These are the sources — each country's National Soil Information Systems (data stays with the country) and the hub's own field and laboratory systems. Hub-produced datasets are stored and cited in Dataverse; the wider commons is made findable through GARDIAN. All of it is served through a secured gateway as the catalogue, under a written, versioned agreement with a named owner.
The technical view: every system and federated layer that produces the hub's data products — with its metadata, field schema, lineage, freshness and how to request access. Each card is coloured by its pipeline stage — sources, governed core, processing, delivery. Search, filter, or open any entry for the full panel.
No systems match your search.
The colours above — and on the diagram below — map to where a system sits in the pipeline.
Field, laboratory and environmental systems that generate the raw record — soil samples, trials, weather and map layers. They hand data to the core through a contract; they never write to it directly.
Everything is validated on arrival (bad records are quarantined, not dropped), catalogued with clear ownership and lineage, and kept as one canonical, versioned model — replacing extracts assembled by hand.
Features are prepared with reproducible joins across soil, trial and weather records. Models are versioned, return their uncertainty, and every result is traceable to the exact data that produced it.
Served through a secured gateway: advice for the field, exchange with national systems that keep their own data, evidence for the record, and maps for programmes.
Sources flow into the governed core — validated on arrival, catalogued, kept as one versioned source of truth. Models learn from it and write results back; every product is then served through the discovery & interoperability layer — a secured API gateway built on open standards. Down the right, governance applies at every tier. Nothing is openly exposed: access is by agreement.
Systems integrate through the platform rather than with one another, and every connection is a written, versioned contract with a named owner on both sides. Six rules hold across every stage of the diagram.
Data is shared under agreements, and partner countries keep control and ownership of their own data. Sovereignty is a default, not a favour.
Every dataset and interface has a named owner on both sides. Access is granted deliberately — consumers hold no credentials beyond what their role needs.
Models are reviewed for responsible use before their outputs inform decisions in the field.
Nothing goes out externally without expert agronomic sign-off — the science is checked before it reaches a farmer.
Every result is traceable to the data that produced it. The metadata catalogue records definitions, vocabularies, ownership and lineage — FAIR made demonstrable.
Each connection between systems is a written, versioned contract with named owners — so nothing changes silently underneath anyone.
Every partnership also runs under the data-protection law of the country involved and the region's frameworks (ECOWAS and the African Union), with international best practice — including the GDPR principles of consent, clear purpose and collecting only what's needed — as the baseline. The platform mainly holds soil, trial and environmental data; personal details are kept to a minimum and stay under each partner's control.
A shared platform only helps if people can actually find the data and plug it into their own tools. So everything here is made to be searchable, clearly explained, and ready to connect — using open, international ways of sharing data instead of one-off formats that lock information away.
One place lists every dataset, so you can tell at a glance what the hub holds — no guessing, no hunting.
Each dataset comes with a plain description, who owns it, and where it came from — so you know what you're looking at.
Systems share data in common, open formats, so your tools can read it straight away instead of needing special code.
Because it follows open international standards, your software is never tied to ours — today or in the future.
Built on open international standards — DCAT, OGC API, STAC, OpenAPI, FAO GloSIS (ISO 28258) and the FAIR data principles — so the platform stays compatible with the wider world of data tools.
Every dataset is published to a live, machine-readable DCAT catalogue (data.json) with schema.org metadata — so the hub's data is discoverable in Google Dataset Search and harvestable by CKAN, HDX and CGIAR GARDIAN. This is the federating registry other portals plug into, not another silo.
Discover datasets from the machine-readable catalogue and call the secured API gateway. The quickstart, full endpoints reference, self-service API keys, authentication and data standards all live in the docs.
Want to use a system, connect a national soil system, or set up a bilateral exchange? Tell us what you need and how you'll use it — nothing is openly exposed, so every partnership begins with a short, formal request and a data-sharing agreement.
regionalfsh@cgiar.org