Distributed solar photovoltaic array location and extent dataset for remote sensing object identification

Area of Science:

• Earth Science
• Renewable Energy
• Computer Science

Background:

• Earth-observing remote sensing data provides insights into natural resources and the built environment.
• Machine learning methods can automatically assess energy systems from aerial and satellite imagery.
• Limited public data exists on small-scale distributed solar photovoltaic (PV) deployments.

Purpose of the Study:

• To address the information gap in distributed solar PV deployments.
• To create a dataset for automatically identifying solar PV arrays using remote sensing imagery.
• To facilitate machine learning applications for PV system analysis.

Main Methods:

• Utilized high-resolution aerial and satellite imagery from four California cities.
• Developed a process to automatically identify solar PV locations.
• Created a dataset containing geospatial coordinates and border vertices for over 19,000 solar panels.

Main Results:

• A comprehensive dataset of distributed solar PV arrays was successfully generated.
• The dataset includes detailed information on over 19,000 individual solar panels.
• The data covers 601 images across four distinct urban areas.

Conclusions:

• The created dataset is valuable for training machine learning algorithms for PV detection.
• Applications include estimating installed PV capacity and analyzing socioeconomic correlates of PV deployment.
• This work advances the automated assessment of distributed solar energy systems using remote sensing.