- Email: [email protected]
A review on the Spanish Method of visual impact assessment of wind farms: SPM2
Renewable and Sustainable Energy Reviews 49 (2015) 756–767
Contents lists available at ScienceDirect
Renewable and Sustainable Energy Reviews journal homepage: www.elsevier.com/locate/rser
A review on the Spanish Method of visual impact assessment of wind farms: SPM2 Cristina Manchado n, Valentin Gomez-Jauregui, César Otero Civil Engineering School of Santander, University of Cantabria, 39005 Cantabria, Spain
art ic l e i nf o
a b s t r a c t
Article history: Received 5 April 2014 Received in revised form 15 August 2014 Accepted 24 April 2015
This work offers a review of the so-called Spanish Method for the visual impact assessment of wind farms. The five coefficients originally proposed in the method have been analysed and discussed from several approaches: validity, efficiency, limitations and need of actualisation, among others. As a result, we establish a set of new proposals that update or modify the definition or calculation of these coefficients, but always trying to retain their original meaning. The work is complemented by a short case study in which we compare the values of the coefficients of the original Spanish Method with those arisen from our new proposal. The difference is often relevant, both in the numerical value of the coefficients and in the improvement of their ability to describe the visual effect. Finally, the new formulation of the Spanish Method opens a possibility for the public participation in several moments of the process. & 2015 Elsevier Ltd. All rights reserved.
Keywords: VIA Visual impact assessment LVIA Visual impact indicators
Contents 1. 2.
3.
4.
n
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A review of the Spanish Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Visibility coefficient of the wind farm from village (a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Visibility coefficient of village from wind farm (b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Visibility coefficient of the wind farm taken as a cuboid (c) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4. Distance coefficient between the wind farm and the village (d). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5. Population coefficient of the village (e) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6. Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion and methodological proposal for the Spanish Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Coefficient a: subdivision of the villages into areas of uniform visual impact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Coefficient a: an alternative expression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Coefficient b: visibility of the buildings of the village from the wind farm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Coefficient c: procedure for univocally obtaining the enveloping cuboid of the park . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5. Coefficient c: a discussion on the values of visibility. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6. Coefficient c: a new definition for the coefficient v. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7. Coefficient c: review and proposal for coefficient n. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8. Coefficient d: some considerations about the effect of distance on visibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9. Coefficient e: review and proposal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.10. Partial assessment factors PA1 and PA2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Visual effects of different hypotheses of design. A case study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Coefficient a . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Corresponding author. E-mail addresses: [email protected] (C. Manchado), [email protected] (V. Gomez-Jauregui), [email protected] (C. Otero).
http://dx.doi.org/10.1016/j.rser.2015.04.067 1364-0321/& 2015 Elsevier Ltd. All rights reserved.
757 757 757 757 757 759 759 759 759 759 759 759 761 761 761 762 762 763 764 764 765
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
4.2. Coefficient c: coefficients v and n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3. Coefficient d . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. The method as a software tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1. Introduction In 2004 the journal Renewable and Sustainable Energy Reviews published an article explaining “a methodology to predict, before its construction, the visual impact that a wind farm can have” [1]. The paper was groundbreaking and somehow surprising: having elapsed only 18 days between its reception and its approval, having cited just seven references (all of them national regulations related to the general problem of the deployment of wind farms in Spain), the method did not offer a case study, nor a method for its validation. In spite of this, it established a very right framework to assess several visual and perceptual aspects that were then a subject of analysis for the scientific community, the stakeholders and some other groups involved in the planning, development and use of wind farms. Moreover, the work was greatly concise, original, intuitive and logical. It proposed a rather easily replicable methodology, as well as a way to quantify some aspects considered critical in the process of visual impact assessment of wind farms. Since then, the Spanish Method has become a reference very often cited in the reviews of the contributions studying the visual impact caused by wind farms. More precisely, several works report to have made use of this method to assess impacts: (i) by contrasting their results with other ones obtained by means of public acceptance's methodologies [2,3], (ii) by comparing the visual effect of several wind farms, and the variation of their Spanish Method coefficients under a particular mitigation hypothesis [4] and (iii) by studying the time of visual exposure of a wind farm by the observers travelling along a particular motorway [5]. The Spanish Method can be integrally programmed [4]. That makes it possible to characterise the effect of one – or several – wind farms over the whole visual inventory, either to a local, regional or trans-regional level. In general, the computation time of this task is high but acceptable. The Spanish Method results more reliable to compare different scenarios than to obtain an absolute visual assessment, but that happens with most of the methods based on visual indicators [6–9]. As a scientific work, the Spanish Method needs corrections and updates. In this article we show how, once deeply reviewed, it helps to get a full meaning of the numerical characterisation of the visual effect caused by a wind farm. It is also possible to compare different mitigation proposals and gives a way to address a participatory process, what always helps to give transparency to the whole project. The work is arranged as follows: the Spanish Method is reviewed in Section 2. In Section 3, several improvements to the method are discussed. Section 4 shows a brief case study, useful to compare the values of the coefficients before and after the proposal of improvement carried out in Section 3. Section 5 describes how and where this improvement of the Spanish Method can be used currently, into a computational free use environment. Finally, in Section 6, the paper offers some conclusions as well as some reflections about the limitations of the present work, the strategy of improvement and about the research itself. 2. A review of the Spanish Method The Spanish Method was established on the basis of five coefficients, which are later integrated into a single value. It
757
765 765 766 766 766
enables access to a table that gives back the definitive impact assessment, expressed as a number into the interval [0–1], whose extreme values represent respectively a null visual impact or an unacceptable visual impact. Next, the aforementioned five coefficients are presented. 2.1. Visibility coefficient of the wind farm from village (a) The method establishes a first coefficient for each one of the villages to be analysed. If the visual impact is presumed to vary along a village, this one is divided into homogeneous subareas. Each subarea must have a uniform sight of the wind farm (this is, it must see the same number of turbines). The number of turbines seen from each area is expressed as a much per one of the total of turbines of the park. Therefore, if the coefficient a has, for example, a value of 0.6, it means that a 60% of the turbines of the park are seen from the village. Thus, such a value of 0.6 can be understood as an average value or as a probability. As usual, it is accepted that a partially seen tower is computed as a visible one. However, the next considerations cannot remain unnoticed: - No procedure is reported to obtain a subdivision of the village in homogeneous areas. So, this aspect of the method is really asystematic. - The areas having whole visibility or null visibility over the wind farm are not studied at all. However, these extreme values should be always reported in some manner. Not doing it could be misunderstood as a way to deliberately bias or hide the impact in these significant parts of the village. These issues are analysed in Sections 3.1 and 3.2. 2.2. Visibility coefficient of village from wind farm (b) The method establishes a coefficient b defined as the ratio between the number of buildings that are visible from the park and the total number of buildings in the village. This definition is particularly imprecise with respect to: – The meaning of the property “buildings seen from the park” provided that the turbines can see very different sets of buildings of a village. – The meaning of “visibility” provided that the upper floors of a building can be seen at the same time than its lower ones remain hidden by other buildings placed forward. – The way to capture the geometry of buildings and to assign them the attribute of a certain height. We will deal with this matter in Section 3.3. 2.3. Visibility coefficient of the wind farm taken as a cuboid (c) The method sets up a coefficient c that assesses the combined effect of the observer's position with respect to the park (coefficient ν) and the number of turbines it has (coefficient n). Coefficient ν forces to divide the surrounding area of the wind
758
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
farm into three different types of areas called front, diagonal and longitudinal. To determine them, it is necessary to consider an enveloping cuboid of the wind farm, as it is suggested in Fig. 1. The coefficient ν receives a value of 1.0 if the village is seeing the park from the front, a value of 0.5 if it is doing it from the diagonal zone and a value of 0.2 if it is doing it from the longitudinal area. No other intermediary positions of the village are considered. The value of the coefficient n, regarding the number of turbines of the park, is shown in Table 1. The sought coefficient c is the product ν n. But we should pay attention to these remarks: – There are an infinite amount of possible cuboids enveloping a wind farm (see for example Fig. 2) and each solution gives rise to different sets of front, diagonal and longitudinal areas. This circumstance is not admissible and should be corrected so that such areas are invariant for a given park. – The coefficient ν has always the only same three values: 1.0, 0.5 and 0.2 for respectively front, diagonal and longitudinal areas. However, this proposal is not consistent (see Fig. 3). Indeed, if the enveloping rectangle is a square, it results undefined which area is front and which one is longitudinal. In fact, in this particular situation the four edges should give rise to the same type of area, front or longitudinal. Furthermore, in such a case, maybe the diagonal area (not the front one) deserves the maximum score for coefficient ν.
– It is not unusual that a village is placed just on the edge of two different areas. For such a case, the value of the coefficient ν has not been defined.
Regarding to the table that defines the coefficient n, the ranks should be, at least, updated. As an example, the method gives the neutral effect (coefficient n¼ 1.00) to parks containing between 11 and 20 turbines, which does not fit to the population sensibility nowadays nor to the estimations of other similar tables. In Sections 3.5–3.7 we will discuss all these considerations.
Fig. 3. If the cuboid becomes a square, front and longitudinal areas can be swappable.
Fig. 1. Wind farm with different sides of viewing.
Table 1 Coefficient n for calculating coefficient c. Number of windmills
n factor
1–3 4–10 11–20 21–30 430
0.50 0.90 1.00 1.05 1.10
Fig. 2. For a given wind farm layout, there are infinite possible enveloping cuboids.
Table 2 Coefficient d, depending on the distance. Distance
Coefficient d
xo 500 m 500 m o xo 6000 m x46000 m
1.00 1.05–0.0002x 0.10
Fig. 4. Linear variation of coefficient d.
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
2.4. Distance coefficient between the wind farm and the village (d) The method defines a coefficient d that considers the incidence of the distance between the wind farm and the village. The same as with the coefficient n, these values (see Table 2) need to be updated, as well as the thresholds of visibility themselves. Furthermore, the method posits the variation of this coefficient according to a linear law (see Fig. 4). The effect of the distance on the visibility is mainly depending on the size of the object and on the changing conditions of their contrast against the background. The laws describing these two phenomena are not linear; not even both are polynomial. In Section 3.9, we will discuss all these considerations. In Section 3.8 this coefficient is discussed more in depth. 2.5. Population coefficient of the village (e) A last coefficient, e, reflects the effect of population. This coefficient is nil for unpopulated villages, it has a value of 0.20 when population varies between 0 and 5 inhabitants and gets its maximum (e¼ 1.00) for 300 or more inhabitants. It can be argued that: – As this estimation is often subject of controversy, it is recommended that these values can result from a participatory process. – It could be advisable to consider a value of the coefficient e bigger than 1.00 for populations of some particularly fragile villages (having special historic, cultural or social values or, as well, having a highly damaged or altered landscape already). In Section 3.9 this coefficient is discussed more in depth. 2.6. Evaluation Once these five coefficients have been calculated, the method sets up the overall assessments factors PA1 and PA2, expressed as follows: PA1 ¼ a b c d PA2 ¼ a b c d e Finally, the method offers a table to express the assessment in qualitative terms (high, medium, low, etc.) and gives an expression to average the total visual effect when all the villages in the area of influence of the wind farm are considered. These last details are not particularly relevant for our work and can be read in the original paper.
