publicdata.au publicdata.au on GitHub

Road crash locations, South Australia, version 2025-09-19

latest
Rows 43,522Fields 40Fetched 29 September 2026Source SHA-256 b78b3d9e8fb19be2…

This version is kept for as long as this site exists, and its URL can be cited. Its files change only for the reasons the terms give, and each change is recorded in its notes. The publisher's file is here as source.zip, byte for byte.

Files

  • data.arrow3.1 MB Arrow: Arrow IPC file (Feather v2), zstd compressed, with the provenance in the schema metadata. Memory-maps in pyarrow, Polars, Arrow JS and R without parsing.
  • data.csv9.9 MB CSV: UTF-8 without a byte-order mark, RFC 4180 quoting, one header row using the normalised field names. Opens in Excel, Numbers and Sheets.
  • data.csv-metadata.json10 KB
  • data.csv.gz1.3 MB CSV (gzip): The same CSV, gzipped, at about a tenth of the size. curl, pandas, DuckDB and R read it directly.
  • data.duckdb DuckDB: One DuckDB database with a records table of typed columns, a fields table and a publicdata metadata table. It attaches read-only over HTTPS, so DuckDB, Python and R can query it without downloading it.
  • data.geo.parquet2.0 MB GeoParquet: the Parquet table with a WKB geometry column and its coordinate system in the file's geo metadata. Reads directly in DuckDB, GeoPandas, QGIS and GDAL.
  • data.geojson43 MB GeoJSON: A FeatureCollection with one Point per row and every field as a property. Rows without coordinates have a null geometry. Ready for QGIS, Leaflet, MapLibre and geopandas.
  • data.gpkg11 MB GeoPackage: A GeoPackage with one point layer, records, in the publisher's coordinate system, plus the fields and publicdata tables. Drag it into QGIS or ArcGIS.
  • data.json38 MB JSON: One object with a publicdata header (publisher, licence, attribution, version, source hash), a fields list and a records array of 43,522 objects with typed values. Blank cells are null.
  • data.ndjson38 MB NDJSON: The first line is the provenance header, then one JSON object per line. For streaming and for tools that read line by line.
  • data.parquet1.6 MB Parquet: Columnar and typed. The smallest download and the fastest to query. Reads directly in DuckDB, pandas, Polars, Arrow and Spark. The provenance header is in the file metadata.
  • data.sqlite8.4 MB SQLite: One database with a records table, a fields table and a publicdata metadata table. Open it with any SQLite client.
  • data.xlsx6.7 MB Excel: One workbook with a records sheet of 43,522 typed rows, a fields sheet and a publicdata sheet with the provenance. Dates are real Excel dates. Opens in Excel, Numbers, LibreOffice and Sheets.
  • manifest.json2.6 KB
  • schema.json8.6 KB
  • schema.sql4.5 KB
  • source.zip2.7 MB

SHA256SUMS holds the SHA-256 of every file in this version under the name it downloads as. Check what you downloaded with sha256sum -c --ignore-missing SHA256SUMS in the same folder. On a Mac, use shasum -a 256 -c --ignore-missing SHA256SUMS. Each list is signed by the deploy that wrote it: gh attestation verify SHA256SUMS --repo National-Digital/publicdata.au --source-ref refs/heads/main shows it came from this project's main branch.

Attribution the licence asks for

Department for Infrastructure and Transport, Government of South Australia, Road Crash Locations in SA, https://data.sa.gov.au/data/dataset/road-crashes-in-sa, accessed 29 September 2026, licensed under CC BY 4.0.

Provenance

Fetched from data.sa.gov.au on 29 September 2026. The portal listed the licence as Creative Commons Attribution at that moment. manifest.json has every detail.