Any photovoltaic (PV) panels designed for electricity generation, whether roof or ground mounted. Solar hot water systems are explicitly excluded from the definition.
Characteristics and Recommended Use
The Solar Panel AI Layer (and associated information summarised in AI Parcels) is exceptionally high quality. It is often difficult for a human to distinguish between photovoltaic and water heating panels from imagery, and the model is usually correct in these cases. Most panels have very high confidence (>90%), and almost certainly represent photovoltaic panels. The few with lower confidence are typically solar hot water, or panel-like roof structures such as skylights. You can use a confidence cut-off to include or exclude those panels that might represent hot water systems.
Where we state a parcel has solar panels, almost all of the false positives are solar hot water, with just a small portion of regular roof structures and skylights. The missed panels are due primarily to extremely glary reflections, very old style panels (small 0.5-1kW arrays from before 2010), and the rare panels under extreme tree shadow.
Note the panels with glary reflections are successfully detected, but with lower confidence.
Solar Panel Classification Typically Include
PV panels on houses
PV panels on commercial buildings
PV panels mounted on the ground
|Borderless black solar panels||Moderate panel glare|
Solar Panel Classification Typically Exclude
|Traditional domestic solar hot water (twin panels plus tank)|
Most Common Errors
|Some glass roof structures and skylights||Some solar pool hot water systems|
|Legacy low KW solar systems||Extreme panel glare|
Confidence Distribution by Region
Please refer to Confidence for a further explanation of this value and best practices in using this score to filter the data according to your use case.