3. Discussion and methodological proposal for the Spanish Method Initially, we programmed the Spanish Method according to their original specifications [4]. In fact, most of the comments shown in Section 2 arise as a result of the different proofs of concept carried out to test such implementation. The visibility calculations for the coefficients a and b make necessary the computation of the ZVI (Zones of Visual Influence) associated to each turbine of the wind farm; as it has been detailed in [4,10], these ZVI are recorded into a numerical structure that keeps in each pixel of the DEM (Digital Elevation Model) the visibility conditions in each and every one of the turbines. This structure is called CHESSBOARD. Some computational properties of CHESSBOARD give the visibility status of any turbine i with respect to a pixel (x, y). The index i is applied to every one of the turbines
759
and the coordinates x and y cover all the DEM considered in the study. 3.1. Coefficient a: subdivision of the villages into areas of uniform visual impact CHESSBOARD is particularly suitable to solve the intended subdivision of the villages into areas of uniform visual impact simply considering that each one of its pixels is one of such areas. Indeed, we then only need to select the collection of pixels belonging to the village and, next, to obtain for each one of these pixels its accumulated value of visibility. The computational property vis(i) has the value 0 if the turbine is not visible and 1 if it is seen from the pixel. The index i goes over each and every turbine of the park (k in total). Each pixel p(x, y) of the village has then an associated value Sp: SP ¼
k X
CHESSBOARDðx; yÞ:visðiÞ
ð1Þ
i¼1
This makes it possible to know how many pixels have a value Sp¼0 (these pixels have no visibility over the park), a value Sp ¼1, Sp¼2, etc. until Sp ¼k (these pixels have a full visibility of the park). Next, the variable SUMA0 stores the amount of pixels that have a value Sp¼0, SUMA1 keeps those that have a value Sp ¼1, etc., up to SUMAk. Being Np the total number of pixels that belong to the village and k the number of turbines of the park, the coefficient a is unambiguously defined as: Pk i ¼ 0 ði SUMAi Þ=k a¼ ð2Þ NP 3.2. Coefficient a: an alternative expression An indicator obtained simply by computing the visibility of the pixels of a ZVI has been reported by other researchers [8,11–13]. Thus, we suggest that the coefficient a can be alternatively expressed by means of a raw visibility value such as Sp (Eq. (1)). This leads to an expression for a as the one expressed in Eq. (3). Using the value of Sp for each pixel in the village provides two added advantages: (i) the coefficient a can be graphically represented in a thematic map and (ii) the areas free of visual effect or the ones that have visibility to the whole wind farm can be directly depicted. a¼
Number of pixels of the village having visibility to the park Number of pixels of the village ð3Þ
3.3. Coefficient b: visibility of the buildings of the village from the wind farm The three comments stated in Section 2.2 lead to a conclusion: the coefficient b cannot be calculated, not even approximately, by means of algorithms based on ZVI techniques (and in general, in computational techniques based on raster data). Fig. 6 helps to understand that techniques such as ray-tracing, hidden surface removal or similar [14] are the suitable ones to solve adequately this problem. Nevertheless, sometimes we have adapted the idea sketched by this coefficient b in order to express the visual effect specifically concentrated on the zones of greater density of population into the village. Once a threshold of density of population by pixel is defined, the set S of those pixels in the village having a density of population equal to or bigger than the threshold is considered. From this set, the subset Sv of those pixels that are seen from the
760
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
Fig. 5. Villages with different degrees of visibility. White pixels represent subareas free of visual effect.
Fig. 6. Group of buildings with different degrees of visibility.
Fig. 7. Unambiguous algorithm for obtaining the enveloping cuboid of the wind farm WF. (a) LE ¼WT1–WT6, Pr ¼ WT4, Ps ¼WT3. (b) LE¼ WT4–WT5, Pr ¼WT2.
wind farm gives the value to the coefficient b. The condition of visibility is given by the ZVI itself. b¼
Sv : Number of pixels of the set S that have visibility from the park S : pixels having a density of population upper than a threshold
ð4Þ
In addition, fortunately Eq. (4) enables the Spanish Method to be applied to other elements of the visual inventory different from villages, such as roads, protected areas, cultural sites, panoramic itineraries, etc. In these other cases, the coefficient b can be applied to those singular subareas that can exist inside the area
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
object of the study (for example, the areas of bigger amount of visitors inside a protected area).
3.4. Coefficient c: procedure for univocally obtaining the enveloping cuboid of the park Let us call WF the wind farm and WT1, WT2,… WTk the set of its turbines (see Fig. 7). This is a procedure to obtain univocally its enveloping cuboid: – Obtain the set of distances dij ¼distance(WTi–WTj) between couples of turbines of WF.
761
– Then the longest edge of the cuboid is defined by the segment LE¼WTi–WTj whose associated distance is maximum: LE ¼MAX (dij). – Next, obtain all the distances between the k turbines WTi and the segment k. – Two possible situations can arise: There are turbines in both sides of the segment LE (Fig. 5a). In such a case, the enveloping cuboid is defined by the parallel lines to LE containing the points Pr and Ps. These points have the largest distance to LE from left and right. There are no turbines in both sides of the segment LE (Fig. 5b). In such a case, the cuboid is defined by the same segment LE and its parallel containing the most distant turbine Pr.
3.5. Coefficient c: a discussion on the values of visibility Front and longitudinal areas cover angles of 151 from the edges of the cuboid (see Fig. 1). This leads to the diagonal areas being angles of 601 applied on the vertices of the enveloping cuboid. The coefficient ν is valued to 0.2, 0.5 and 1.0, respectively for the longitudinal, diagonal or frontal areas (even though these values are not any way justified). However, in general the shape of the cuboid can vary from a perfect square to a really oblong rectangle. In particular (as it was exposed in Section 2.3) if the cuboid becomes a square, this procedure gives inconsistent results. Our first try to solve this problem took into account the ratio r ¼SE/LE between the shortest and the longest edges of the rectangle, so that the coefficient ν was redefined this way: – – –
ν ¼ 1.0 for villages placed on the front area. ν ¼ r ¼SE/LE for villages placed on the longitudinal area. ν ¼ (1 þr)/2 for villages placed on the diagonal area.
This designation results again unsatisfactory; in many cases, the maximum visual effect is perceived from the diagonal. That could lead to try a second redefinition of the coefficient ν; D being the measure of the diagonal of the cuboid: – – –
ν ¼ LE/D for villages placed on the front area. ν ¼ SE/D for villages placed on the longitudinal area. ν ¼ 1.0 for villages placed on the diagonal area.
However, this expression remains based on two statements – why choose only three types of areas and why choose an angle of 151 to define the border lines among them – that are not justified. Such a division in three types of areas reduces the nature of the problem, simplifying it excessively. Maybe a division into only three areas (front, diagonal and longitudinal) is not the best way to represent the variation of the visibility coefficient ν, and some other approaches should be considered. 3.6. Coefficient c: a new definition for the coefficient v
Fig. 8. Coefficient d modelled by Hermite's Splines. (a) Boundary conditions for the first example. (b) Solution for the first example (Eq. (8)). (c) Boundary conditions for the second example. (d) Solution for the second example (Eq. (10)).
For modelling the effect of the visual exposure of the park, several works draw on the magnitude of the visibility angle [8] or the solid angle [9]. Both are continuous variables. It is therefore advisable to abandon the resource of the enveloping cuboid and to represent the coefficient directly by means of the visibility angle spanned by the wind farm from a camera point. An algorithm for its calculation is described in [15]. It can be argued that this new coefficient ν does not have a normalised range of variation between 0 and 1. Nevertheless, this can be easily achieved, for example, assigning to the visual angle of π radians the value ν ¼ 1. It can be noticed as well that this new
762
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
coefficient ν is affected by distance (the previous one was not). This is true and we consider it is better to effectively measure the visual effect according to the position of the observer. Besides, we can always consider a homogenised expression such as ν d (where d is the distance from the observer to the wind farm) for this coefficient. The final coefficient ν associated to a village (or any other element of the visual inventory) is the average of the ones obtained for each one of their pixels. 3.7. Coefficient c: review and proposal for coefficient n A lot of scientific literature [16–18], good practice guides [19–25], case studies [26], regional and national regulations offer proposals to this respect, usually expressed as tables of assessment. It is really unnecessary that the Spanish Method opts just for one of these tables in particular; this choice can come from a participatory agreement. Moreover, if the Spanish Method is completely implemented as a computational application, it means that it is very easy to run different hypotheses of assessment according to different tables of values for this coefficient n. We should think about this other issue as well: if a wind farm has 8 turbines (corresponding to n ¼0.9) but from the village V only 3 (corresponding to n ¼ 0.5) are seen, should we give n ¼0.9 or n ¼0.5 to V? We show some comparisons to this respect in Section 4.2. 3.8. Coefficient d: some considerations about the effect of distance on visibility
model for the coefficient d can be proposed by techniques of Computer Aided Geometric Design (CAGD). This branch of mathematics has been used – and it is still used – in many different processes where the geometry of objects, or the law governing a certain phenomenon, has to be modelled depending on the boundary conditions, because they are the only source of available data. The CAGD offers many standard procedures to interpolate or approximate curves to many different combinations of boundary conditions. One of these combinations is the problem that we face here. In the example of Fig. 8a, we only know the threshold of maximum visibility (the coefficient d is set to 1.0 for the first 8 km) and the threshold of minimum visibility (it is considered that the coefficient d remains constant with a value of 0.1 for a distance of 22 km onwards, until the studio area is over). Then, the objective is to define a polynomial curve that passes through two specific points with the given tangents. This is a standard problem in CAGD [34] and it is formulated as a Hermite's spline, which is expressed, if we choose a third-degree polynomial, as follows: P ðuÞ ¼
3 X
P i Bi ðuÞ
ð5Þ
i¼0
By definition, the parameter u varies on the unit interval [0, 1]. The functions Bi(u) are, in this case, the Hermite's third-degree polynomials whose expressions [34] are: B0 ðuÞ ¼ 2u3 3u2 þ 1 B1 ðuÞ ¼ 2u3 þ 3u2 B2 ðuÞ ¼ u3 2u2 þ u
