GeoParquet (Apache Parquet)
Open BetaGeoParquet is a columnar geospatial format based on Apache Parquet. Ideal for big data analytics, data lake applications, and processing in DuckDB, Spark, BigQuery, and other analytical systems.
Key Features
About GeoParquet Format
Parquet is a columnar storage format from Apache: data lies by column, each one compressed separately, and block headers carry statistics, so the engine reads only the columns it needs and skips the rest of the rows. GeoParquet (the OGC standard, version 1.0) adds to that an agreement on how to store geometry and the coordinate system.
It is a format for analytics and long-term storage: a data lake in object storage, a data mart in BigQuery or Snowflake, heavy computation in DuckDB and GeoPandas. On large selections it is both smaller on disk and faster in queries than row-based formats.
The output is a single .parquet: all the layers merged into one table with a _layer column you can split them by. Geometry is stored in binary WKB form, and the WGS84 (EPSG:4326) coordinate system is described in the file metadata per the GeoParquet specification. The default compression is snappy.
You cannot look at such a file with your own eyes, it is binary. But it is read by DuckDB, GeoPandas, Spark, BigQuery, QGIS and ogr2ogr, usually without loading it into a database at all.
What Program Opens GeoParquet Files
The following programs and applications can open, edit, and work with GeoParquet (Apache Parquet) (.parquet) files exported from OSM2CDR:
Format specifications
- Open standard
- 3D coordinates supported
- Lossless (no quality loss)
- Stores feature attributes
- Output: 1 file
- GeoParquet 1.0 (OGC). Columnar Apache Parquet with WKB geometry. Optimised for analytics: DuckDB, BigQuery, GeoPandas.
Who Uses GeoParquet Maps
How to Export GeoParquet
How to Open After Export
- DuckDB — INSTALL spatial; LOAD spatial; SELECT * FROM 'map.parquet' WHERE _layer='buildings'; the query runs straight against the file, no import needed.
- Python — geopandas.read_parquet("map.parquet"), then the usual GeoDataFrame work, spatial joins included.
- QGIS 3.28 and newer — opens GeoParquet if the GDAL build has Parquet support; if the format is not visible, convert it: ogr2ogr -f GPKG out.gpkg map.parquet.
- Cloud storage — the file goes into S3 or GCS and is read from there directly by the DuckDB, Spark and BigQuery engines, with no database to stand up.
- Splitting by layer — save separate files from GeoPandas with a filter on _layer if the consumer only needs one layer.
Popular maps in GeoParquet
Get GeoParquet (Apache Parquet) map exports
Download professional GeoParquet maps from OpenStreetMap. Instant generation, any area.
Frequently Asked Questions
Related Formats
Frequently Exported Together
Related Articles
GeoParquet — Columnar Format for Big-Data GIS: History, OGC Status, DuckDB and OSM Export
GeoParquet deep dive: Apache Parquet + OGC geo extension, the 1.1 release (Sept 2024), bbox covering, querying via DuckD
guideGeoParquet: The Format for Fast Geospatial Analytics
What is GeoParquet, why it outperforms Shapefile for analytics, and how to use it with DuckDB, Python, and QGIS. A compl
formatGIS Data Formats in 2026: From Shapefile to Cloud-Native
GIS format overview: Shapefile, GeoPackage, FlatGeobuf, PMTiles, GeoParquet, COG — evolution and choice in 2026.