Research Paper
The implications for visual simulation and analysis of temporal variation in the visibility of wind turbines

https://doi.org/10.1016/j.landurbplan.2018.12.004Get rights and content

Highlights

  • Visual simulation of wind farms should include at least 30° variation in blade orientation.

  • Impact analysis can incorporate visibility data, and its effect on color contrast, through the extinction coefficient.

  • Use of the worst-case scenario in simulation and analysis can be moderated.

Abstract

The visual impact of wind turbines is a central issue in their public acceptance. New wind farm proposals are commonly subject to visual simulation and visual impact assessment. Guidelines for both these processes are already used in a number of jurisdictions and there is widespread interest in making simulation and impact assessments as meaningful as possible. To a large degree the guidelines for both processes tend to be based on worst-case (full-frontal) conditions. There are two quite good reasons for this. Firstly, the worst-case sets a boundary for visual impact. Secondly, we seldom know enough about how visibility or visual impact changes over time and what ‘typical’ conditions are like – or how the range of conditions is distributed. This paper argues that if we can address this knowledge gap, then we may also be able to take a more nuanced approach to simulation and analysis. It should be possible, using widespread atmospheric visibility data, to determine the temporal distribution of visual impacts at a particular location rather computing a single estimate. It may also be valid to create simulations of both worst-case and more typical conditions. This paper explores the key variables affecting visual impacts – visual magnitude and color difference – and how they may be monitored and analyzed efficiently and effectively. A wind farm in southern Victoria, Australia is used as the case study. Recommendations are made on how the approaches could be used more widely.

Introduction

The generation of energy for domestic and industrial usage it a controversial topic in many parts of the world. All energy production options have external (non-market) costs which affect their acceptance. For wind energy, whether onshore or offshore, one of the major externalities is their impact on visual amenity. Understanding of, and ideally quantification of, this factor has been the subject of considerable research over the last 30 years (at least back to Wolsink, 1989). There are two main aspects to determination of visual impacts of wind energy: simulation and analysis. Visual simulation involves creation of a virtual wind farm such that people who may be affected can see how the installation will appear when built. This may involve the use of photo montage (Bureau of Ocean Energy Management, 2012, Landscape Institute, 2018, Scottish Natural Heritage, 2017), 3D creation of a virtual exploration environment (Bishop and Stock, 2010, Ruotolo et al., 2012, Jallouli and Moreau, 2011) or augmented reality (Bishop, 2015, Grassi and Klein, 2016). Simulation may also include acoustics (Ribe et al., 2018) and mitigation measures (Bishop & Stock, 2010). Analysis involves determination of the impacts based on visibility analysis (widely available in popular geographic information systems and generally based on algorithms first implemented by Travis, Elsner, Iverson, & Johnson, 1975) and can include factors such as the nature of the local landscape (Lothian, 2008), the number of turbines (Torres-Sibille et al., 2009, Tsoutsos et al., 2009) their size (Tsoutsos et al., 2009), distance (Bishop & Miller, 2007), design (Torres-Sibille et al., 2009) and mitigation measures (Manchado et al., 2013). Design factors include shape (Furze, 2002), color(s) and whether lights or other obstruction markers are attached (Pohl et al., 2012). Less well considered is the influence of local atmospheric conditions and lighting on the visual impact of a wind farm (with exceptions including Bishop and Miller, 2007, Sullivan et al., 2013). This gap applies in both simulation and analysis. In the United Kingdom guidelines for both processes have been published (Landscape Institute, 2018, Scottish Natural Heritage, 2017) while the New Zealand Institute of Landscape Architects (2010) have produced a best practice guide for visual simulation. However, these tend to give little recognition to specific effects of atmospheric and lighting conditions, and even less attention to procedures for incorporating these influences. The New Zealand document by-passes the issue, stating: “While variations in light and atmospheric conditions can influence the appearance and visibility of elements within the images, the simulation technique does provide an accurate representation of location, scale and general appearance, even though there may be variations in light and atmospheric conditions at various times of the day, differing seasons and under varying weather conditions.” (p13).