The tables originally proposed by the Spanish Method have become obsolete since the size of turbines is currently much bigger. With respect to the thresholds of distance, the literature is wide [6,27–32], thus it is again advisable to decide these values in a participatory process. Such events always need a labour of scientific and technical dissemination previously, in which it is possible to analyse as well the intermediate thresholds of distance in which the visual effects are considered of high, medium or low intensity. As a simple example, for turbines 140 m high (tower and blades), we usually define an area of high visual effect at 5 miles around the machine, an area of medium effect at a distance between 5 and 10 miles and an area of low intensity at a distance from 10 to 16 miles. These values can be a reasonable extrapolation of the ranks proposed in Refs. [21,24,27]. It is also convenient to think about the reason why this law (see Table 2 and Fig. 4) is linear and consider if this hypothesis is the best. Let us focus on how the visual effect has been defined at a distance of 500 m. At this point, the visual intrusion suddenly and sharply starts to decrease, even though these types of natural phenomena are better described by laws having a smooth and completely continuous gradient. Visibility and visual effect of an object vary with the distance according to two main causes: the variation of its size and the variation of its contrast with the background. The first effect follows a well-known quadratic law; the second effect is known as scattering, which follows an exponential law [27,33]. Obviously, this latter law accepts to be approximated, but the approximant polynomial does not need to be of the order 2; it can be cubic, quartic, etc. as well. Some works, mainly based on the perceptions and assessments of photographs made by expert [7] or non-expert observers [27] give this law a quadratic expression. It seems clear that these second order polynomials are more suitable than the linear ones. Nevertheless this model must be expressed by a law consistent with the boundary conditions, which are the visibility thresholds. Since a definition of the visibility thresholds exists, the mathematical
B3 ðuÞ ¼ u3 u2
ð6Þ
The boundary conditions, defined by the threshold values, are as follows (see Fig. 8a): P 0 ¼ ð8; 1Þ; which means that the coefficient d is 1:0 at 8 km: P 1 ¼ ð22; 0:1Þ; which means that the coefficient d is 0:1 at 22 km: P 2 ¼ ð1; 0Þ; which means that the tangent in point P 0 is horizontal: P 3 ¼ ð1; 0Þ; which means that tangent in point P 1 is horizontal:
ð7Þ
Also by definition, the Hermite's spline is set to be P0 at u ¼0 and P1 at u ¼1, being the value of the tangent P2 at u ¼ 0 and P3 at u¼ 1. As a result, the desired curve can be expressed with these parametric expressions: X ðuÞ ¼ 8 B0 ðuÞ þ 22 B1 ðuÞ þ 1 B2 ðuÞ þ 1 B3 ðuÞ ¼ 26 u3 þ 39 u2 þ u þ 8 Y ðuÞ ¼ 1 B0 ðuÞ þ 0:1 B1 ðuÞþ 0 B2 ðuÞþ 0 B3 ðuÞ ¼ 1:8 u3 2:7 u2 þ 1
ð8Þ
The sought curve (Fig. 8b) fits the proposed conditions for the visibility thresholds. The above described formulation can be generalised to any type visibility threshold. Thus, as a simple example, if we would like to model the visual effect (coefficient d) for other atmospheric conditions that would give a constant threshold of maximum visibility for the first four kilometres, and a threshold of minimum visibility linearly attenuated from a value of coefficient d equal to 0.1 at a distance of 15 km until it wears off (coefficient d¼0), at a distance of 20 km, the model described maintains the expressions [Eqs. (5) and (6)]. It also maintains the domain of the parameter u on the unit interval [0, 1], but the conditions of [Eq. (7)] become those exposed in [Eq. (9)], below (see Fig. 8c): P 0 ¼ ð4; 1Þ; which means that coefficient d is 1 at 4 km: P 1 ¼ ð15; 0:1Þ; which means that coefficient d is 0:1 at 15 km: P 2 ¼ ð1; 0Þ; which means that the tangent at point P 0 is horizontal P 3 ¼ ð1; 0:02Þ; which means that the tangent at ð9Þ point P 1 is 0:1=ð20 15Þ
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
763
Fig. 9. Layout of the wind farm, villages and background map. The crosses (X) represent the turbines. The circles represent thresholds at 8, 22 and 30 km. The legend at left describes the visibility map for the 8 turbines.
Table 3 Coefficient a calculated according to Eq. (1) (left column) and Eq. (2) (right column). Name
Coefficient a according to Eq. (1)
Coefficient a according to Eq. (2)
ESCOBEDO FRESNEDO HAZAS DE CESTO LIERGANES RIAÑO RIBA SARÓN SOLORZANO SOMO
0.217 0.000 0.898 0.369 0.997 0.000 0.058 0.778 0.802
0.305 0.000 1.000 0.533 1.000 0.000 0.120 1.000 0.823
These conditions would give these new parametric expressions: X ðuÞ ¼ 4 B0 ðuÞ þ 15 B1 ðuÞ þ1 B2 ðuÞ þ 1 B3 ðuÞ ¼ 20 u3 þ 30 u2 þ u þ 4:0 Y ðuÞ ¼ 1 B0 ðuÞ þ 0:1 B1 ðuÞ þ 0 B2 ðuÞ 0:02 B3 ðuÞ ¼ 1:78 u3 –2:68 u2 þ1:0
ð10Þ
The result is shown in Fig. 8d. In a strict mathematical sense, the curve may have a degree different to third-degree. That is a decision of the expert using this method. If, for example, a Hermite's interpolation with fourth-degree polynomials is used, sets of four terms as expressed in [Eq. (7) or (9)] are no longer sufficient to mathematically define the model. The researcher or designer gains a new degree of freedom (DoF) that, properly chosen, will enable to debug the model scientifically or technically. It is obvious now that this approach opens the possibility of testing models for the coefficient d of quartic Hermite splines (degree four) wherein the fifth degree of freedom can be the coefficient for the atmospheric attenuation of the scattering. That is the research line in which part of our current work is focused on this moment.
Fig. 10. Villages considered in the case study and front, diagonal and longitudinal areas.
3.9. Coefficient e: review and proposal The definition of this indicator admits the same reflections expressed in Section 3.7 with respect to the coefficient n of the indicator c. As there are not few scientific studies [6,12,13,35] and national or regional regulations [19–25] about it, we consider that the most advisable for the designer is to elaborate the best dissemination sessions or material and try to achieve a definitive table for the indicator e by means of participation and agreement.
764
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
3.10. Partial assessment factors PA1 and PA2 These global indicators have not been studied in this review. Our opinion is that the Spanish Method is rich and valuable when coefficients a–e are individually analysed, but we do not trust too much on the meaning of the integration of these coefficients into a single number. Among many other reasons, such a general coefficient could compensate possible extreme values of their individual coefficients, what not always helps to make a good analysis. Moreover, the
Spanish Method can be a good tool to understand and to express the visual effects, but the final assessment must be conducted by mean of participation and agreement in some manner.
4. Visual effects of different hypotheses of design. A case study The next case study obtains the original coefficients of the Spanish Method and compares them to those arising from the new Table 5 Comparison of values of coefficient d for original method and new proposals.
Table 4 Comparison of values for coefficient ν. (N.V. means “no visibility”) and coefficient n. Name
Original coeff. ν
Modified coeff. ν (rad)
Original coeff. n
Modified coeff. n
ESCOBEDO FRESNEDO HAZAS DE CESTO LIERGANES RIAÑO RIBA SARÓN SOLORZANO SOMO
0.50 0.20 1.00
0.06 N.V. 0.52
0.90 0.90 0.90
0.90 0.00 0.90
1.00 0.20 0.20 1.00 0.50 0.50
0.12 0.24 N.V. 0.05 0.55 0.15
0.90 0.90 0.90 0.90 0.90 0.90
0.90 0.90 0.00 0.50 0.90 0.90
Village
Distance (km)
Original coefficient d
Updated coefficient d (linear)
Updated coefficient d (cubic)
ESCOBEDO FRESNEDO HAZAS DE CESTO LIERGANES RIAÑO RIBA SARÓN SOLORZANO SOMO
23036.71 25446.46 2570.99
0.100 0.100 0.538
0.100 0.100 1.000
0.100 0.100 1.000
11606.57 4938.12 13067.61 20687.98 2753.63 11642.27
0.100 0.055 0.100 0.100 0.494 0.100
0.770 1.000 0.670 0.180 1.000 0.770
0.852 1.000 0.732 0.122 1.000 0.849
Fig. 11. Horizontal angle from the village of Sarón.
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
proposal, to provide an order of the magnitude of the change produced. The experiment is focused on coefficients a, c and d. We consider a project of wind farm in Cantabria (Spain), made up by eight turbines disposed as a cluster (non linear), as it is shown in Fig. 9. The villages considered in the area are representative of different conditions of visibility, distances, relative situations with respect to the park, etc. 4.1. Coefficient a In Table 3, the central column gives the values according to Eq. (1) (see Section 3.1) and the right column shows the ones obtained according to Eq. (2) (see Section 3.2). The first expression gives more attenuated results. The difference obtained can have proportions up to 44% (Liérganes) or 40.5% (Escobedo): the choice of one option or another is something relevant and must be justified. Eq. (1) had this development, for example, in the village of Solórzano: Pk i ¼ 0 ði SUMAi Þ=WM a¼ NP ð1 4 þ 27 5 þ 176 6 þ 110 7 þ 8 8Þ=8 ¼ 0:778 ¼ 322 However, Eq. (2) gives a coefficient a ¼1.0 because any pixel in Solórzano is free of visibility towards the wind farm.
765
4.2. Coefficient c: coefficients v and n Fig. 10 shows the disposition of the villages with respect to the front, diagonal and longitudinal areas, according to the specification of the original method. The original values of coefficient ν are presented in the second column of Table 4. The third column shows the value of ν expressed as visibility angle, such as it has been proposed in Section 3.7. The fourth column shows the coefficient n according to its original definition and the fifth column shows the new value, when not all but only the viewed turbines are considered for obtaining n in Table 1. The values of the coefficient ν differ, sometimes intensely, from those obtained in the new proposal. For example, let us consider the case of the villages of Hazas de Cesto and Sarón (see Table 4 and Fig. 11). Even though both are situated in a front area of the park (i.e., ν ¼1.00 in the original specification), they have a very different visibility angle (0.52 rad in Hazas de Cesto and 0.05 in Sarón). This difference is due to: (i) the angle of visibility decreases with the distance and (ii) the angle of visibility considers strictly the number of turbines actually visible from each village. However, the original method assumes that the whole wind farm is unconditionally seen from anywhere in the village, which is not always true at all. Indeed, a deeper inspection for Sarón shows that at most 4 of the 8 turbines are seen, which considerably reduces the visual effect experienced and makes it clear that the original definition of the method resulted to be too conservative. The same happens with coefficient n. It makes no sense to assign for Sarón a coefficient n ¼0.9, associated to 8 turbines (the whole park), when it is impossible to see from any pixel of the village more than four. Altogether, considering ν and n, the original Spanish Method tended to overestimate the coefficient c.
4.3. Coefficient d
Fig. 12. Law of variation of coefficient d: a comparison between the linear model (blue), and the cubic model (red). Differences for several villages are depicted. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Table 5 shows a comparison among the original value of the coefficient d, a new value of this coefficient once the thresholds of distance have been updated to 8 and 22 km and a third value, considering a cubic variation of the visual effect (see Fig. 12). As expected, updating the law involves deep changes in the values of this coefficient. When this law changes from linear to cubic, the variations are lower but not irrelevant. In villages such as
Fig. 13. A short summary of this review of visual impact indicators.
766
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
Liérganes, Riba or Somo the updated coefficient reaches increments up to 11%.
