Image: Tom Brewster, Wikimedia Commons
An academic article published in Communication Earth & Environment has introduced a systematic methodology that uses deep learning and high-resolution aerial imagery quantify the land required for utility-scale solar projects in the US Western Interconnection.
The study aims to resolve previous inconsistencies in land use estimates by analyzing 719 solar installations and providing crucial data for future energy system planning and life cycle assessments.
The core of the research involved training a neural network to delineate the entire project footprint – including access roads, fencing and buffer zones – rather than just the immediate panel area, a limitation in previous analyses. The resulting dataset covers more than 13 GW of installed capacity and reveals vital statistics on land efficiency.
The average capacity-based land use efficiency across the sample was quantified at 24.7 W per square meter, with significant variability between projects, the study found.
The study found that dual-axis tracking systems surprisingly exhibited lower project-level land use efficiency compared to fixed-rack or single-axis systems. The article attributes this to the need for greater spacing between tracking panels to prevent shading and allow full movement, resulting in a larger overall footprint.
The study also classified the land cover used by these projects. Although 65% of total installed capacity is now on ‘developed’ land, historical data indicates that 38% of these locations were previously agricultural land.
These findings highlight the potential for ‘land-saving’ strategies. Developers can limit environmental impacts by prioritizing installation on already disturbed surfaces, such as existing roofs or brownfields.
The researchers said the prevalence of projects on former agricultural land also highlights the potential for expanding agrivoltaic energy, the dual use of land for both solar energy generation and agricultural activities, to optimize resource use. Find the full study and methodology here.
This content is copyrighted and may not be reused. If you would like to collaborate with us and reuse some of our content, please contact: editors@pv-magazine.com.
Popular content
