Global Inventory of Photovoltaic Solar Generating Units from Satellite Imagery

Kruitwagen et al. (2021) provide a global inventory of photovoltaic solar energy generating units. The repository contains training, cross-validation, test, and predicted data sets, including 68,661 predicted polygons capturing global deployment status at the end of 2018. The data was published in Nature and is associated with the University of Oxford.

Use Cases

• Training machine learning models for solar panel detection based on satellite imagery patches.
• Cross-validating solar energy unit detection algorithms using seeded regions from the WRI GPPDB.
• Evaluating model performance on a test set of utility-scale solar generating units.
• Analyzing the global spatial distribution and density of solar photovoltaic capacity as of 2018.

Strengths

• Includes 68,661 predicted polygons for global deployment analysis.
• Training data is based on 36,882 polygons from OpenStreetMap (OSM).
• Test set contains 7,263 polygons specifically for utility-scale (>10kW) units.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.
• Data reflects a specific temporal snapshot (end of 2018) and may not reflect recent installations.

Provenance

Source

University of Oxford (Kruitwagen, L., et al.)

Collection Method

Derived from satellite imagery and OpenStreetMap (OSM) data, with predictions generated by a model described in the associated Nature publication.

Time Range

Data captures status at the end of 2018.

Geography

Global