5. The method as a software tool The Spanish Method has been implemented into a tool called MOYSES [4,5,10] that is currently available in http://193.144.189. 246/moyses. A user can create input layers and run the programme assisted by a comprehensive tutorial. MOYSES computes more visibility coefficients than those studied in this article: apart from the Spanish Method, it can give indicators of prominence, contrast, area, length, duration of the intrusion and some other more features. Fractality [7,36], rythm or uniformity [7] and movement of blades [6,37] are indicators not yet obtained but currently in progress.
6. Conclusion We have reviewed, discussed and reformulated different issues on the Spanish Method for the visual assessment of wind farms. Next, we summarise the most relevant conclusions: – In general, the 5 coefficients used in the Spanish Method are applicable to the whole visual inventory, not only to villages. – Coefficient a was not univocally defined: it has been done in this contribution. – Coefficient a can express an average value for the visibility effect or a raw value as well. Both values give complementary information and may differ considerably from each other. – Coefficient b can be very complicated, expensive to compute and not always reliable enough. Nonetheless, a singular case, when this coefficient is calculated considering the height of buildings equal to cero, is full of meaning, easy to obtain and very advisable. – Coefficient ν of the coefficient c was not univocally defined: it has been done in this contribution. – The coefficient ν of the coefficient c does not successfully express the effect of visual incidence by means of a discrete variable and can be substituted by the visibility angle, frequently considered in similar studies. – The coefficient n of the coefficient c needs to be updated. Even though there are several references to carry out such an update, we suggest taking this issue as the basis for a participatory decision. – A wind farm can be seen from a village only partially. In such a case the coefficient n of the coefficient c should not be computed considering the total number of turbines of the park but only those that are really seen. – Coefficient d needs to be updated; several references can be used to do it even though we suggest taking this issue as the basis for a participatory decision. Moreover, the law of variation for this coefficient according to the distance can be not linear. Quadratic or cubic laws represent the thresholds of visibility better. – Coefficient e needs to be updated as well. Again, this can be addressed by means of a participatory process. In Fig. 13 we have integrated these proposals into an only user form. Each one of these values or options makes up a different setting chosen by the designer for each particular hypothesis of design. To state that the application MOYSES (see Section 5) uses the Spanish Method is not completely exact, since in fact it currently implements the modifications explained in this paper. We refer
this derived method as SPM2. This name (coming from Spanish Method 2) is an acknowledgement to the original methodology; it tries however to remove any explicit reference of nationality (despite of these authors are not less Spanish than the Method itself). References [1] Hurtado JP, Fernández J, Parrondo JL, Blanco E. Spanish method of visual impact evaluation in wind farms. Renew Sustain Energy Rev 2004;8:483–91. http://dx.doi.org/10.1016/j.rser.2003.12.009. [2] Theocharis T, Tsouchlaraki A, Tsiropoulos M, Serpetsidakis M. Visual impact evaluation of a wind park in a Greek island. Appl Energy 2009;86:546–53. http://dx.doi.org/10.1016/j.apenergy.2008.08.013. [3] Quianna W, MwirigiM’Ikiugu M, Isami K. Visual impact evaluation of wind farms: a case study of Choshi City, Japan. Civ Environ Res 2013;3(7):97–106. [4] Manchado C, Otero C, Gómez-Jáuregui V, Arias R, Bruschi V, Cendrero A. Visibility analysis and visibility software for the optimisation of wind farm design. Renew Energy 2013;60:388–401. http://dx.doi.org/10.1016/j.renene.2013.05.026. [5] Manchado C, Gomez-Jauregui V, Otero C. Moyses v4.0 modeller and simulator for visual impact assessment. In: Proceedings of the international congress in energy and environment engineering and management. Lisbon, Portugal; 2013. 978-989-98406-4-5. [6] Bishop ID, Miller D. Visual assessment of off-shore wind turbines: the influence of distance, contrast, movement and social variables. Renew Energy 2007;32(5):814–31. http://dx.doi.org/10.1016/j.renene.2006.03.009. [7] Torres Sibille AC, Cloquell-Ballester VA, Cloquell-Ballester VA, Darton R. Development and validation of a multicriteria indicator for the assessment of objective aesthetic impact of wind farms. Renew Sustain Energy Rev 2009;13(1):40–66. http://dx.doi.org/10.1016/j.rser.2007.05.002. [8] Rodrigues M, Montañés C, Fueyo R. A method for the assessment of the visual impact caused by the large-scale deployment of renewable-energy facilities. Environ Impact Assess 2010;30:240–6. http://dx.doi.org/10.1016/j.eiar.2009.10.004. [9] Molina-Ruiz J, Martínez-Sánchez MJ, Pérez-Sirvent C, Tudela-Serrano ML, García ML. Developing and applying a GIS-assisted approach to evaluate visual impact in wind farms. Renew Energy 2011;36(3):1125–32. http://dx. doi.org/10.1016/j.renene.2010.08.041. [10] Otero C, Cendrero A, Bruschi V, Togores R, Manchado C, Arias R. Moyses v3.0 (Modeller and Simulator for Visual Impact assessment). In: Proceedings of the 2007 international workshop on computing in civil engineering. Pittsburgh, Pennsylvania. ISBN: 978-0-7844-0937-4; 2007. p. 486–93. [11] Mollër B. Changing wind power landscapes: regional assessment of visual impact on land use and population in Northern Jutland, Denmark. Appl Energy 2006;83:477–94. http://dx.doi.org/10.1016/j.apenergy.2005.04.004. [12] Mollër B. Spatial analyses of emerging and fading wind energy landscapes in Denmark. Land Use Policy 2010;27:233–41. http://dx.doi.org/10.1016/j. landusepol.2009.06.001. [13] Wood G. Is what you see what you get? post-development auditing of methods used for predicting the zone of visual influence in EIA Environ Impact Assess 2000;20:537–56. http://dx.doi.org/10.1016/S0195-9255(00) 00055-X. [14] Hearn D, Baker P. Computer graphics. Upper Saddle River, New Jersey: Prentice Hall PTR; 1994. [15] Otero C, Manchado C, Arias R, Iglesias A, Gálvez A. Method for mitigating visual impact. Patent WO2012049340A1.PCT/ES2011/000295. Published 19/04/2012. [16] Ladenburg J, Möller B. Attitude and acceptance of offshore wind farms – The influence of travel time and wind farm attributes. Renew Sustain Energy Rev 2011;18:4223–35. http://dx.doi.org/10.1016/j.rser.2011.07.130. [17] Ladenburg J, Termansen M, Hasler B. Assessing acceptability of two onshore wind power development schemes: a test of viewshed effects and the cumulative effects of wind turbines. Energy 2013;54:45–54. http://dx.doi. org/10.1016/j.energy.2013.02.021. [18] Meyerhoff J, Ohl C, Hartje H. Landscape externalities from onshore wind power. Energy Policy 2010;38:82–92. http://dx.doi.org/10.1016/j.enpol.2009.08.055. [19] Landscape Institute and Institute of Environmental Management & Assessment. Guidelines for Landscape and Visual Impact Assessment. 3rd Edition. Routledge, Oxford, UK, 2013. ISBN: 978-0-415-68004-2. [20] Scottish National Heritage. Visual Assessment of Wind Farms: Best Practice. Edinbourgh: University of Newcastle; 2002 Report No. F01AA303A. [21] Scottish National Heritage. Visual Representation of Wind Farms. Good Practice Guidance. March. 2006 Report No: FO3AA308/2, Edinbourgh. [22] Scottish National Heritage. Siting and designing wind farms in the landscape. December; 2009. [23] Scottish National Heritage. Assessing the cumulative impact of onshore wind energy developments. March; 2012. [24] National Research Council. Environmental Impact of Wind Energy Projects. Washington D.C.: National Academy of Sciences, 〈http://nap.edu/catalog.php? record_id=11935〉; 2007 [accessed September 2013 ]. [25] Environment Protection and Heritage Council. National Wind Farm Development Guidelines. Australia, 〈http://www.scew.gov.au/resource/future-natio nal-wind-farm-development-guidelines〉 July; 2010 [accessed September 2013].
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
[26] Union Fenosa Wind Australia. Berry Bank wind farm landscape and visual impact assessment. Final Report. July; 2009. [27] Shang H, Bishop ID. Visual thresholds for detection, recognition and visual impact in landscape settings. J Environ Psychol 2000;20:125–40. http://dx.doi. org/10.1006/jevp.1999.0153. [28] Bishop ID. Determination of thresholds of visual impact: the case of wind turbines. Environ Plan B 2002;29:707–18. http://dx.doi.org/10.1068/b12854. [29] European Union. First Report. Annex 3. Memorandum by the campaign for the Protection of Rural Wales (CPRW); 1999. [30] Campaign for the Protection of Rural Wales. Annex 3: Memorandum by the Campaign for the Protection of Rural Wales (CPRW). House of Lords Select Committee on European Union, First Report. 〈http://www.parliament.the-sta tioneryoffice.co.uk/pa/ld199900/ldselect/ldeucom/18/18a04.htm〉 [accessed September 2013] and repeated as CPRW (Campaign for the Protection of Rural Wales) (1999). Appendix 1: The Potential Visual Impact of Wind Turbines in Relation to Distance. 〈http://www.cprw.org.uk/wind/Hlords/ hlapp1.htm〉; 1999 [accessed September 2013]. [31] Sinclair G. The potential visual impact of wind turbines in relation to distance: an approach to the environmental assessment of planning proposals. Environmental Information Services; 2001.
767
[32] Sullivan R, Kirchler L, Lahti T, Roché S, Beckman K, Cantwell Bet al. Wind Turbine Visibility and Visual Impact Threshold Distances in Western Landscapes. Argonne National Laboratory. U. S. Department of Energy Office of Science Laboratory, 〈http://visualimpact.anl.gov/windvitd/docs/WindVITD. pdf〉 [accessed September 2013]. [33] Iglesias C, Asenjo V, Castellana J, Castillo E, de la Fuente G, Puente A, et al. Estudios de Paisaje: Ámbitos de Estudio y Aplicaciones Prácticas. ECOPÁS (Ed.). Madrid; 2010. [34] Farin G. Curves and surfaces for CAGD: a practical guide (The Morgan Kaufmann series in computer graphics), 5th edition, Chapter 7. ISBN: 1-55860-737-4. Academic Press, Burlington MA, Massachussetts. [35] Cowell R. Wind Power, landscape and strategic, spatial planning – the construction of “acceptable locations” in Wales. Land Use Policy 2010;27 (2):222–32. http://dx.doi.org/10.1016/j.landusepol.2009.01.006. [36] Hagerhall C, Purcell T, Taylor R. Fractal dimension of landscape silhouette outlines as a predictor of landscape preference. Environ Psychol 2004;24:247–55. http://dx.doi.org/10.1016/j.jenvp.2003.12.004. [37] Jalloulli J, Moreau G. An immersive path-based study of wind turbines' landscape: a French case in Plouguin. Renew Energy 2009;34(3):597–607 http://dx.doi.org/ 10.1016/j.renene.2008.05.036.
