Discover the data.
Bring your own workflow.

Public metadata API

Use GET /api/v1/datasets to list scene descriptions and retained media inventories. Responses include a total count, page number and next-page URL.

GET /api/v1/datasets?country=united-states&media=point-clouds&limit=50&page=1
GET /api/v1/datasets/{datasetId}
GET /api/v1/stats

Supported filters: q, country, region, city, category, media, aligned, minPhotos, from, to and sort. Place and category filters use the slugs visible in catalog URLs. Media values include photos, orthomosaics, dsms, dtms, point-clouds, 3d-meshes and camera-data.

Dates use ISO 8601. limit is 1–100. Sort options are newest, photos and captured. Respect HTTP 429 and Retry-After; avoid parallel crawling. Catalog changes can shift offset-based pages. Store records by dataset ID and version, deduplicate during crawling, and periodically reconcile the complete catalog.

STAC and DCAT discovery

The STAC catalog exposes STAC 1.1.0 Catalog, Collection and Item documents. Only scenes with a known capture date appear in STAC. Follow catalog child/item links, or use the paginated item collection. This is not advertised as a complete STAC API implementation.

GET /stac
GET /stac/collections/scenes
GET /stac/collections/scenes/items?page=1&limit=50
GET /stac/collections/scenes/items/{datasetId}
GET /catalog/dcat.json?page=1

STAC assets link to public metadata and reviewed thumbnails, not private binaries. Point geometry indicates a scene location; it is not a footprint. Place-only listings have null geometry. The paginated DCAT JSON-LD feed exposes the broader public catalog. Dataset landing pages also carry Schema.org Dataset metadata and canonical URLs.

Licensed customer access

After access is assigned, sign in and create a scoped token under account security. Keep the token out of URLs and browser-side code.

Authorization: Bearer ad_YOUR_TOKEN

GET  /api/access/v1/datasets
GET  /api/access/v1/datasets/{datasetId}/manifest?page=1
POST /api/access/v1/assets/{assetId}/link

A manifest contains up to 500 file records, safe relative delivery paths, byte sizes, available checksums and a next-page URL. Obtain a short-lived download URL for each file immediately before downloading it. Do not cache or publish signed URLs. Tokens only access your existing active grants; they cannot administer users or create access rights.

The delivery manifest is pinned to a dataset version. Recheck that every manifest page has the same version. Save the manifest alongside the files. A sample Python bulk-download client is available as part of the deployment package; your contact can provide it with delivery instructions.

Data semantics and limits

Photograph counts are counts of archived files. File totals include overlaps between bundles and individually retained products. Capture observations such as GPS or gimbal angles are not the same thing as calibrated, reconstructed camera poses. Consult the camera files, image map and capture-observation metadata for each scene.

Generated JSON files have SHA-256 checksums where indicated. S3 ETags are not presented as SHA-256 or guaranteed content hashes. No public metadata field implies independent survey validation, precise coverage, suitability for a particular task, or a blanket right to train a model.

Public feeds omit private storage keys, contributor identity, consent evidence and access grants. Geographic coordinates follow the same reviewed disclosure level as the website.