Interactive PV database#

For batteries, open the worldwide BESS projects viewer.

Explore the published Parquet snapshot: all records, metadata, evidence and dates. The browser loads one table at a time using a JSON projection generated directly from the downloadable Parquet file; no external service or API key is required.

Loading database catalogue…

Reproduce and extend#

The projects dashboard shows the sum of reported capacity_ac_mw for all matching rows, independently of the selected numeric field and table page. DC capacity is summed separately; neither DC nor unspecified capacity is used to fill missing AC. All-unknown capacity displays Unknown, while a reported zero remains zero. These are catalogue totals, not deduplicated operating fleet estimates: use the status filter and inspect project phases before analysis.

The map uses only finite latitude/longitude pairs within geographic bounds; missing or invalid locations remain in the table and capacity totals. Markers group nearby projects at the current zoom, and each group provides paginated access to every project, including projects at identical coordinates. Search and filters update the map and KPIs together; map panning and table pagination do not change the totals. Equipment tables do not show a project map.

Map navigation uses bundled Leaflet 1.9.4 and the repository’s existing public-domain Natural Earth 5.1.1 country outlines (1:110m). No external tiles, geocoding, CDN requests or API keys are required. Boundaries are an overview; they do not establish site-level accuracy. The original source’s location_accuracy remains visible in each project’s details.

The Database gallery contains runnable examples for database inventory, field completeness, country coverage, separate AC/DC capacity statistics and equipment queries. The database guide explains provenance and licensing. Dataset source licenses apply to exports, including CSVs.

The viewer is rebuilt on every documentation build, including future database updates. It contains every row; pagination limits display only. Browser files carry the exact source Parquet checksum and are verified before display. Null fields display as Unknown; false and zero are retained. Each record exposes source links, field-level evidence, UTC creation/update/verification dates and nested extension metadata. CSV exports include the selected columns and all matching rows, with empty cells for nulls and formula-like strings escaped for spreadsheet safety. Record JSON and Parquet retain the original values.

The browser does not parse arbitrary uploaded Parquet files or run SQL. For full Python analysis, download the Parquet files or rebuild them locally:

uv sync --locked --all-packages --group docs
uv run python -m energy_data_retrieval.pv_catalogue build --output artifacts/pv-analysis
uv run python examples/database/plot_database_statistics.py
uv run python examples/database/plot_project_coverage.py
from pathlib import Path
from energy_analysis.pv_database import field_profile, grouped_statistics

snapshot = Path("artifacts/pv-analysis")
print(field_profile(snapshot / "modules.parquet"))
print(grouped_statistics(snapshot / "projects.parquet", "country_code", "capacity_dc_mw"))

For local viewing, build the documentation and serve it over HTTP; browsers restrict local file:// fetches:

uv run sphinx-build -b html docs docs/_build/html
uv run python -m http.server 8000 --directory docs/_build/html

Open http://localhost:8000/database-viewer.html. The entire generated site is self-contained for this viewer; no CDN scripts, remote queries or uploads are used.