Eressos-Antissa 10 Hatziargyriou, N.D., Tsikalakis, A., Androutsos, A., 2006. Status of distributed Michou Evergetoulas 10 generation in the Greek islands. In: IEEE, Proceedings of the Power Engineering Mistegna Loutropolis Thermis 20 Society General Meeting, Montreal, Canada. Mithimna Mithimna 40 Hatziargyriou, N.D., Tsikalakis, A., Kilias, V., 2007. The effect of island intercon- Moria Mytilene 30 nections on the increase of Wind Power penetration in the Greek System. In: Mytilene Mytilene 50 IEEE, Proceedings of the Power Engineering Society General Meeting, Tampa, Nees Kidonies Loutropolis Thermis 20 Florida, USA. Neochori Plomari 5 Hellenic Ministry of Economy and Finance (HMEF), Ministerial Circular, 2008. “P L ” Paleochori Plomari 5 O .1068/3.4.2008 . Adjustment values for the objective of determining the Paleokipos Gera 20 taxable value of any cause transferred land areas outside the city or town that has no specific housing conditions. Pamfila Mytilene 30 Hellenic Ministry of Environment, Physical Planning and Public Works (HMEPPPW), Panagiouda Mytilene 30 2008. Special Regulatory Framework for Land-planning and Sustainable Papados Gera 20 Development in Renewable Energy Source, Athens, Greece. Pelopi Mantamados 15 Kablan, M.M., 2004. Decision support for energy conservation promotion: an Perama Gera 20 analytic hierarchy process approach. Energy Policy 32, 1151e1158. Petra Petra 25 Kaldellis, J.K., 2005. Social attitude towards wind energy applications in Greece. Pigi Loutropolis Thermis 10 Energy Policy 33, 595e602. Pirgi Thermis Loutropolis Thermis 30 Kaldellis, J.K., Zafirakis, D., 2007. Present situation and future prospects of electricity Plagia Plomari 5 generation in Aegean Archipelago islands. Energy Policy 35, 4623e4639. Plakados Gera 10 Kontos, Th., Komilis, D., Halvadakis, K., 2005. Siting MSW landfills with a spatial Plomari Plomari 5 multiple criteria analysis methodology. Waste Management 25, 818e832. Polichnitos Polichnitos 20 Koroneos, C., Michailidis, M., Moussiopoulos, N., 2004. Mutli-objective optimization Pterounta Eressos-Antissa 5 in energy systems: the case study of Lesvos island, Greece. Renewable and e Sigri Eressos-Antissa 5 Sustainable Energy Reviews 8, 91 100. Laboratory of Natural Disasters, Department of Geography, University of the Sikaminea Mithimna 10 Aegean, Greece. Sikounta Evergetoulas 10 Lejeune, P., Feltz, C., 2008. Development of a decision support system of setting up Skalochori Kalloni 5 a wind energy policy across the Walloon Region (southern Belgium). Renewable Skopelos Gera 20 Energy 33, 2416e2422. Stavros Polichnitos 5 Ma, J., Scott, N., DeGloria, S., Lembo, A., 2005. Siting analysis of farm-based Stipsi Petra 10 centralized anaerobic digester systems for distributed generation using GIS. Taxiarches Mytilene 20 Biomass and Bioenergy 28, 591e600. Trigonas Plomari 3 Malczewski, J., 1999. GIS and Multicriteria Decision Analysis. John Wiley & Sons, Vasilika Polichnitos 15 New York. Vatoussa Eressos-Antissa 5 Masera, O., Ghilardi, A., Drigo, R., Trossero, M.-A., 2006. WISDOM: AGIS-based Vrissa Polichnitos 20 supply demand mapping tool for woodfuel management. Biomass and Bio- energy 30, 618e637. Meszaros, Cs, Rapcsak, T., 1996. On sensitivity analysis for a class of decision systems. Decision Support Systems 16, 231e240. Miranda, V., 2006. Wind power, distributed generation: new challenges, new References solutions. Turkish Journal of Electrical Engineering 14. Nath, S.S., Bolte, J.P., Ross, L.G., Aguilar-Manjarrez, J., 2000. Applications of Amador, J., Dominguez, J., 2006. Spatial analysis methodology applied to rural geographical information systems (GIS) for spatial decision support in aqua- electrification. Renewable Energy 31, 1505e1520. culture. Aquacultural Engineering 23, 233e278. Aras, H., Erdogmus, S., Koc, E., 2004. Multi-criteria selectionforawind observationstation National Statistical Service of Greece (NSSG), 2008. Available from: http://www. location using analytic hierarchy process. Renewable Energy 29, 1383e1392. statistics.gr/portal/page/portal/ESYE (last access date 27.08.09.). Aydin, N.Y., Kentel, E., Duzgum, S., 2010. GIS-based environmental assessment of Nekhay, O., Arriaza, M., Guzman-Alvarez, J.R., 2009. Spatial Analysis of the suit- wind energy systems for spatial planning: a case study from Western Turkey. ability of olive plantations for wildlife habitat restoration. Computers and Renewable and Sustainable Energy Reviews 14, 364e373. Electronics in Agriculture 65, 49e64. Baban, S., Parry, T., 2001. Developing and applying a GIS-assisted approach to Ntziachristos, L., Kouridis, C., Samaras, Z., Pattas, K., 2005. A wind-power fuel-cell locating wind farms in the UK. Renewable Energy 24, 59e71. hybrid system study on the