Contents lists available at ScienceDirect
Renewable and Sustainable Energy Reviews journal homepage: www.elsevier.com/locate/rser
A review on the Spanish Method of visual impact assessment of wind farms: SPM2 Cristina Manchado n, Valentin Gomez-Jauregui, César Otero Civil Engineering School of Santander, University of Cantabria, 39005 Cantabria, Spain
art ic l e i nf o
a b s t r a c t
Article history: Received 5 April 2014 Received in revised form 15 August 2014 Accepted 24 April 2015
This work offers a review of the so-called Spanish Method for the visual impact assessment of wind farms. The five coefficients originally proposed in the method have been analysed and discussed from several approaches: validity, efficiency, limitations and need of actualisation, among others. As a result, we establish a set of new proposals that update or modify the definition or calculation of these coefficients, but always trying to retain their original meaning. The work is complemented by a short case study in which we compare the values of the coefficients of the original Spanish Method with those arisen from our new proposal. The difference is often relevant, both in the numerical value of the coefficients and in the improvement of their ability to describe the visual effect. Finally, the new formulation of the Spanish Method opens a possibility for the public participation in several moments of the process. & 2015 Elsevier Ltd. All rights reserved.
Keywords: VIA Visual impact assessment LVIA Visual impact indicators
Contents 1. 2.
3.
4.
n
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A review of the Spanish Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Visibility coefficient of the wind farm from village (a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Visibility coefficient of village from wind farm (b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Visibility coefficient of the wind farm taken as a cuboid (c) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4. Distance coefficient between the wind farm and the village (d). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5. Population coefficient of the village (e) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6. Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion and methodological proposal for the Spanish Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Coefficient a: subdivision of the villages into areas of uniform visual impact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Coefficient a: an alternative expression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Coefficient b: visibility of the buildings of the village from the wind farm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Coefficient c: procedure for univocally obtaining the enveloping cuboid of the park . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5. Coefficient c: a discussion on the values of visibility. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6. Coefficient c: a new definition for the coefficient v. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7. Coefficient c: review and proposal for coefficient n. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8. Coefficient d: some considerations about the effect of distance on visibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9. Coefficient e: review and proposal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.10. Partial assessment factors PA1 and PA2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Visual effects of different hypotheses of design. A case study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Coefficient a . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Corresponding author. E-mail addresses: [email protected] (C. Manchado), [email protected] (V. Gomez-Jauregui), [email protected] (C. Otero).
http://dx.doi.org/10.1016/j.rser.2015.04.067 1364-0321/& 2015 Elsevier Ltd. All rights reserved.
757 757 757 757 757 759 759 759 759 759 759 759 761 761 761 762 762 763 764 764 765
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
4.2. Coefficient c: coefficients v and n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3. Coefficient d . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. The method as a software tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1. Introduction In 2004 the journal Renewable and Sustainable Energy Reviews published an article explaining “a methodology to predict, before its construction, the visual impact that a wind farm can have” [1]. The paper was groundbreaking and somehow surprising: having elapsed only 18 days between its reception and its approval, having cited just seven references (all of them national regulations related to the general problem of the deployment of wind farms in Spain), the method did not offer a case study, nor a method for its validation. In spite of this, it established a very right framework to assess several visual and perceptual aspects that were then a subject of analysis for the scientific community, the stakeholders and some other groups involved in the planning, development and use of wind farms. Moreover, the work was greatly concise, original, intuitive and logical. It proposed a rather easily replicable methodology, as well as a way to quantify some aspects considered critical in the process of visual impact assessment of wind farms. Since then, the Spanish Method has become a reference very often cited in the reviews of the contributions studying the visual impact caused by wind farms. More precisely, several works report to have made use of this method to assess impacts: (i) by contrasting their results with other ones obtained by means of public acceptance's methodologies [2,3], (ii) by comparing the visual effect of several wind farms, and the variation of their Spanish Method coefficients under a particular mitigation hypothesis [4] and (iii) by studying the time of visual exposure of a wind farm by the observers travelling along a particular motorway [5]. The Spanish Method can be integrally programmed [4]. That makes it possible to characterise the effect of one – or several – wind farms over the whole visual inventory, either to a local, regional or trans-regional level. In general, the computation time of this task is high but acceptable. The Spanish Method results more reliable to compare different scenarios than to obtain an absolute visual assessment, but that happens with most of the methods based on visual indicators [6–9]. As a scientific work, the Spanish Method needs corrections and updates. In this article we show how, once deeply reviewed, it helps to get a full meaning of the numerical characterisation of the visual effect caused by a wind farm. It is also possible to compare different mitigation proposals and gives a way to address a participatory process, what always helps to give transparency to the whole project. The work is arranged as follows: the Spanish Method is reviewed in Section 2. In Section 3, several improvements to the method are discussed. Section 4 shows a brief case study, useful to compare the values of the coefficients before and after the proposal of improvement carried out in Section 3. Section 5 describes how and where this improvement of the Spanish Method can be used currently, into a computational free use environment. Finally, in Section 6, the paper offers some conclusions as well as some reflections about the limitations of the present work, the strategy of improvement and about the research itself. 2. A review of the Spanish Method The Spanish Method was established on the basis of five coefficients, which are later integrated into a single value. It
757
765 765 766 766 766
enables access to a table that gives back the definitive impact assessment, expressed as a number into the interval [0–1], whose extreme values represent respectively a null visual impact or an unacceptable visual impact. Next, the aforementioned five coefficients are presented. 2.1. Visibility coefficient of the wind farm from village (a) The method establishes a first coefficient for each one of the villages to be analysed. If the visual impact is presumed to vary along a village, this one is divided into homogeneous subareas. Each subarea must have a uniform sight of the wind farm (this is, it must see the same number of turbines). The number of turbines seen from each area is expressed as a much per one of the total of turbines of the park. Therefore, if the coefficient a has, for example, a value of 0.6, it means that a 60% of the turbines of the park are seen from the village. Thus, such a value of 0.6 can be understood as an average value or as a probability. As usual, it is accepted that a partially seen tower is computed as a visible one. However, the next considerations cannot remain unnoticed: - No procedure is reported to obtain a subdivision of the village in homogeneous areas. So, this aspect of the method is really asystematic. - The areas having whole visibility or null visibility over the wind farm are not studied at all. However, these extreme values should be always reported in some manner. Not doing it could be misunderstood as a way to deliberately bias or hide the impact in these significant parts of the village. These issues are analysed in Sections 3.1 and 3.2. 2.2. Visibility coefficient of village from wind farm (b) The method establishes a coefficient b defined as the ratio between the number of buildings that are visible from the park and the total number of buildings in the village. This definition is particularly imprecise with respect to: – The meaning of the property “buildings seen from the park” provided that the turbines can see very different sets of buildings of a village. – The meaning of “visibility” provided that the upper floors of a building can be seen at the same time than its lower ones remain hidden by other buildings placed forward. – The way to capture the geometry of buildings and to assign them the attribute of a certain height. We will deal with this matter in Section 3.3. 2.3. Visibility coefficient of the wind farm taken as a cuboid (c) The method sets up a coefficient c that assesses the combined effect of the observer's position with respect to the park (coefficient ν) and the number of turbines it has (coefficient n). Coefficient ν forces to divide the surrounding area of the wind
758
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
farm into three different types of areas called front, diagonal and longitudinal. To determine them, it is necessary to consider an enveloping cuboid of the wind farm, as it is suggested in Fig. 1. The coefficient ν receives a value of 1.0 if the village is seeing the park from the front, a value of 0.5 if it is doing it from the diagonal zone and a value of 0.2 if it is doing it from the longitudinal area. No other intermediary positions of the village are considered. The value of the coefficient n, regarding the number of turbines of the park, is shown in Table 1. The sought coefficient c is the product ν n. But we should pay attention to these remarks: – There are an infinite amount of possible cuboids enveloping a wind farm (see for example Fig. 2) and each solution gives rise to different sets of front, diagonal and longitudinal areas. This circumstance is not admissible and should be corrected so that such areas are invariant for a given park. – The coefficient ν has always the only same three values: 1.0, 0.5 and 0.2 for respectively front, diagonal and longitudinal areas. However, this proposal is not consistent (see Fig. 3). Indeed, if the enveloping rectangle is a square, it results undefined which area is front and which one is longitudinal. In fact, in this particular situation the four edges should give rise to the same type of area, front or longitudinal. Furthermore, in such a case, maybe the diagonal area (not the front one) deserves the maximum score for coefficient ν.
– It is not unusual that a village is placed just on the edge of two different areas. For such a case, the value of the coefficient ν has not been defined.
Regarding to the table that defines the coefficient n, the ranks should be, at least, updated. As an example, the method gives the neutral effect (coefficient n¼ 1.00) to parks containing between 11 and 20 turbines, which does not fit to the population sensibility nowadays nor to the estimations of other similar tables. In Sections 3.5–3.7 we will discuss all these considerations.
Fig. 3. If the cuboid becomes a square, front and longitudinal areas can be swappable.
Fig. 1. Wind farm with different sides of viewing.
Table 1 Coefficient n for calculating coefficient c. Number of windmills
n factor
1–3 4–10 11–20 21–30 430
0.50 0.90 1.00 1.05 1.10
Fig. 2. For a given wind farm layout, there are infinite possible enveloping cuboids.
Table 2 Coefficient d, depending on the distance. Distance
Coefficient d
xo 500 m 500 m o xo 6000 m x46000 m
1.00 1.05–0.0002x 0.10
Fig. 4. Linear variation of coefficient d.
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
2.4. Distance coefficient between the wind farm and the village (d) The method defines a coefficient d that considers the incidence of the distance between the wind farm and the village. The same as with the coefficient n, these values (see Table 2) need to be updated, as well as the thresholds of visibility themselves. Furthermore, the method posits the variation of this coefficient according to a linear law (see Fig. 4). The effect of the distance on the visibility is mainly depending on the size of the object and on the changing conditions of their contrast against the background. The laws describing these two phenomena are not linear; not even both are polynomial. In Section 3.9, we will discuss all these considerations. In Section 3.8 this coefficient is discussed more in depth. 2.5. Population coefficient of the village (e) A last coefficient, e, reflects the effect of population. This coefficient is nil for unpopulated villages, it has a value of 0.20 when population varies between 0 and 5 inhabitants and gets its maximum (e¼ 1.00) for 300 or more inhabitants. It can be argued that: – As this estimation is often subject of controversy, it is recommended that these values can result from a participatory process. – It could be advisable to consider a value of the coefficient e bigger than 1.00 for populations of some particularly fragile villages (having special historic, cultural or social values or, as well, having a highly damaged or altered landscape already). In Section 3.9 this coefficient is discussed more in depth. 2.6. Evaluation Once these five coefficients have been calculated, the method sets up the overall assessments factors PA1 and PA2, expressed as follows: PA1 ¼ a b c d PA2 ¼ a b c d e Finally, the method offers a table to express the assessment in qualitative terms (high, medium, low, etc.) and gives an expression to average the total visual effect when all the villages in the area of influence of the wind farm are considered. These last details are not particularly relevant for our work and can be read in the original paper.
