Introducing geoparquet-io — new CLI tool for converting, validating & optimizing GeoParquet files

[Louisa Diggs]

Hi all

Sharing a recent blog post that Chris Holmes and Nissim Lebovits published on the Cloud-Native Geo blog, introducing geoparquet-io (or gpio), an opinionated command-line tool for converting, validating, and optimizing GeoParquet files.

gpio uses DuckDB, PyArrow, and obstore under the hood and enforces cloud-native best practices by default – Hilbert curve spatial ordering, bbox columns, ZSTD compression, and smart row group sizes.

Read the full post here: https://cloudnativegeo.org/blog/2026/03/introducing-geoparquet-io

Community update: Since the post went out, there's been active discussion in the community Slack. One highlight: gpio can now extract WFS layers with DuckDB under the hood. Youssef Harby demonstrated pulling 309k building parcels from Salta province in Argentina in just 2 minutes and 14 seconds. Credit to Youssef for his original write-up on querying WFS with DuckDB.

gpio is currently in v1.0-beta. The team is looking for early users to help with stress-testing, bug reports, feature requests, and PRs. You can check out the open GitHub issues to see what's currently planned.

Best,

Louisa Diggs

Radiant Earth

radiant.earth | cloudnativegeo.org | source.coop