OpenPVMapper: A Multi-source, Nationwide Database of Rooftop Photovoltaic Systems in France
The paper introduces OpenPVMapper, an open, nationwide database of over 1.1 million rooftop photovoltaic installations in mainland France, created by aggregating and reconciling multiple independent data sources to achieve a comprehensive, high-confidence installation-level dataset that surpasses the limitations of existing public registries and single-method remote sensing efforts.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine trying to count every single bird in a massive, bustling forest, but you can't walk through the trees. You can't ask the birds to raise their hands, and the forest rangers only keep a rough tally of the giant eagles, ignoring the tiny sparrows. This is the challenge scientists face when trying to track rooftop solar panels. These panels are the "sparrows" of the energy world: millions of them, popping up on houses and small businesses every day, often without anyone in a central office knowing about them. While big solar farms are easy to spot, these tiny, decentralized power generators are scattered everywhere, making them incredibly hard to map.
To solve this, researchers use two main tricks. First, they use "remote sensing," which is like taking a giant, high-resolution photo of the whole forest from a drone or satellite and using a super-smart computer brain (called deep learning) to spot the shiny blue rectangles of solar panels. Second, they use "crowdsourcing," where volunteers on a global map (like OpenStreetMap) manually draw where they see panels, kind of like birdwatchers logging sightings. The problem is that the computer brain sometimes gets distracted by shadows or blue roofs, and the volunteers can't be everywhere at once. So, the big question is: how do we get a complete, accurate picture of this hidden energy network without missing the small stuff or counting fake panels?
Enter OpenPVMapper, a new project that acts like a super-detective for solar panels across France. Instead of relying on just one detective, the team combined three different sources of information to build the most complete map of rooftop solar ever made for the country. They took the computer's "best guess" from aerial photos, added a second computer method that looks at building records, and sprinkled in the human-verified spots from the global map.
Here is the magic trick: when two or more of these independent detectives agree that a solar panel exists, the team becomes almost certain it's real. If only one detective spots it, they flag it as "maybe." By mixing these sources, they created a database of 1,135,850 individual solar installations, totaling about 15.01 GWp of power. This map covers nearly every rooftop solar system under 36 kWp (the size of a typical home system), a category that official government records often miss.
The team didn't just guess; they tested their work. They manually checked a sample of 1,862 installations and found that the map is about 74–75% accurate overall. But here is the exciting part: when two different sources agreed on a location, the accuracy jumped to 97%, and when three sources agreed, it hit 98%. This proves that combining different methods is far better than trusting just one. The map is now open for everyone to use, helping researchers and planners understand exactly how much solar power is actually being generated across France, down to the individual roof.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.