3. Discussion and methodological proposal for the Spanish Method Initially, we programmed the Spanish Method according to their original specifications [4]. In fact, most of the comments shown in Section 2 arise as a result of the different proofs of concept carried out to test such implementation. The visibility calculations for the coefficients a and b make necessary the computation of the ZVI (Zones of Visual Influence) associated to each turbine of the wind farm; as it has been detailed in [4,10], these ZVI are recorded into a numerical structure that keeps in each pixel of the DEM (Digital Elevation Model) the visibility conditions in each and every one of the turbines. This structure is called CHESSBOARD. Some computational properties of CHESSBOARD give the visibility status of any turbine i with respect to a pixel (x, y). The index i is applied to every one of the turbines
759
and the coordinates x and y cover all the DEM considered in the study. 3.1. Coefficient a: subdivision of the villages into areas of uniform visual impact CHESSBOARD is particularly suitable to solve the intended subdivision of the villages into areas of uniform visual impact simply considering that each one of its pixels is one of such areas. Indeed, we then only need to select the collection of pixels belonging to the village and, next, to obtain for each one of these pixels its accumulated value of visibility. The computational property vis(i) has the value 0 if the turbine is not visible and 1 if it is seen from the pixel. The index i goes over each and every turbine of the park (k in total). Each pixel p(x, y) of the village has then an associated value Sp: SP ¼
k X
CHESSBOARDðx; yÞ:visðiÞ
ð1Þ
i¼1
This makes it possible to know how many pixels have a value Sp¼0 (these pixels have no visibility over the park), a value Sp ¼1, Sp¼2, etc. until Sp ¼k (these pixels have a full visibility of the park). Next, the variable SUMA0 stores the amount of pixels that have a value Sp¼0, SUMA1 keeps those that have a value Sp ¼1, etc., up to SUMAk. Being Np the total number of pixels that belong to the village and k the number of turbines of the park, the coefficient a is unambiguously defined as: Pk i ¼ 0 ði SUMAi Þ=k a¼ ð2Þ NP 3.2. Coefficient a: an alternative expression An indicator obtained simply by computing the visibility of the pixels of a ZVI has been reported by other researchers [8,11–13]. Thus, we suggest that the coefficient a can be alternatively expressed by means of a raw visibility value such as Sp (Eq. (1)). This leads to an expression for a as the one expressed in Eq. (3). Using the value of Sp for each pixel in the village provides two added advantages: (i) the coefficient a can be graphically represented in a thematic map and (ii) the areas free of visual effect or the ones that have visibility to the whole wind farm can be directly depicted. a¼
Number of pixels of the village having visibility to the park Number of pixels of the village ð3Þ
3.3. Coefficient b: visibility of the buildings of the village from the wind farm The three comments stated in Section 2.2 lead to a conclusion: the coefficient b cannot be calculated, not even approximately, by means of algorithms based on ZVI techniques (and in general, in computational techniques based on raster data). Fig. 6 helps to understand that techniques such as ray-tracing, hidden surface removal or similar [14] are the suitable ones to solve adequately this problem. Nevertheless, sometimes we have adapted the idea sketched by this coefficient b in order to express the visual effect specifically concentrated on the zones of greater density of population into the village. Once a threshold of density of population by pixel is defined, the set S of those pixels in the village having a density of population equal to or bigger than the threshold is considered. From this set, the subset Sv of those pixels that are seen from the
760
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
Fig. 5. Villages with different degrees of visibility. White pixels represent subareas free of visual effect.
Fig. 6. Group of buildings with different degrees of visibility.
Fig. 7. Unambiguous algorithm for obtaining the enveloping cuboid of the wind farm WF. (a) LE ¼WT1–WT6, Pr ¼ WT4, Ps ¼WT3. (b) LE¼ WT4–WT5, Pr ¼WT2.
wind farm gives the value to the coefficient b. The condition of visibility is given by the ZVI itself. b¼
Sv : Number of pixels of the set S that have visibility from the park S : pixels having a density of population upper than a threshold
ð4Þ
In addition, fortunately Eq. (4) enables the Spanish Method to be applied to other elements of the visual inventory different from villages, such as roads, protected areas, cultural sites, panoramic itineraries, etc. In these other cases, the coefficient b can be applied to those singular subareas that can exist inside the area
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
object of the study (for example, the areas of bigger amount of visitors inside a protected area).
3.4. Coefficient c: procedure for univocally obtaining the enveloping cuboid of the park Let us call WF the wind farm and WT1, WT2,… WTk the set of its turbines (see Fig. 7). This is a procedure to obtain univocally its enveloping cuboid: – Obtain the set of distances dij ¼distance(WTi–WTj) between couples of turbines of WF.
761
– Then the longest edge of the cuboid is defined by the segment LE¼WTi–WTj whose associated distance is maximum: LE ¼MAX (dij). – Next, obtain all the distances between the k turbines WTi and the segment k. – Two possible situations can arise: There are turbines in both sides of the segment LE (Fig. 5a). In such a case, the enveloping cuboid is defined by the parallel lines to LE containing the points Pr and Ps. These points have the largest distance to LE from left and right. There are no turbines in both sides of the segment LE (Fig. 5b). In such a case, the cuboid is defined by the same segment LE and its parallel containing the most distant turbine Pr.
3.5. Coefficient c: a discussion on the values of visibility Front and longitudinal areas cover angles of 151 from the edges of the cuboid (see Fig. 1). This leads to the diagonal areas being angles of 601 applied on the vertices of the enveloping cuboid. The coefficient ν is valued to 0.2, 0.5 and 1.0, respectively for the longitudinal, diagonal or frontal areas (even though these values are not any way justified). However, in general the shape of the cuboid can vary from a perfect square to a really oblong rectangle. In particular (as it was exposed in Section 2.3) if the cuboid becomes a square, this procedure gives inconsistent results. Our first try to solve this problem took into account the ratio r ¼SE/LE between the shortest and the longest edges of the rectangle, so that the coefficient ν was redefined this way: – – –
ν ¼ 1.0 for villages placed on the front area. ν ¼ r ¼SE/LE for villages placed on the longitudinal area. ν ¼ (1 þr)/2 for villages placed on the diagonal area.
This designation results again unsatisfactory; in many cases, the maximum visual effect is perceived from the diagonal. That could lead to try a second redefinition of the coefficient ν; D being the measure of the diagonal of the cuboid: – – –
ν ¼ LE/D for villages placed on the front area. ν ¼ SE/D for villages placed on the longitudinal area. ν ¼ 1.0 for villages placed on the diagonal area.
However, this expression remains based on two statements – why choose only three types of areas and why choose an angle of 151 to define the border lines among them – that are not justified. Such a division in three types of areas reduces the nature of the problem, simplifying it excessively. Maybe a division into only three areas (front, diagonal and longitudinal) is not the best way to represent the variation of the visibility coefficient ν, and some other approaches should be considered. 3.6. Coefficient c: a new definition for the coefficient v
Fig. 8. Coefficient d modelled by Hermite's Splines. (a) Boundary conditions for the first example. (b) Solution for the first example (Eq. (8)). (c) Boundary conditions for the second example. (d) Solution for the second example (Eq. (10)).
For modelling the effect of the visual exposure of the park, several works draw on the magnitude of the visibility angle [8] or the solid angle [9]. Both are continuous variables. It is therefore advisable to abandon the resource of the enveloping cuboid and to represent the coefficient directly by means of the visibility angle spanned by the wind farm from a camera point. An algorithm for its calculation is described in [15]. It can be argued that this new coefficient ν does not have a normalised range of variation between 0 and 1. Nevertheless, this can be easily achieved, for example, assigning to the visual angle of π radians the value ν ¼ 1. It can be noticed as well that this new
762
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
coefficient ν is affected by distance (the previous one was not). This is true and we consider it is better to effectively measure the visual effect according to the position of the observer. Besides, we can always consider a homogenised expression such as ν d (where d is the distance from the observer to the wind farm) for this coefficient. The final coefficient ν associated to a village (or any other element of the visual inventory) is the average of the ones obtained for each one of their pixels. 3.7. Coefficient c: review and proposal for coefficient n A lot of scientific literature [16–18], good practice guides [19–25], case studies [26], regional and national regulations offer proposals to this respect, usually expressed as tables of assessment. It is really unnecessary that the Spanish Method opts just for one of these tables in particular; this choice can come from a participatory agreement. Moreover, if the Spanish Method is completely implemented as a computational application, it means that it is very easy to run different hypotheses of assessment according to different tables of values for this coefficient n. We should think about this other issue as well: if a wind farm has 8 turbines (corresponding to n ¼0.9) but from the village V only 3 (corresponding to n ¼ 0.5) are seen, should we give n ¼0.9 or n ¼0.5 to V? We show some comparisons to this respect in Section 4.2. 3.8. Coefficient d: some considerations about the effect of distance on visibility
model for the coefficient d can be proposed by techniques of Computer Aided Geometric Design (CAGD). This branch of mathematics has been used – and it is still used – in many different processes where the geometry of objects, or the law governing a certain phenomenon, has to be modelled depending on the boundary conditions, because they are the only source of available data. The CAGD offers many standard procedures to interpolate or approximate curves to many different combinations of boundary conditions. One of these combinations is the problem that we face here. In the example of Fig. 8a, we only know the threshold of maximum visibility (the coefficient d is set to 1.0 for the first 8 km) and the threshold of minimum visibility (it is considered that the coefficient d remains constant with a value of 0.1 for a distance of 22 km onwards, until the studio area is over). Then, the objective is to define a polynomial curve that passes through two specific points with the given tangents. This is a standard problem in CAGD [34] and it is formulated as a Hermite's spline, which is expressed, if we choose a third-degree polynomial, as follows: P ðuÞ ¼
3 X
P i Bi ðuÞ
ð5Þ
i¼0
By definition, the parameter u varies on the unit interval [0, 1]. The functions Bi(u) are, in this case, the Hermite's third-degree polynomials whose expressions [34] are: B0 ðuÞ ¼ 2u3 3u2 þ 1 B1 ðuÞ ¼ 2u3 þ 3u2 B2 ðuÞ ¼ u3 2u2 þ u
