Datasets API

These routes cover raster registration, CSW-assisted ingestion, metadata repair, thumbnails, and point-query helpers.

Listing and Catalog Inspection

Method	Path	Purpose	Auth
GET	/datasets	list datasets with pagination and search filters	yes
POST	/datasets/csw-search	query a CSW catalogue with a structured JSON payload and paginated response	no
GET	/datasets/duplicates	find duplicate datasets by hash, identifier, or source linkage	yes
GET	/datasets/hyspex-missing-variants	report HySpex identifiers missing RGB or Z variants	yes

Registry Maintenance

Method	Path	Purpose	Auth
POST	/datasets/rehash	recompute SHA-256 hashes for all registered datasets	yes
POST	/datasets/prune-missing	remove manifest entries whose source files are gone	yes
POST	/datasets/resync	rebuild the manifest from files on disk and optionally refresh CSW metadata	yes
DELETE	/thumbnails	clear cached thumbnail files and memory entries	yes

Dataset Registration

Method	Path	Purpose	Auth
POST	/datasets	upload a GeoTIFF with multipart form data	yes
POST	/datasets/from-string	register a GeoTIFF that already exists in a mounted runtime path	yes
POST	/datasets/nbs-sentinel-quicklook-csw	query the NBS Sentinel CSW by identifier, detect platform, register quicklook GeoTIFFs, and return combined WMS/WFS service metadata	no
POST	/datasets/nbs-sentinel-nc-csw	compatibility route for Sentinel CSW records: for S1 it prefers direct quicklook GeoTIFF registration and automatically falls back to SAFE processing when quicklooks are missing; non-S1 records keep the NC flow	no
POST	/datasets/nbs-sentinel-safe-csw	query the NBS Sentinel CSW by identifier and prepare or submit SAFE-based Sentinel pipelines using mounted local archives: S1 amplitude quicklooks or S2 RGB product combinations	no
POST	/datasets/sentinel-csw	resolve one Sentinel record from CSW and register discovered GeoTIFFs	no
POST	/datasets/hyspex-csw	build or optionally submit an nc-to-geotiff job from a HySpex CSW child record	yes
POST	/datasets/hyspex-csw-parent	synchronously resolve and queue child jobs from one HySpex parent record	yes
POST	/datasets/from-point-cloud-config	create a GeoTIFF from point-cloud config and register it	delegated async route also exists
POST	/datasets/from-nc-config	alias for point-cloud config conversion workflow	delegated async route also exists

Dataset Inspection and Derived Assets

Method	Path	Purpose	Auth
GET	/datasets/{dataset_id}/sample	sample raster values at one lon/lat	no
GET	/datasets/{dataset_id}/extent.geojson	return the dataset footprint as GeoJSON	no
GET	/datasets/{dataset_id}/thumbnail.png	return a generated thumbnail PNG	no
DELETE	/datasets/{dataset_id}	remove the dataset and optionally its terrain assets	yes
POST	/datasets/{dataset_id}/seed-mapcache	queue cache seeding for one dataset	no
POST	/thumbnails/generate	pre-generate and cache thumbnail images for one or more datasets	no

NetCDF and OPeNDAP Point Queries

Method	Path	Purpose	Auth
POST	/datasets/{dataset_id}/query-point	query one point against the dataset-linked NetCDF source	no
POST	/datasets/{dataset_id}/query-points	batch query points against the dataset-linked NetCDF source	no
POST	/datasets/query-point	query one point against an arbitrary NetCDF URL	no
POST	/datasets/query-points	batch query points against an arbitrary NetCDF URL	no

NBS Sentinel route responses include a profile object derived from the detected platform. Profiles can provide per-platform metadata defaults and point-query variable defaults. The built-in S1 family profile currently defaults SAFE fallback preparation and dataset-linked point queries to Amplitude_VH and Amplitude_VV when the caller does not supply variables explicitly.

Structured CSW Search

POST /datasets/csw-search accepts a typed JSON body and builds an AND-combined FES filter from whichever fields are present.

Supported request fields:

• endpoint: optional CSW endpoint override
• keywords: list of keyword strings, matched against dc:subject
• start_datetime: lower bound applied to apiso:TempExtent_begin
• end_datetime: upper bound applied to apiso:TempExtent_end
• bbox: { minx, miny, maxx, maxy }, encoded as a CRS84 bounding box filter
• anytext: wildcard match against csw:AnyText
• limit: response page size, default 25, max 100
• offset: zero-based response offset, default 0

Example request:

{
  "endpoint": "https://nbs.csw.met.no",
  "keywords": ["sentinel", "s2"],
  "start_datetime": "2026-01-01T00:00:00Z",
  "end_datetime": "2026-01-31T23:59:59Z",
  "bbox": {
    "minx": 10.0,
    "miny": 60.0,
    "maxx": 11.0,
    "maxy": 61.0
  },
  "anytext": "ice",
  "limit": 20,
  "offset": 0
}

The response is paginated and includes:

• total: total records reported by the CSW service
• returned: number of records in the current page
• limit and offset: echoed request pagination values
• next_offset and prev_offset: cursors for the next and previous pages
• constraints_applied: number of top-level FES constraints sent to CSW
• items: normalized CSW records including identifier, title, abstract, subjects, bbox, and references

Style Output

Method	Path	Purpose	Auth
GET	/styles/{dataset_id}/{style_id}.sld	return an SLD document for a dataset/style pairing	no

Notes

• Upload and registration routes attempt duplicate detection to avoid re-registering identical GeoTIFFs.
• HySpex child ingestion can be preview-only or submit jobs immediately, depending on parameters.
• POST /datasets/hyspex-csw-parent is useful for synchronous inspection or small batches, but large campaigns are better driven through the async jobs endpoint family.