GeoParquet

• Standard Geospatial Data in Parquet

  Following GeoParquet's structure enables interoperability between any system that reads or writes spatial data in Parquet

• Columnar Data for Geo

  Data science workflows benefit from columnar data formats, and geospatial analysis can tap into its innovations

• Cloud Data Warehouse Interoperability

  Snowflake, BigQuery, RedShift, DataBricks can all work together seamlessly with the same geospatial data format

Tools

• Browser-based converter: powered by the GPQ library, you can convert GeoJSON to GeoParquet and vice-versa, from within your browser.
• GeoPandas (Python) extends the datatypes used by pandas to allow spatial operations on geometric types and supports reading and writing GeoParquet.
• QGIS Windows and Linux ship with GeoParquet support, and Mac can work installing with conda (from the terminal with conda activated run 'conda config --add channels conda-forge', 'conda install qgis libgdal-arrow-parquet', and then just type 'qgis' in the terminal).
• Scribble Maps is a full-featured web app that supports both import & export of GeoParquet.
• BigQuery Converter provides Python scripts to read and write GeoParquet files with Google BigQuery.
• CARTO is a geospatial platform and supports import of GeoParquet.
• gpq provides a command-line interface to validate and describe any GeoParquet file. It can also convert GeoParquet to and from GeoJSON
• stac-geoparquet converts STAC catalogs into GeoParquet.
• Apache Sedona is a cluster computing system for processing large-scale spatial data that extends existing cluster computing systems like Apache Spark and Apache Flink. It can load and save GeoParquet with Scala, Java, Python or R.
• Esri's ArcGIS GeoAnalytics Engine 'delivers spatial analysis to your big data by extending Apache Spark with ready-to-use SQL functions and analysis tools'. It can load or save GeoParquet with the Python library or the Spark plugin, see their GeoParquet page for more details.
• FME: by Safe Software is a no code platform that effortlessly integrates your data, including read and write support for GeoParquet starting in version 23.1
• SeerAI's Geodesic Platform is a cloud-native, planetary scale Spatiotemporal Data Mesh and Data Fusion platform. Geodesic's Boson Service Mesh supports GeoParquet natively and can expose massive GeoParquet datasets as compatible formats to other analytical systems and geospatial software via APIs. All tabular and feature data outputs are written in Parquet/GeoParquet format.
• Wherobots provides a fully-managed cloud spatial data lakehouse that can manage and analyze geospatial data at any scale. All data on Wherobots can be saved in GeoParquet format and cataloged by its Havasu Spatial Table Format.
• pygeoapi is a Python server implementation of the OGC API suite of standards. It now supports a Parquet provider that allows publishing a GeoParquet file as an OGC API - Features collection.

Libraries

• geoarrow (R)
• sfarrow (R)
• GDAL/OGR (C++, bindings in several languages)
• GeoParquet.jl (Julia)
• gpq (Go and WASM)
• Fiona (Python - as of version 1.9.4. Note the GeoParquet driver will only be available if your system's GDAL library links libarrow; fiona wheels on PyPI do not include libarrow as it is rather large.)
• .NET 6 library (.NET)
• C++ example code - see this discussion topic for more info.
• GeoParquet.js (JavaScript)

• Microsoft provides access to the Planetary Computer STAC items as GeoParquet, see this quickstart guide for more information. Their Building Footprints are also distributed as GeoParquet.
• There is also a sample dataset nz-building-outlines.parquet that has been used in early testing, converted from GeoJSON downloaded from the LINZ Data Service.
• source.coop provides two datasets in cloud-native geospatial formats, including GeoParquet. The Google Open Buildings cloud-native distribution has over 800 million building footprints across Africa and SE Asia. And the Eurocrops cloud-native distribution provides over 20 million harmonized field boundaries across 16 different European countries.
• emotional.byteroad.net provides most of its +100 datasets in GeoParquet. The GeoParquet files all linked through the metadata records.