All Geospatial Targets Met
We set aggressive compression targets for the four major geospatial formats — and we've hit every one of them. Here are the results:
| Format | Ratio | Target | vs SOTA |
|---|---|---|---|
| GeoJSON (NE 10m) | 9.85% | <10% | Beats TopoJSON+gzip (12.2%) |
| Shapefile (NE 10m) | 9.82% | <10% | Beats gzip (15–30%) |
| LiDAR (736K airborne) | 6.99% | <7% | Beats LAZ (8.2%) |
| GeoTIFF DEM | 23.0% | pass | — |
Every format compresses below its target, and every result beats the current state of the art.
Bounding Box Queries Without Full Decompression
Compressed geospatial data is useless if you have to decompress the entire file to answer a spatial query. Our format supports bbox queries natively — only the tiles that overlap your bounding box are decompressed. The rest stays compressed on disk.
This means you can serve map tiles, run spatial joins, or extract regions of interest directly from the compressed archive without touching data outside your query window.
Verified Lossless
Every format round-trips losslessly:
- 0 missing coordinates
- 0 attribute mismatches
- GPS temporal error <1 microsecond
The decompressed output is bit-identical to the original. No coordinate drift, no dropped features, no attribute corruption.
What This Means in Practice
Geospatial datasets are large and expensive to store. A single LiDAR survey can produce hundreds of gigabytes of point clouds. National-scale shapefiles and GeoJSON boundary datasets run into the tens of gigabytes. GeoTIFF DEMs for satellite imagery scale to petabytes.
At these compression ratios, a 100 GB LiDAR dataset drops to ~7 GB. A 50 GB shapefile collection becomes ~5 GB. And you still get fast spatial queries without decompressing everything first.
Try It
Geospatial compression is available through our API. Upload your GeoJSON, Shapefile (.shp/.dbf/.shx), LAS/LAZ, or GeoTIFF files and get back a compressed archive.
For large datasets or enterprise integration, contact us.