Consequently, when wind turbines are simulated in the landscape, they most commonly appear as if bathed in bright sunlight and with the rotor blades at right angles to the line of sight – in other words, at their maximum color difference and maximum visual magnitude: the worst-case scenario. In reality, this situation seldom occurs. Even in areas area with a great deal of sunshine, the angle of the sun is continually changing creating partial shadow on a turbine. At dawn and dusk the color of the background sky may be quite different from at noon. Also, many wind farms are in coastal locations (either onshore or offshore) and weather conditions are changeable. Even if the turbines are bathed in sunshine, at least 50% of the time the plane of the blades will be at 45° or more to the line-of-sight.

Choosing the right language for this discussion is itself problematic. Terms like ‘worst-case’ imply that the effect is a bad one even though in some circumstances the effect of wind turbines on the landscape is seen, by some, as positive. However, an alternative mode of expression (e.g. ‘conditions of most influence’) are not as easily understood and so worst-case is used here to denote the condition of both greatest contrast and largest visual magnitude (for a given distance) without any a priori judgement of whether the visual affect is positive or negative.

Simulation of the worst-case conditions can be justified on the basis that it provides two end points to consider (best-case and worst-case) and it can then be pointed out that the reality will typically be somewhere in between. It is easier for stakeholders to understand that the situation may be somewhere between zero impact and the worst-case level, than it is to accept a simulation or analysis that might be ‘typical’, but could be worse at times. If a ‘typical-case’ is simulated (however that is defined) then people would have to be asked to consider that sometimes the outcome is more extreme, without providing them with a boundary condition. This could lead to viewers imagining situations worse that is ever really the case.

This is very likely the rationale applied by organization such as The Scottish Natural Heritage (2017) who recommend that simulations be created such that: “a representation of the proposal that is accurate enough for the potential impacts to be fully understood” (p22). In their guideline document they also insist that: “turbines are rendered with sufficient contrast against the backdrop (whether this is the sky or the landform)” (p24). The SNH is also concerned that: “the methodology …. ameliorate[s] the lack of contrast and depth in printed images to ensure that they provide the best representation of the wind farm proposal.” (p22)

While it is not explicit the SNH guidelines are making it clear that high levels of contrast are expected from the simulators regardless of the frequency with which such conditions occur in the real world. Fig. 1 is taken from the cover of the SNH report. It shows the typical orientation and lighting conditions used in, and recommended for, many simulations. In this image an existing, much smaller, turbine is present in the lower right. The difference in appearance of this turbine is striking because both its contrast and its orientation are quite different from the simulated turbines. This paper looks at whether this ‘full-frontal’ approach is really appropriate.

The key factors changing the appearance of a wind farm included numbers of turbines and their distance from the viewer. However, those are not the concern here because these are considered in all simulation and assessment studies. The focus here is on those aspects of appearance (and hence impact) which are not generally well considered. The significance of visibility and weather conditions has already been well documented, but the patterns of temporal change in visibility and, especially, mechanisms for their assessment are less well explored. The other key aspect is the orientation of the turbine blades relative to line-of-sight. This can change the perceived visual magnitude of the turbines considerably and should be considered in both development of simulations and impact analysis.

Lavallee (2012) documents a very thorough analysis of data from weather stations along the North Carolina shore as part of a process of simulating the appearance of large offshore wind facilities from local beaches. Hourly data over a ten-year period was analyzed. From these data estimates were made of:

  • The average number of days that there is visibility to 10 nm (18.5 km), 15 nm (27.8 km), and 20 nm (37.0 km).

  • The average number of days that are sunny.

  • The average percent of each day that is sunny, cloudy and foggy.

  • The average number of days that are foggy/cloudy for at least 50 percent of the day.

Each of these was computed as an annual figure. Visibility and sunniness were also computed by season ((Dec 22-Mar 21; Mar 22-June21; June 22- Sept21; Sept 22-Dec21). The visibility metrics were however uncertain because the weather stations (airport based) that assessed visibility only did so to a distance of 16 km. Estimates for longer distance were extrapolations of the rates of change in visibility in the time period that preceded reaching 16 km (clearing visibility) or in the time period after visibility fell below 16 km.

The report’s conclusions took this form: “Summer days have the lowest visibility and winter nights have the highest. During the day there is visibility to 10 nm at least 50% of the day 34.8% of the time, or 127 days per year. This drops to 27.3% of the days in the summer. In general, the sky is clear 67.8% of the time and cloudy the remaining 32.2% of the time during the daytime hours. It is rarely foggy.” (p7)

In parallel with this analysis, 234 offshore wind turbine simulations were created (Bureau of Ocean Energy Management, 2012). Each simulation consists of an array configuration of 200 turbines with 1000-meter turbine spacing. The simulations included:

  • 18 different locations;

  • Four lighting conditions (morning, afternoon, starlit night and misty nights);

  • Three distances (10, 15, and 20 nautical miles (18.5, 27.8 and 37.0 km from shore); and

  • Two turbine models (Siemens 3.6 MW and Vestas 7 MW).

