Germany's PV Atlas

Best estimate: 88,709 MWp, with a 90% range of 78,597–101,104 MWp. That range covers two specific, measured sources of uncertainty -- but not everything that could move this number. Below: what's inside the range, what isn't, and how it compares to unrelated data sources.

What's inside the range:

• The two "panel area to power" conversion numbers. One converts rooftop panel area to kWp, the other converts open-ground solar-farm land to kWp. Both are measured against real, confirmed power plants rather than assumed, but each carries its own uncertainty.
• How well the detection model finds panels of different sizes. Its measured precision and recall were checked installation-size by installation-size, and that check itself has a margin of error.

What's outside the range -- and can't be added back in with more arithmetic:

• Getting the national total right is not evidence the geography is right. Cross-validated, this method's German national figure lands within a few percent of the register while the median municipality is out by roughly half, because over-prediction of the densest tenth of municipalities cancels under-prediction of the other ninety percent. Three different models whose municipal errors spanned 38–64% all produced nearly the same national number. What a grid model is consumed for is the geography, not the total.
• In Germany, multiplying roof area by a constant does as well as the classifier. Against a per-municipality roof-area baseline, the classifier's median error (48.4%) is worse than the baseline's (37.8%). Where rooftop PV is close to ubiquitous, capacity is nearly proportional to roof area by construction. The classifier earns its keep where PV is rare -- in Pakistan it cuts the same error from 88.9% to 26.5% -- which is the regime it was built for and Germany is not.

Treat this as an early-stage estimate from an active research pipeline, not a finished census. What's genuinely new here -- a reproducible way to estimate distributed solar from free satellite imagery and open-source AI, in places where official statistics are sparse or absent -- holds regardless of whether any single number on this page turns out exactly right. Expect these figures to keep moving as the evidence behind them grows.

There are no hand-mapped ground-truth areas behind this page, and for the small-panel half there do not need to be. Registration in Germany's Marktstammdatenregister is legally mandatory rather than voluntary, and it publishes totals per municipality -- which is exactly the quantity the small-panel estimator produces. The conversion from flagged roof area to kilowatts was therefore fitted against the register across 9,674 fully-covered municipalities, at 0.2793 kWp per m² of credited roof area, and applied countrywide as a genuine out-of-sample test. What a register cannot substitute for is a precision check on small installations: below 30 kWp it publishes no coordinates at all, so there is no way to ask whether a particular flagged roof really carries panels.

A complete national register puts the Best estimate above too high. The Marktstammdatenregister records roughly 74.8 GWp of PV in Germany, below the figure this page reports, so read these tiers as geography -- where capacity sits -- rather than as a national total. The hand-mapped OpenStreetMap component was reconciled against the register in September 2026 and both placements now sit within about 10% of it; what remains unreconciled is the model's own detections and the small-panel half. Every other country in this project would need a register of this kind to make the same statement, and does not have one.