Researchers from the United States and United Kingdom have used machine learning to map every large solar plant in the world. The team behind the map sees it as an opportunity to consider the future trajectory of solar expansion and to inform decisions on what land uses can be best supplanted by solar farms by looking at historic trends. The findings were published in the journal Nature by researchers from the University of Oxford in the UK, Descartes Labs Inc in San Francisco, and the World Resources Institute in Washington DC.

The map was made possible by the construction of a machine learning system which could detect solar facilities in satellite imagery; a system that the researchers scaled to a planet-wide search using 550 terabytes of imagery. To streamline their results, the researchers limited their search to solar facilities which generate at least 10 kilowatts during peak operation: approximately twice as much as a small residential rooftop system. The researchers also refined their search to areas of likely human habitation.

“We searched almost half of Earth’s land surface area, filtering out remote areas far from human populations,” explained research co-author Lucas Kruitwagen in an article on The Conversation. “In total, we detected 68,661 solar facilities. [These] ranged in size from sprawling gigawatt-scale desert installations in Chile, South Africa, India and north-west China, through to commercial and industrial rooftop installations in California and Germany, rural patchwork installations in North Carolina and England, and urban patchwork installations in South Korea and Japan.”

The team used the map to generate an array of insightful findings. Among them was the finding that solar PV generating capacity grew by 81% between 2016 and 2018, with particularly strong growth in India (184%), Turkey (142%), and China (120%). The team was also able to use archived satellite imagery to estimate the installation dates for 30% of the facilities.

Finally, the team was able to conclude that solar power plants are most often constructed in agricultural areas, followed by grasslands and deserts. “This highlights the need to carefully consider the impact that a ten-fold expansion of solar PV generating capacity will have in the coming decades on food systems, biodiversity, and lands used by vulnerable populations,” Kruitwagen explains. “Policymakers can provide incentives to instead install solar generation on rooftops which cause less land-use competition, or other renewable energy options.”

The publication of the research comes only weeks after the U.S. Department of Energy published a view that solar energy could produce 40% of the United States’ electricity by 2035. On a state level, a law in California is set to come into force in 2023 which will mandate solar panels for new buildings, while researchers from UC Santa Cruz and UC Merced have separately floated the idea of building solar panels over California’s canals.