Visibility distance is certainly one influence on the level of contrast of a wind turbine, but also influential are factors such as the degree, direction and elevation of sun light, the background cloud cover, and haze which can reduce contrast even at distances within the range of visibility. These factors do not operate independently and so contrast cannot be assessed by analysis of visibility distance alone.

Based on prior analysis of the perceptual size of a wind turbine (Bishop, 2002), and combining this with logistic regression models of visibility (Shang & Bishop, 2000), 130 m turbines (such as those in this analysis) would, in clear air, not even be detected by half the people when beyond 45 km distant and would have visual impact for less than half the people beyond 22 km. This applies in circumstances of an elevated view point from which earth curvature does not affect visibility. The result is different for offshore turbines. If an observer is 4 m above sea level, such turbines (total height 130 m) have disappeared at distances over 42 km. Sullivan et al. (2013) reported visibility out to 44 km for offshore turbines but that was at a somewhat higher elevation.

The perceptual magnitude of a whole wind farm is a separate issue and depends on the way individual impacts accumulate. This depends not only on numbers of turbines but also their design, consistency and layout relative to each other and to the landscape.

No studies were found specifically looking at the effect of blade orientation on visual impact. Some insight into the significant of this factor can be deduced from Bishop (2002). That work found that the perceived visual magnitude of a turbine at 90° to the line of sight can be estimated “by taking the actual area of the stationary blades and adding between 10% and 20%” (p710). So, when the blades are parallel to the line-of-sight not only is the direct visual magnitude reduced (since the blades are now seen side on and partially marking each other, but this additional 10–20% component arising from their movement is also likely to reduce substantially).

Clearly, a range of procedural questions must be addressed before any quantification of the strength of wind farm visual effect can be reliably estimated. These same questions also influence the ways in which simulations are designed. The objectives here therefore are:

  • Identify a mechanism to determine blade orientation relative to line-of-sight

  • Test the distribution of orientations within a wind farm and suggest guidelines for simulation and assessment based on these findings

  • Identify a mechanism for determination of color difference between turbines and their background under a range of lighting and visibility conditions

  • Test the distribution of visual contrasts across time and suggest guidelines for simulation and assessment based on these findings.

A digital SLR (Panasonic DMC-G3) with Lumix 14–42 mm lens set to 42 mm (end of range) was mounted on a tripod with a clear view of the Bald Hills wind farm in southern Victoria, Australia (Fig. 2). For this exercise the focal length of the lens has no effect on outcomes. Bald Hills consists of 52 turbines (Senvion MM92 WECs), each with a capacity of 2.05 megawatts (MW) giving a total installed capacity of 106.6 MW. Hub heights are up to 85 m and rotor diameters up to 92.5 m. The towers have a diameter of 4.3 m at the base and 3.0 m at the top. The turbines are in three groups. Within the groups the turbines are typically 500–600 m apart. The farm was fully commissioned in 2015. The minimum distance of the turbines from the camera locations was 6.0 km.

A YouPro YP-870 remote control unit was used to trigger the camera so as to record images at regular intervals. Two images, two seconds apart, were taken every 15 min. Two images were used at each time interval to determine if the rotor blades were moving at that time. The resolution of the images was 4952 by 3448 pixels. The camera ISO was set to 1600 and settings were such that the speed was set to a constant 1/500th of a second and the aperture allowed to vary automatically according to the lighting conditions. As light changes the human eye also adjusts its aperture and so this was considered to best approximate the viewing characteristics of the human eye. It is contrary to the advice normally given for time lapse photography (which is to keep the aperture constant and allow the speed to adjust such that the depth of field is not changing). However, in this situation, there was intention to play the images back as a movie. The independent images simply needed to approximate the viewer experience.

Some additional images were taken from a view point to the north of the wind farm in order to also have examples with the turbines in full sun.