The tables originally proposed by the Spanish Method have become obsolete since the size of turbines is currently much bigger. With respect to the thresholds of distance, the literature is wide [6,27–32], thus it is again advisable to decide these values in a participatory process. Such events always need a labour of scientific and technical dissemination previously, in which it is possible to analyse as well the intermediate thresholds of distance in which the visual effects are considered of high, medium or low intensity. As a simple example, for turbines 140 m high (tower and blades), we usually define an area of high visual effect at 5 miles around the machine, an area of medium effect at a distance between 5 and 10 miles and an area of low intensity at a distance from 10 to 16 miles. These values can be a reasonable extrapolation of the ranks proposed in Refs. [21,24,27]. It is also convenient to think about the reason why this law (see Table 2 and Fig. 4) is linear and consider if this hypothesis is the best. Let us focus on how the visual effect has been defined at a distance of 500 m. At this point, the visual intrusion suddenly and sharply starts to decrease, even though these types of natural phenomena are better described by laws having a smooth and completely continuous gradient. Visibility and visual effect of an object vary with the distance according to two main causes: the variation of its size and the variation of its contrast with the background. The first effect follows a well-known quadratic law; the second effect is known as scattering, which follows an exponential law [27,33]. Obviously, this latter law accepts to be approximated, but the approximant polynomial does not need to be of the order 2; it can be cubic, quartic, etc. as well. Some works, mainly based on the perceptions and assessments of photographs made by expert [7] or non-expert observers [27] give this law a quadratic expression. It seems clear that these second order polynomials are more suitable than the linear ones. Nevertheless this model must be expressed by a law consistent with the boundary conditions, which are the visibility thresholds. Since a definition of the visibility thresholds exists, the mathematical
B3 ðuÞ ¼ u3 u2
ð6Þ
The boundary conditions, defined by the threshold values, are as follows (see Fig. 8a): P 0 ¼ ð8; 1Þ; which means that the coefficient d is 1:0 at 8 km: P 1 ¼ ð22; 0:1Þ; which means that the coefficient d is 0:1 at 22 km: P 2 ¼ ð1; 0Þ; which means that the tangent in point P 0 is horizontal: P 3 ¼ ð1; 0Þ; which means that tangent in point P 1 is horizontal:
ð7Þ
Also by definition, the Hermite's spline is set to be P0 at u ¼0 and P1 at u ¼1, being the value of the tangent P2 at u ¼ 0 and P3 at u¼ 1. As a result, the desired curve can be expressed with these parametric expressions: X ðuÞ ¼ 8 B0 ðuÞ þ 22 B1 ðuÞ þ 1 B2 ðuÞ þ 1 B3 ðuÞ ¼ 26 u3 þ 39 u2 þ u þ 8 Y ðuÞ ¼ 1 B0 ðuÞ þ 0:1 B1 ðuÞþ 0 B2 ðuÞþ 0 B3 ðuÞ ¼ 1:8 u3 2:7 u2 þ 1
ð8Þ
The sought curve (Fig. 8b) fits the proposed conditions for the visibility thresholds. The above described formulation can be generalised to any type visibility threshold. Thus, as a simple example, if we would like to model the visual effect (coefficient d) for other atmospheric conditions that would give a constant threshold of maximum visibility for the first four kilometres, and a threshold of minimum visibility linearly attenuated from a value of coefficient d equal to 0.1 at a distance of 15 km until it wears off (coefficient d¼0), at a distance of 20 km, the model described maintains the expressions [Eqs. (5) and (6)]. It also maintains the domain of the parameter u on the unit interval [0, 1], but the conditions of [Eq. (7)] become those exposed in [Eq. (9)], below (see Fig. 8c): P 0 ¼ ð4; 1Þ; which means that coefficient d is 1 at 4 km: P 1 ¼ ð15; 0:1Þ; which means that coefficient d is 0:1 at 15 km: P 2 ¼ ð1; 0Þ; which means that the tangent at point P 0 is horizontal P 3 ¼ ð1; 0:02Þ; which means that the tangent at ð9Þ point P 1 is 0:1=ð20 15Þ
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
763
Fig. 9. Layout of the wind farm, villages and background map. The crosses (X) represent the turbines. The circles represent thresholds at 8, 22 and 30 km. The legend at left describes the visibility map for the 8 turbines.
Table 3 Coefficient a calculated according to Eq. (1) (left column) and Eq. (2) (right column). Name
Coefficient a according to Eq. (1)
Coefficient a according to Eq. (2)
ESCOBEDO FRESNEDO HAZAS DE CESTO LIERGANES RIAÑO RIBA SARÓN SOLORZANO SOMO
0.217 0.000 0.898 0.369 0.997 0.000 0.058 0.778 0.802
0.305 0.000 1.000 0.533 1.000 0.000 0.120 1.000 0.823
These conditions would give these new parametric expressions: X ðuÞ ¼ 4 B0 ðuÞ þ 15 B1 ðuÞ þ1 B2 ðuÞ þ 1 B3 ðuÞ ¼ 20 u3 þ 30 u2 þ u þ 4:0 Y ðuÞ ¼ 1 B0 ðuÞ þ 0:1 B1 ðuÞ þ 0 B2 ðuÞ 0:02 B3 ðuÞ ¼ 1:78 u3 –2:68 u2 þ1:0
ð10Þ
The result is shown in Fig. 8d. In a strict mathematical sense, the curve may have a degree different to third-degree. That is a decision of the expert using this method. If, for example, a Hermite's interpolation with fourth-degree polynomials is used, sets of four terms as expressed in [Eq. (7) or (9)] are no longer sufficient to mathematically define the model. The researcher or designer gains a new degree of freedom (DoF) that, properly chosen, will enable to debug the model scientifically or technically. It is obvious now that this approach opens the possibility of testing models for the coefficient d of quartic Hermite splines (degree four) wherein the fifth degree of freedom can be the coefficient for the atmospheric attenuation of the scattering. That is the research line in which part of our current work is focused on this moment.
Fig. 10. Villages considered in the case study and front, diagonal and longitudinal areas.
3.9. Coefficient e: review and proposal The definition of this indicator admits the same reflections expressed in Section 3.7 with respect to the coefficient n of the indicator c. As there are not few scientific studies [6,12,13,35] and national or regional regulations [19–25] about it, we consider that the most advisable for the designer is to elaborate the best dissemination sessions or material and try to achieve a definitive table for the indicator e by means of participation and agreement.
764
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
3.10. Partial assessment factors PA1 and PA2 These global indicators have not been studied in this review. Our opinion is that the Spanish Method is rich and valuable when coefficients a–e are individually analysed, but we do not trust too much on the meaning of the integration of these coefficients into a single number. Among many other reasons, such a general coefficient could compensate possible extreme values of their individual coefficients, what not always helps to make a good analysis. Moreover, the
Spanish Method can be a good tool to understand and to express the visual effects, but the final assessment must be conducted by mean of participation and agreement in some manner.
4. Visual effects of different hypotheses of design. A case study The next case study obtains the original coefficients of the Spanish Method and compares them to those arising from the new Table 5 Comparison of values of coefficient d for original method and new proposals.
Table 4 Comparison of values for coefficient ν. (N.V. means “no visibility”) and coefficient n. Name
Original coeff. ν
Modified coeff. ν (rad)
Original coeff. n
Modified coeff. n
ESCOBEDO FRESNEDO HAZAS DE CESTO LIERGANES RIAÑO RIBA SARÓN SOLORZANO SOMO
0.50 0.20 1.00
0.06 N.V. 0.52
0.90 0.90 0.90
0.90 0.00 0.90
1.00 0.20 0.20 1.00 0.50 0.50
0.12 0.24 N.V. 0.05 0.55 0.15
0.90 0.90 0.90 0.90 0.90 0.90
0.90 0.90 0.00 0.50 0.90 0.90
Village
Distance (km)
Original coefficient d
Updated coefficient d (linear)
Updated coefficient d (cubic)
ESCOBEDO FRESNEDO HAZAS DE CESTO LIERGANES RIAÑO RIBA SARÓN SOLORZANO SOMO
23036.71 25446.46 2570.99
0.100 0.100 0.538
0.100 0.100 1.000
0.100 0.100 1.000
11606.57 4938.12 13067.61 20687.98 2753.63 11642.27
0.100 0.055 0.100 0.100 0.494 0.100
0.770 1.000 0.670 0.180 1.000 0.770
0.852 1.000 0.732 0.122 1.000 0.849
Fig. 11. Horizontal angle from the village of Sarón.
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
proposal, to provide an order of the magnitude of the change produced. The experiment is focused on coefficients a, c and d. We consider a project of wind farm in Cantabria (Spain), made up by eight turbines disposed as a cluster (non linear), as it is shown in Fig. 9. The villages considered in the area are representative of different conditions of visibility, distances, relative situations with respect to the park, etc. 4.1. Coefficient a In Table 3, the central column gives the values according to Eq. (1) (see Section 3.1) and the right column shows the ones obtained according to Eq. (2) (see Section 3.2). The first expression gives more attenuated results. The difference obtained can have proportions up to 44% (Liérganes) or 40.5% (Escobedo): the choice of one option or another is something relevant and must be justified. Eq. (1) had this development, for example, in the village of Solórzano: Pk i ¼ 0 ði SUMAi Þ=WM a¼ NP ð1 4 þ 27 5 þ 176 6 þ 110 7 þ 8 8Þ=8 ¼ 0:778 ¼ 322 However, Eq. (2) gives a coefficient a ¼1.0 because any pixel in Solórzano is free of visibility towards the wind farm.
765
4.2. Coefficient c: coefficients v and n Fig. 10 shows the disposition of the villages with respect to the front, diagonal and longitudinal areas, according to the specification of the original method. The original values of coefficient ν are presented in the second column of Table 4. The third column shows the value of ν expressed as visibility angle, such as it has been proposed in Section 3.7. The fourth column shows the coefficient n according to its original definition and the fifth column shows the new value, when not all but only the viewed turbines are considered for obtaining n in Table 1. The values of the coefficient ν differ, sometimes intensely, from those obtained in the new proposal. For example, let us consider the case of the villages of Hazas de Cesto and Sarón (see Table 4 and Fig. 11). Even though both are situated in a front area of the park (i.e., ν ¼1.00 in the original specification), they have a very different visibility angle (0.52 rad in Hazas de Cesto and 0.05 in Sarón). This difference is due to: (i) the angle of visibility decreases with the distance and (ii) the angle of visibility considers strictly the number of turbines actually visible from each village. However, the original method assumes that the whole wind farm is unconditionally seen from anywhere in the village, which is not always true at all. Indeed, a deeper inspection for Sarón shows that at most 4 of the 8 turbines are seen, which considerably reduces the visual effect experienced and makes it clear that the original definition of the method resulted to be too conservative. The same happens with coefficient n. It makes no sense to assign for Sarón a coefficient n ¼0.9, associated to 8 turbines (the whole park), when it is impossible to see from any pixel of the village more than four. Altogether, considering ν and n, the original Spanish Method tended to overestimate the coefficient c.
4.3. Coefficient d
Fig. 12. Law of variation of coefficient d: a comparison between the linear model (blue), and the cubic model (red). Differences for several villages are depicted. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Table 5 shows a comparison among the original value of the coefficient d, a new value of this coefficient once the thresholds of distance have been updated to 8 and 22 km and a third value, considering a cubic variation of the visual effect (see Fig. 12). As expected, updating the law involves deep changes in the values of this coefficient. When this law changes from linear to cubic, the variations are lower but not irrelevant. In villages such as
Fig. 13. A short summary of this review of visual impact indicators.
766
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
Liérganes, Riba or Somo the updated coefficient reaches increments up to 11%.
