Formats and integration
A practical guide to cloud-native geospatial formats: COG, STAC, GeoParquet, and GeoJSON
Cloud-native formats let software read just the part of a dataset it needs, straight from object storage. Four of them cover most Earth observation workflows: COG for rasters, STAC for catalogs, and GeoParquet and GeoJSON for vectors.
2 minute read. Written by the GeoVeyra team.
Why cloud-native matters
Satellite scenes are large. Downloading whole files to answer a question about one field is slow and expensive. Cloud-native formats are organized so that a client can request only the bytes it needs, usually with HTTP range requests against object storage.
Cloud Optimized GeoTIFF (COG)
A COG is a regular GeoTIFF with an internal layout designed for partial reading: the image is split into tiles and includes lower-resolution overviews. A map viewer can fetch a zoomed-out preview or a single tile without reading the full file.
Use COG for imagery, analysis rasters such as change heatmaps, elevation models, and any gridded data you want to stream.
STAC: SpatioTemporal Asset Catalog
STAC is an open specification for describing geospatial assets with JSON. An Item describes one scene or product with its footprint, time, and properties, and links to its files. Collections group related Items, and the STAC API standardizes searching by place, time, and properties.
Use STAC to make an archive of imagery and derived products discoverable, for people and for software.
GeoParquet
GeoParquet is a specification for storing vector geometries in Apache Parquet, a compressed, columnar file format widely used in analytics. Columnar storage makes it efficient to read only the attributes you need across millions of features.
Use GeoParquet for large vector datasets such as parcels, building footprints, or historical change polygons that feed analytical workloads.
GeoJSON
GeoJSON, standardized as RFC 7946, encodes points, lines, and polygons with properties in plain JSON, using longitude and latitude coordinates. It is human-readable and supported almost everywhere, from web maps to desktop GIS.
Use GeoJSON for exchanging modest numbers of features, such as the findings for a site, and for web applications and webhooks.
Choosing between them
A typical GeoVeyra-style workflow uses all four together.
- Rasters and heatmaps: Cloud Optimized GeoTIFF
- Finding imagery and products: STAC
- Large vector archives for analysis: GeoParquet
- Sharing findings with people and apps: GeoJSON
Want to see these ideas applied to your area of interest? Book a demo or browse the glossary.