Skip to main content

Open Standards for Nature & Biodiversity Data

Overview​

Working with UK nature data means stitching together dozens of datasets from different publishers — Natural England, the Environment Agency, Historic England, local authorities, citizen-science networks. Anyone who has done it hits the same wall: getting the files is easy; making them agree with each other is hard. That gap is a story about open standards — which ones exist, what they actually solve, and the specific thing they don't.

The key distinction this page draws, because almost every frustration lives on the wrong side of it:

  • Transport & access standards — how you get the bytes. Largely solved.
  • Semantic standards — what the bytes mean, and how much to trust them. Largely unsolved for habitat and condition data.

An open standard is a specification that is publicly documented, freely usable, and vendor-neutral — so anyone can implement it and the outputs interoperate. The nature- data world has excellent open standards for moving data and weak ones for agreeing what it means.

Why it matters for nature strategy​

BNG is fundamentally an integration problem. A baseline pulls habitat identity from one dataset, area from another, condition from a third, constraints from several more — and the statutory metric demands they line up. Every place two datasets disagree is manual, judgement-heavy reconciliation:

  • Classification mismatch — one dataset's "improved grassland" is another's something-else; schemes don't map 1:1 (see Reading a Habitat Classification System).
  • Condition is undefined across sources — there is no shared vocabulary for what "moderate condition" means dataset-to-dataset.
  • Provenance and confidence are invisible — nothing standard tells you, in a machine-readable way, that one polygon was surveyed and another was machine-guessed (see Reliability, Confidence, and What a Dataset Doesn't Say).

Solve the first problem — moving files — and you have done the easy 20%. The other 80% is semantics, and that is where an open standard would actually pay off.

The standards that already exist​

Transport & access — the solved layer​

This is the mature part of the landscape, and it is genuinely good:

  • OGC standards (Open Geospatial Consortium) — the vendor-neutral backbone: WFS/ WMS web services, GeoPackage and GeoJSON as open file formats, and the modern OGC API – Features for querying features over the web. These are open, widely implemented, and increasingly the "good practice" baseline.
  • Esri / ArcGIS — see the dedicated note below; the ArcGIS REST API and ArcGIS Hub / Open Data portals are how a large share of this data is actually published (many datasets in the WildStack stack come from *.opendata.arcgis.com).
  • Licensing — the Open Government Licence (OGL) is the de-facto open standard for permission, and it is why most of this data is usable at all.
  • Metadata & discovery — UK GEMINI (the UK profile of ISO 19115), carried over from INSPIRE, plus DCAT for data catalogues, standardise how datasets are described and found on data.gov.uk.

Semantic & content — the weak layer​

Here the picture is patchy, and this is where the real friction lives:

  • Species records: solved, and it's the proof it can be done. Darwin Core (from the TDWG / Biodiversity Information Standards community) is a mature open standard for occurrence records. It is precisely why the NBN Atlas and GBIF interoperate — a shared vocabulary for "this species, here, on this date." Species data has its open semantic standard.
  • Habitats: emerging. UKHab is becoming the shared habitat-classification vocabulary, and the statutory metric's habitat list + distinctiveness bands act as a de-facto controlled vocabulary for scoring. But adoption is uneven and translation between schemes still loses information.
  • Condition: essentially none. There is no widely-adopted open standard for encoding habitat condition consistently across datasets.
  • Provenance & confidence: the biggest gap. There is no cross-dataset standard for machine-readable "how was this derived, and how reliable is it?" Living England's per-feature reliability field is a good single-dataset answer — but it is bespoke, not a shared standard other publishers follow.

The most relevant live effort​

The closest thing to a purpose-built open standard for this kind of data is the UK government's Planning Data Platform (planning.data.gov.uk), run by MHCLG. It is actively standardising planning datasets from every local authority — including conservation areas, tree preservation orders, and article 4 directions — into consistent, openly-licensed schemas. It is the direct fix for exactly the fragmentation described on the Conservation Areas page, where each LPA's data currently arrives differently (or not at all). Its limit for BNG: it is planning-data-first, not ecology-first — so habitat classification, condition, and confidence still sit outside its remit.

A note on ArcGIS — why it helps but doesn't finish the job​

Your instinct is right: ArcGIS does provide something like an open data-sharing standard — and it genuinely didn't solve the whole problem. Both halves of that are true, and the reason is the transport-vs-semantics split:

  • What it solves. Esri published the GeoServices REST Specification openly, and ArcGIS Hub / Open Data gives public bodies a turnkey way to publish downloads, live services, and metadata. For much of the UK nature-data estate, ArcGIS Hub is the distribution layer. It solves access — discovery, download, an API — very well. (The vendor-neutral equivalent is OGC API – Features; Esri also supports INSPIRE and DCAT-style outputs, which shows it implementing open standards rather than being one.)
  • What it doesn't. ArcGIS Hub is a platform, not a semantic standard. It will hand you a habitat polygon flawlessly and tell you nothing standard about whether it was surveyed or modelled, what its "condition" field means, or how to join it to a polygon from a different publisher's portal. It moves meaning around; it does not harmonise it. That is why the challenges that remained after ArcGIS are the semantic ones — because that was never the layer it operated on.

Nation differences​

The standards here are mostly UK-wide or international. INSPIRE / UK GEMINI applied across the UK and was retained after Brexit; OGC, Darwin Core, and GeoPackage are international. Where nations diverge is at the portal and habitat-scheme level — each nation runs its own data portals and may lean on different habitat classifications — rather than in the transport standards themselves. The planning.data.gov.uk effort is England-specific.

  • NBN Atlas — the worked example of a solved semantic standard (Darwin Core).
  • Living England — its reliability field is the closest thing to a provenance/confidence standard, and shows what a cross-dataset one could look like.
  • Priority Habitat Inventory — the survey-grade counterpart whose confidence is implicit, for want of a standard to make it explicit.
  • Conservation Areas — the fragmentation that planning.data.gov.uk is actively trying to standardise away.

See also Coordinate Reference Systems for Nature Data: CRS is the one semantic dimension that is effectively standardised (EPSG:27700), and it shows how much smoother life is when a standard genuinely lands.

WildStack's take​

WildStack's take

The bottleneck in BNG data was never getting the files — OGC, ArcGIS Hub, and the Open Government Licence solved distribution years ago. The bottleneck is that every dataset speaks its own private dialect of meaning and confidence, so harmonising them is slow, manual, expert work. That work — deciding which source wins for a given parcel, carrying "surveyed vs modelled" through the calculation, reconciling classifications — is exactly what WildStack's confidence hierarchy and cross-dataset attribution do by hand, dataset by dataset. It is, in effect, an unstandardised semantic layer rebuilt privately by every serious practitioner.

The instructive precedent is Darwin Core: species records interoperate across the whole world because one open semantic standard was agreed. Nothing equivalent exists for habitat, condition, and provenance. The statutory metric is already a de-facto standard for scoring; what's missing is an open standard for the evidence underneath the score — machine-readable classification, condition, provenance, and confidence that join cleanly across publishers. If that standard existed (and planning.data.gov.uk hints at the shape of it), a huge amount of bespoke reconciliation would collapse into interoperability. Naming that gap precisely — access is solved, meaning is not — is the first step to closing it, and it's why we treat provenance and confidence as first-class data, not metadata footnotes.

Official sources​

Last reviewed​

5 July 2026. Revisit as the Planning Data Platform expands its dataset coverage and standards, if a shared open standard for habitat provenance/condition emerges, or if UK geospatial-standards policy (INSPIRE successor arrangements) changes materially.

Last updated on