Section snippets

Assessment of orientation changes

When the rotor blades are at 90 degrees to the line-of-sight they have the greatest visual magnitude. At 0 degrees, the viewer sees the blades in a single, relatively narrow, line (not quite a straight line because of the curved design of the blades). The raw visual magnitude of the rotor can be determined either by counting pixels, or by estimating its angle relative to line-of-sight and interpolating. To determine the relative angle from an image one needs to (a) measure the width (w) to

Aspects of color difference

Color difference is often determined in an industrial context from the LAB expression of colors (Robertson, 1977) and was used in the landscape context by Bishop (1997). However, as wind turbines are typically painted white, there is no innate difference in hue to influence the overall color difference and consideration can be limited to variation in (a) lightness, and (b) saturation. A turbine may appear somewhat colored in morning or evening light, but this analysis has been simplified by

Cloudy afternoon

Fig. 7 shows images of one turbine as it changed across a somewhat hazy partially cloudy, and not particularly windy, afternoon in winter. In none of the images were sunlit/shaded parts of the turbines visually distinct. There were just four turbines visible within the field of view, they were all moving, and their variation in orientation was up to 29° (see Table 2). The change from predominantly grey tones, to a strong red component, when nearing sunset, is quite clear. As Table 3 shows, when

Discussion

The unexpectedly high degree of differences in the orientation of the turbines within the Bald Hills windfarm, including adjacent turbines with common differences in orientation of 40 degrees, is a key point of consideration for simulators. While lower figures were found in the test applications in Section 4, these measured the angles of just four turbines and in the morning example the winds were stronger than when the other test images were taken. It is therefore not clear if 40 degrees

Conclusion

As result of a careful monitoring of a particular wind farm, and development of measures reflecting temporal changes in visual magnitude and color difference (and hence visual contrast), it is reasonable to conclude, for both onshore and offshore windfarms, that:

  • Simulations should better reflect natural variability in blade angles, such that even when deliberately presenting the worst-case representation, the rotor blades should be shown as diverging by 15° or more on either side of the

References (33)

  • H.-D. Shang et al.

    Visual thresholds for detection, recognition and visual impact in landscape settings

    Journal of Environmental Psychology

    (2000)
  • T. Tsoutsos et al.

    Visual impact evaluation of a wind park in a Greek island

    Applied Energy

    (2009)
  • Anon (2014). What makes a good print? PhotoReview (June-August 2014) online at...
  • I.D. Bishop

    Determination of thresholds of visual impact: The case of wind turbines

    Environment and Planning B: Planning and Design

    (2002)
  • Bureau of Ocean Energy Management

    Offshore North Carolina Visualization Study

  • F. Cozman et al.

    Depth from scattering

  • Cited by (15)

    • Deconstructing viewshed analysis makes it possible to construct a useful visual impact map for wind projects

      2022, Landscape and Urban Planning
      Citation Excerpt :

      The validity of the results presented appear to be strong. However, it is acknowledged that other factors could be considered, such as atmospheric visibility (Bishop 2019b) or the scenic quality of the surrounding landscape (Betakova et al. 2015; Lothian, 2008). This study follows Valiant’s (2013) admonition: “It is sufficient to aim to capture the fundamental parts of a specific reproducible phenomenon.”

    • Modelling potential visibility of wind turbines: A geospatial approach for planning and impact mitigation

      2021, Renewable and Sustainable Energy Reviews
      Citation Excerpt :

      Wrozynski, Sojka [24], reported that an effective distance for the visibility of a 150 m high turbine is 12 km in case of Poland. Visibility and noise annoyance effects were also investigated by experimental research including computer simulations and perception studies [25–31]. Wind turbine arrays have also impact on visibility.

    • Analysis and visualization of temporal variation in visual impacts

      2021, Landscape and Urban Planning
      Citation Excerpt :

      The study reported here looked more closely at how this can be done, i.e. how ‘typical’ may be defined and what impact mapping or imagery should be presented to stakeholders in addition to the worst-case. Much of the pertinent literature was discussed in (Bishop, 2019) and little additional work on this topic since that time could be identified for this paper. Although, Hallan and González (2020) analyzed changes over time in the character and visibility of wind farms in Ireland, their analysis involved year-by-year changes, as new infrastructure was developed, then assumed a uniform level of theoretical visibility (out to 15 km) and simply added together the number of turbines visible within this range.

    View all citing articles on Scopus
    View full text