5. The method as a software tool The Spanish Method has been implemented into a tool called MOYSES [4,5,10] that is currently available in http://193.144.189. 246/moyses. A user can create input layers and run the programme assisted by a comprehensive tutorial. MOYSES computes more visibility coefficients than those studied in this article: apart from the Spanish Method, it can give indicators of prominence, contrast, area, length, duration of the intrusion and some other more features. Fractality [7,36], rythm or uniformity [7] and movement of blades [6,37] are indicators not yet obtained but currently in progress.
6. Conclusion We have reviewed, discussed and reformulated different issues on the Spanish Method for the visual assessment of wind farms. Next, we summarise the most relevant conclusions: – In general, the 5 coefficients used in the Spanish Method are applicable to the whole visual inventory, not only to villages. – Coefficient a was not univocally defined: it has been done in this contribution. – Coefficient a can express an average value for the visibility effect or a raw value as well. Both values give complementary information and may differ considerably from each other. – Coefficient b can be very complicated, expensive to compute and not always reliable enough. Nonetheless, a singular case, when this coefficient is calculated considering the height of buildings equal to cero, is full of meaning, easy to obtain and very advisable. – Coefficient ν of the coefficient c was not univocally defined: it has been done in this contribution. – The coefficient ν of the coefficient c does not successfully express the effect of visual incidence by means of a discrete variable and can be substituted by the visibility angle, frequently considered in similar studies. – The coefficient n of the coefficient c needs to be updated. Even though there are several references to carry out such an update, we suggest taking this issue as the basis for a participatory decision. – A wind farm can be seen from a village only partially. In such a case the coefficient n of the coefficient c should not be computed considering the total number of turbines of the park but only those that are really seen. – Coefficient d needs to be updated; several references can be used to do it even though we suggest taking this issue as the basis for a participatory decision. Moreover, the law of variation for this coefficient according to the distance can be not linear. Quadratic or cubic laws represent the thresholds of visibility better. – Coefficient e needs to be updated as well. Again, this can be addressed by means of a participatory process. In Fig. 13 we have integrated these proposals into an only user form. Each one of these values or options makes up a different setting chosen by the designer for each particular hypothesis of design. To state that the application MOYSES (see Section 5) uses the Spanish Method is not completely exact, since in fact it currently implements the modifications explained in this paper. We refer
this derived method as SPM2. This name (coming from Spanish Method 2) is an acknowledgement to the original methodology; it tries however to remove any explicit reference of nationality (despite of these authors are not less Spanish than the Method itself). References [1] Hurtado JP, Fernández J, Parrondo JL, Blanco E. Spanish method of visual impact evaluation in wind farms. Renew Sustain Energy Rev 2004;8:483–91. http://dx.doi.org/10.1016/j.rser.2003.12.009. [2] Theocharis T, Tsouchlaraki A, Tsiropoulos M, Serpetsidakis M. Visual impact evaluation of a wind park in a Greek island. Appl Energy 2009;86:546–53. http://dx.doi.org/10.1016/j.apenergy.2008.08.013. [3] Quianna W, MwirigiM’Ikiugu M, Isami K. Visual impact evaluation of wind farms: a case study of Choshi City, Japan. Civ Environ Res 2013;3(7):97–106. [4] Manchado C, Otero C, Gómez-Jáuregui V, Arias R, Bruschi V, Cendrero A. Visibility analysis and visibility software for the optimisation of wind farm design. Renew Energy 2013;60:388–401. http://dx.doi.org/10.1016/j.renene.2013.05.026. [5] Manchado C, Gomez-Jauregui V, Otero C. Moyses v4.0 modeller and simulator for visual impact assessment. In: Proceedings of the international congress in energy and environment engineering and management. Lisbon, Portugal; 2013. 978-989-98406-4-5. [6] Bishop ID, Miller D. Visual assessment of off-shore wind turbines: the influence of distance, contrast, movement and social variables. Renew Energy 2007;32(5):814–31. http://dx.doi.org/10.1016/j.renene.2006.03.009. [7] Torres Sibille AC, Cloquell-Ballester VA, Cloquell-Ballester VA, Darton R. Development and validation of a multicriteria indicator for the assessment of objective aesthetic impact of wind farms. Renew Sustain Energy Rev 2009;13(1):40–66. http://dx.doi.org/10.1016/j.rser.2007.05.002. [8] Rodrigues M, Montañés C, Fueyo R. A method for the assessment of the visual impact caused by the large-scale deployment of renewable-energy facilities. Environ Impact Assess 2010;30:240–6. http://dx.doi.org/10.1016/j.eiar.2009.10.004. [9] Molina-Ruiz J, Martínez-Sánchez MJ, Pérez-Sirvent C, Tudela-Serrano ML, García ML. Developing and applying a GIS-assisted approach to evaluate visual impact in wind farms. Renew Energy 2011;36(3):1125–32. http://dx. doi.org/10.1016/j.renene.2010.08.041. [10] Otero C, Cendrero A, Bruschi V, Togores R, Manchado C, Arias R. Moyses v3.0 (Modeller and Simulator for Visual Impact assessment). In: Proceedings of the 2007 international workshop on computing in civil engineering. Pittsburgh, Pennsylvania. ISBN: 978-0-7844-0937-4; 2007. p. 486–93. [11] Mollër B. Changing wind power landscapes: regional assessment of visual impact on land use and population in Northern Jutland, Denmark. Appl Energy 2006;83:477–94. http://dx.doi.org/10.1016/j.apenergy.2005.04.004. [12] Mollër B. Spatial analyses of emerging and fading wind energy landscapes in Denmark. Land Use Policy 2010;27:233–41. http://dx.doi.org/10.1016/j. landusepol.2009.06.001. [13] Wood G. Is what you see what you get? post-development auditing of methods used for predicting the zone of visual influence in EIA Environ Impact Assess 2000;20:537–56. http://dx.doi.org/10.1016/S0195-9255(00) 00055-X. [14] Hearn D, Baker P. Computer graphics. Upper Saddle River, New Jersey: Prentice Hall PTR; 1994. [15] Otero C, Manchado C, Arias R, Iglesias A, Gálvez A. Method for mitigating visual impact. Patent WO2012049340A1.PCT/ES2011/000295. Published 19/04/2012. [16] Ladenburg J, Möller B. Attitude and acceptance of offshore wind farms – The influence of travel time and wind farm attributes. Renew Sustain Energy Rev 2011;18:4223–35. http://dx.doi.org/10.1016/j.rser.2011.07.130. [17] Ladenburg J, Termansen M, Hasler B. Assessing acceptability of two onshore wind power development schemes: a test of viewshed effects and the cumulative effects of wind turbines. Energy 2013;54:45–54. http://dx.doi. org/10.1016/j.energy.2013.02.021. [18] Meyerhoff J, Ohl C, Hartje H. Landscape externalities from onshore wind power. Energy Policy 2010;38:82–92. http://dx.doi.org/10.1016/j.enpol.2009.08.055. [19] Landscape Institute and Institute of Environmental Management & Assessment. Guidelines for Landscape and Visual Impact Assessment. 3rd Edition. Routledge, Oxford, UK, 2013. ISBN: 978-0-415-68004-2. [20] Scottish National Heritage. Visual Assessment of Wind Farms: Best Practice. Edinbourgh: University of Newcastle; 2002 Report No. F01AA303A. [21] Scottish National Heritage. Visual Representation of Wind Farms. Good Practice Guidance. March. 2006 Report No: FO3AA308/2, Edinbourgh. [22] Scottish National Heritage. Siting and designing wind farms in the landscape. December; 2009. [23] Scottish National Heritage. Assessing the cumulative impact of onshore wind energy developments. March; 2012. [24] National Research Council. Environmental Impact of Wind Energy Projects. Washington D.C.: National Academy of Sciences, 〈http://nap.edu/catalog.php? record_id=11935〉; 2007 [accessed September 2013 ]. [25] Environment Protection and Heritage Council. National Wind Farm Development Guidelines. Australia, 〈http://www.scew.gov.au/resource/future-natio nal-wind-farm-development-guidelines〉 July; 2010 [accessed September 2013].
C. Manchado et al. / Renewable and Sustainable Energy Reviews 49 (2015) 756–767
[26] Union Fenosa Wind Australia. Berry Bank wind farm landscape and visual impact assessment. Final Report. July; 2009. [27] Shang H, Bishop ID. Visual thresholds for detection, recognition and visual impact in landscape settings. J Environ Psychol 2000;20:125–40. http://dx.doi. org/10.1006/jevp.1999.0153. [28] Bishop ID. Determination of thresholds of visual impact: the case of wind turbines. Environ Plan B 2002;29:707–18. http://dx.doi.org/10.1068/b12854. [29] European Union. First Report. Annex 3. Memorandum by the campaign for the Protection of Rural Wales (CPRW); 1999. [30] Campaign for the Protection of Rural Wales. Annex 3: Memorandum by the Campaign for the Protection of Rural Wales (CPRW). House of Lords Select Committee on European Union, First Report. 〈http://www.parliament.the-sta tioneryoffice.co.uk/pa/ld199900/ldselect/ldeucom/18/18a04.htm〉 [accessed September 2013] and repeated as CPRW (Campaign for the Protection of Rural Wales) (1999). Appendix 1: The Potential Visual Impact of Wind Turbines in Relation to Distance. 〈http://www.cprw.org.uk/wind/Hlords/ hlapp1.htm〉; 1999 [accessed September 2013]. [31] Sinclair G. The potential visual impact of wind turbines in relation to distance: an approach to the environmental assessment of planning proposals. Environmental Information Services; 2001.
767
[32] Sullivan R, Kirchler L, Lahti T, Roché S, Beckman K, Cantwell Bet al. Wind Turbine Visibility and Visual Impact Threshold Distances in Western Landscapes. Argonne National Laboratory. U. S. Department of Energy Office of Science Laboratory, 〈http://visualimpact.anl.gov/windvitd/docs/WindVITD. pdf〉 [accessed September 2013]. [33] Iglesias C, Asenjo V, Castellana J, Castillo E, de la Fuente G, Puente A, et al. Estudios de Paisaje: Ámbitos de Estudio y Aplicaciones Prácticas. ECOPÁS (Ed.). Madrid; 2010. [34] Farin G. Curves and surfaces for CAGD: a practical guide (The Morgan Kaufmann series in computer graphics), 5th edition, Chapter 7. ISBN: 1-55860-737-4. Academic Press, Burlington MA, Massachussetts. [35] Cowell R. Wind Power, landscape and strategic, spatial planning – the construction of “acceptable locations” in Wales. Land Use Policy 2010;27 (2):222–32. http://dx.doi.org/10.1016/j.landusepol.2009.01.006. [36] Hagerhall C, Purcell T, Taylor R. Fractal dimension of landscape silhouette outlines as a predictor of landscape preference. Environ Psychol 2004;24:247–55. http://dx.doi.org/10.1016/j.jenvp.2003.12.004. [37] Jalloulli J, Moreau G. An immersive path-based study of wind turbines' landscape: a French case in Plouguin. Renew Energy 2009;34(3):597–607 http://dx.doi.org/ 10.1016/j.renene.2008.05.036.