non-interconnected Aegean islands grid. Renewable Bennui, A., Rattanamanee, P., Puetpaiboon, U., Phukpattaranont, P., Chetpattana- Energy 30, 1471e1487. nondh, K., 2007. Site selection for large wind turbine using GIS. In: Proceedings Papadopoulos, A.M., Glinou, G.L., Papachristos, D.A., 2008. Developments in the of the PSU-UNS International Conference on Engineering and Environment e utilisation of wind energy in Greece. Renewable Energy 33, 105e110. ICEE, Phuket. Pereira, J.M.C., Duckstein, L., 1993. A multiple criteria decision making approach to Bishop, I.D., Miller, D.R., 2007. Visual assessment of off-shore wind turbines: the GIS based land suitability evaluation. International Journal of Geographical influence of distance, contrast, movement and social variables. Renewable Information Systems 7, 407e424. Energy 32, 814e831. Pohekar, S., Ramachandran, M., 2004. Application of multi-criteria decision making Carrion, J.A., Estella, A.E., Dols, F.A., Toro, M.Z., Rodriguez, M., Ridao, A.R., 2008. to sustainable energy planning e a review. Renewable and Sustainable Energy Environmental decision-support systems for evaluating the carrying capacity of Reviews 8, 365e381. land areas: optimal site selection for grid-connected photovoltaic power plants. Polatidis, H., Haralambopoulos, D., 2007. Renewable energy systems: a societal and Renewable and Sustainable Energy Reviews 12, 2358e2380. technological platform. Renewable Energy 32, 329e341. Cavallaro, F., Ciraolo, L., 2005. A multicriteria approach to evaluate wind energy Ramachandra, T.V., Shruthi, B.V., 2007. Spatial mapping of renewable energy plants on an Italian island. Energy Policy 33, 235e244. potential. Renewable and Sustainable Energy Reviews 11, 1460e1480. Dey, P.K., Ramcharan, E.K., 2008. Analytic hierarchy process helps select site for Ramirez-Rosado, J., Garcia-Garrido, E., Fernandez-Jimenez, L.A., Zorzano- limestone quarry expansion in Barbados. Journal of Environmental Manage- Santamaria, P., Monteiro, C., Miranda, V., 2008. Promotion of new wind farms ment 88, 1384e1395. based on decision support system. Renewable Energy 33, 558e566. Dimitropoulos, A., Kontoleon, A., 2009. Assessing the determinants of local Rodman, L., Meentemeyer, R., 2006. A geographic analysis of wind turbine place- acceptability of wind-farm investment: a choice experiment in the Greek ment in Northern California. Energy Policy 34, 2137e2149. Aegean islands. Energy Policy 37, 1842e1854. Saaty, T., 1980. The Analytic Hierarchy Process. McGraw-Hill, New York. European Commission (EC), 2006. Energy Technologies: Knowledge ePerceptions e Siddiqui, M., Everet, J., Vieux, B., 1996. Landfill siting using geographical information Measures. EUR22396. Directorate-General for Research Sustainable Energy systems: a demonstration. Journal of Environmental Engineering 122, 515e523. L.-I. Tegou et al. / Journal of Environmental Management 91 (2010) 2134e2147 2147

Tsoutsos, T., Maria, E., Mathioudakis, V., 2007. Sustainable siting procedure of Wang, G., Qin, L., Li, G., Chen, L., 2009. Landﬁll site selection using spatial infor- small hydroelectric plants: the Greek experience. Energy Policy 35, mation technologies and AHP: a case study in Beijing, China. Journal of Envi- 2946e2959. ronmental Management 90, 2414e2421. Tsoutsos, T., Tsouchlaraki, A., Tsiropoulos, M., Serpetsidakis, M., 2009. Visual impact Wolsink, M., 2007. Wind power implementation: the nature of public attitudes: evaluation of a wind park in a Greek island. Applied Energy 86, 546e553. equity and fairness instead of ‘backyard motives’. Renewable and Sustainable Tuzkaya, G., Onut, S., Tuzkaya, U.R., Gulsun, B., 2008. An analytic process approach Energy Reviews 11, 1188e1207. for locating undesirable facilities: an example from Istanbul, Turkey. Journal of Yang, F., Zeng, G., Du, C., Tang, L., Zhou, J., Li, Z., 2008. Spatial analyzing system for Environmental Management 88, 970e983. urban land-use management based on GIS and multi-criteria assessment Vaidya, O.S., Kumar, S., 2006. Analytic hierarchy process: an overview of applica- modelling. Progress in Natural Science 18, 1279e1284. tions. European Journal of Operational Research 169, 1e29. Yoon, K., Hwang, C.L., 1995. Multiple Attribute Decision Making: An Introduction. Vajjhala, S., Fischbeck, P., 2007. Quantifying siting difﬁculty: a case study of US Sage Publication Inc., London, UK. transmission line siting. Energy Policy 35, 650e671. Yue, C.-D., Wang, S.S., 2006. GIS-based evaluation of multifarious local renewable Vestas. V52e850 kW Brochure, 2009. Available from: http://www.vestas.com/en/ energy sources: a case study of the Chigu area of southwestern Taiwan. Energy wind-power-solutions/wind-turbines/kw.aspx (last access date 27.08.09.). Policy 34, 730e742.