Summary:
Across Europe, the expansion of wind energy is increasingly shaped by concerns over landscape aesthetics, with visual impact often driving local resistance. A study published in the journal Energy and AI by researchers from ETH Zurich and the Paul Scherrer Institute (PSI) examines how perceived landscape beauty can be integrated into large-scale wind planning. Using a machine learning model trained on more than 200,000 crowdsourced image ratings from Great Britain, the team mapped scenicness across Europe at high spatial resolution.
When applied to wind resource assessments across 29 countries, the results show that excluding highly scenic areas can reduce wind generation potential by more than 60% in some regions. At the continental scale, electricity costs remain relatively stable, with only modest differences between preservation scenarios. At regional level, the effect is more pronounced, particularly in mountainous areas such as the Alps and Norway, where strong wind conditions often coincide with highly valued landscapes.
The study points to the importance of localised planning and indicates that measures such as micro-siting and aligning turbines with existing infrastructure can reduce visual impact while maintaining overall energy production.
— Press Release —
Wind energy and scenic landscapes: Balancing beauty and power through better planning
Researchers at ETH Zurich and the Paul Scherrer Institute PSI have systematically analysed this conflict at the European level. Under the direction of Russell McKenna, Professor of Energy System Analysis at ETH Zurich and Lab Head at PSI, they examined how people perceive landscape beauty and how this aspect can be incorporated into wind energy planning.
“Our aim was to better understand the conflict between energy expansion and social acceptance,” says Ruihong Chen, a doctoral student in McKenna’s group and first author of the study recently published in the journal Energy and AI.
How landscape beauty can be measured
“Beauty is, of course, a highly subjective concept,” Chen adds. In light of this, the researchers trained a machine-learning model using crowdsourced data from Great Britain. The dataset comprises over 200,000 landscape images that users have rated on a scale of 1 to 10.
The model showed the researchers which landscape features are most strongly associated with perceived beauty. Decisive factors include, for example, the type of land use (e.g. glacial and rocky terrains are classified as most beautiful, while agricultural land and settlement areas rank lowest), the natural quality of the landscape, proximity to water bodies and the amount of sunlight.
In the next step, the researchers applied the model to the whole of Europe. “So far, analyses of this kind have only been carried out for individual countries,” says Chen. Now, for the first time, a machine-learning-based map has been produced showing which regions in Europe are considered particularly beautiful and scenic.
Less wind energy – but hardly higher costs
The researchers linked their analysis of landscape quality to a wind energy model. This enabled them to investigate, for the first time, how protecting particularly beautiful landscapes affects wind energy on a large scale.
The result: excluding especially beautiful and scenic landscapes across Europe would substantially reduce the potential for wind energy. However, the cost per unit of electricity generated would remain close to the European average.
The reason is that suitable locations with strong, consistent winds that are easily accessible – for example, thanks to existing infrastructure or proximity to the electricity grid – are often outside areas rated as particularly beautiful. A larger share of electricity production could take place there.
The conflict is evident on the ground
However, this pan-European approach masks conflicts at the regional level, where notable differences exist. Chen explains: “Especially in hotspots like the Alpine region or Norway, excluding beautiful landscapes would significantly reduce wind power potential.”
When good locations are eliminated, generation costs rise substantially because the remaining sites tend to be less efficient. “On a regional level, unfortunately, good wind conditions can coincide with beautiful landscapes,” says Chen. Switzerland and the Alps exemplify this conflict: despite strong wind resources, the potential has been underutilised mainly due to landscape conservation concerns.
Chen adds: “Broad European or national perspectives are insufficient for wind power planning. Our analysis demonstrates that planning must be highly localised to effectively address local conflicts.”
How tensions can be defused
Different methods exist to defuse conflicts, one of which is known as micro-siting. This technique involves carefully positioning each wind turbine, allowing scenic areas to be preserved without automatic exclusion.
“Wind turbines can, for example, be located behind an edge of the terrain or closer to existing infrastructure such as power lines,” says Chen. This significantly reduces the visual impact on the landscape.
An adapted design could also help make wind turbines less conspicuous and better blend into the landscape. “Bundling installations with existing infrastructure is probably the most socially acceptable way,” says Chen.
First attempt subject to limitations
The study is a first attempt at predicting landscape scenic value on a European scale with high spatial resolution and is subject to certain limitations. As the training dataset is biased towards land cover in Great Britain, not all types of European land cover are well represented.
Further improvements could focus on enriching the training data with locations in other European countries. “For example, we could incorporate social media data to make it even more accurate and robust,” says Chen.
The findings can already be applied in other areas; for example, in the planning of alpine solar plants, grid expansion or other infrastructure projects. In any case, the study provides an approach to advancing the energy transition without losing sight of the landscape.
***
Researchers from the Jülich Research Centre in Germany were also involved in this study.
Journal Reference:
Ruihong Chen, Tristan Pelser, Alena Lohrmann, Jann Michael Weinand, Russell McKenna, ‘Data-driven landscape scenicness mapping for continental-scale onshore wind resource assessment’, Energy and AI 24, 100752 (2026). DOI: 10.1016/j.egyai.2026.100752
Article Source:
Press Release/Material by Deborah Kyburz | ETH Zurich
Featured image credit: User_Pascal | Unsplash
