Skip to content

EPT tutorials

From metadata records and datasets to a validated STAC catalog on EDITO — against real EMODnet data.

The goals

  1. Make STAC — your existing ISO19139 records and data files become valid STAC Items, wrapped in Collections and a Catalog, hosted on S3 and discoverable through EDITO.
  2. Make ARCO — the same data converted to cloud-optimized formats (NetCDF → Zarr, GDB/shapefile → GeoParquet), locally or on EDITO, so clients read chunks over HTTP instead of downloading whole files.

Both run through one EPT API — python, UI, or bash.

What is EPT

One API that does the publishing work:

Convert Validate Deliver
ISO19139 → STAC · NetCDF → Zarr · GDB → GeoParquet clean in, valid STAC out — checked against STAC + EDITO rules local paths or S3 — private or public data

An optional frontend ships with it: every capability is a tool generated live from the OpenAPI schema — fill the form, press Run Tool, inspect the JSON response. EPT runs local or in the cloud: as the as the EDITO service (zero install), from docker images, or from a clone — same API, same contract (Install). Under the hood it uses EDITO services (auth, S3 storage, STAC API, publication); what you code against is just EPT.

STAC

Catalog                         ← root, ties it together
└── Collection                  ← shared extents, license
    ├── Item                    ← one dataset, one place, one time
    │   ├── Asset  Zarr         ← the actual data
    │   └── Asset  ISO19139     ← the metadata record
    └── Item

Plain JSON with links — a linked file structure you can host as static files on S3 (crawlable, cacheable, versionable) or serve from a STAC API — typically an OGC API — Features implementation, backed by a database or by those same static files. The EDITO STAC API is one such API, and STAC is what it serves and searches — so your data needs a STAC representation to be found. Your ISO19139 records already hold everything STAC needs — bbox, temporal extent, contacts, constraints. EPT does the mapping: one call per Item, one call per Collection, one call for the Catalog.

ARCO — cloud-optimized data

ARCO (analysis-ready, cloud-optimized) is about how data is stored, not what it contains: cut into chunks a client fetches individually over HTTP, instead of downloading whole files first. Currently, the two ARCO formats EPT produces:

  • Zarr — an N-dimensional array store: the array is split into chunks, each chunk a separate object on S3 with plain JSON metadata describing shape, dtype and compression. Reading one time slice means fetching only that slice's chunks — not the whole NetCDF. xarray.open_zarr reassembles it into the familiar labelled dataset, CF metadata intact. Works for gridded/raster data and time series.
  • GeoParquet — Parquet with a standard geometry column: a columnar table where data is grouped into compressed row groups. A query on two columns reads only those columns, and a spatial or attribute filter can skip whole row groups without reading them. Works for vector tables — points, lines, polygons.

Both live as ordinary objects on your EDITO S3 storage — the same place the STAC Assets link to — so any HTTP client, notebook or GIS can read them without special data infrastructure. Conversion is format-only: no content checks or coordinate transformations; those belong upstream of EPT.

Getting Started

A published catalog is visible to your project, not necessarily public — wider visibility is a request to EDITO support.