GeoParquet to GeoJSON Converter
The reverse of the cloud-native migration: turning a query result from BigQuery, Carto, or Sedona into a portable JSON file you can hand to a web map or partner.
Convert GeoParquet to GeoJSON now
Common issues converting GeoParquet to GeoJSON
- GeoParquet files can carry tens of millions of features. GeoJSON is plain JSON — one giant array — and most parsers OOM past ~500 MB. Filter rows in your warehouse before exporting.
- Parquet INT64 columns survive in GeoJSON as JavaScript Numbers, which lose precision past 2^53. If your IDs are 64-bit integers, cast them to string before exporting.
- We honour the `geo` metadata's primary geometry column. If the file declares multiple geometry columns, only the primary becomes the GeoJSON geometry; the others survive as WKB hex strings on the feature properties.
- CRS is reprojected to WGS 84 because GeoJSON requires it. The reprojection uses the source CRS declared in the `geo` metadata — if that's missing we assume WGS 84 and warn.
Frequently asked questions
Does the converter stream large Parquet files?
The browser converter loads the file fully into memory. For multi-gigabyte Parquet, use the async API which streams via DuckDB-spatial server-side.
What if the GeoParquet file uses a non-WKB encoding?
GeoParquet 1.1 also supports WKT and a native Arrow encoding. We read all three; output is always GeoJSON regardless of source encoding.
Are nested struct columns flattened?
Top-level only. A struct column appears as a JSON object on each feature's properties, preserving its shape.
Can I export only a subset of features?
The browser tool converts everything. For server-side filtered exports ("only features in this bbox") use the workflow editor, which composes a filter step before write.