TNFD Data and Evidence Quality: Primary, Secondary, Proxy and Geospatial Information

Figure 1. Show the evidence hierarchy.

The same dataset can be suitable for one use and unsuitable for another. A sector proxy can select suppliers for engagement but may not support a site claim. A global protected-area layer can screen a coordinate but cannot establish habitat condition or actual impact. A verified coordinate may be primary location evidence, while the ecological layer overlaid on it is secondary. The team should therefore define the decision and required confidence before grading the source.

A supplier questionnaire is primary but may be self-reported, incomplete, based on a different period or interpreted under inconsistent definitions. A site coordinate may identify the office entrance rather than the operating footprint. A grievance register may exclude cases held by contractors. Direct evidence still needs population controls, cut-off, ownership, reconciliation and review.

Nature reporting relies on external datasets because organisations cannot independently measure species distribution, ecosystem condition and hydrological risk everywhere. The source register should preserve publisher, dataset name, edition, extraction date, spatial and temporal coverage, resolution, licence, transformations and known limitations. A score separated from its methodology is weak evidence.

A proxy is a deliberate substitute: sector ratings for a borrower with unknown activities, city coordinates for a supplier with no site address, average water use per tonne, or estimated commodity origin. The register should explain why the proxy is representative, how uncertainty is treated, what decision it supports and when it will be replaced or reassessed.

Coordinate provenance. GPS survey, cadastral record, counterparty submission, public address or automated geocoder.

Geometry. Point, polygon, concession, catchment, supply area or administrative centroid.

Reference system. Coordinate reference system, transformation and spatial-join method.

Layer date and resolution. Dataset vintage, pixel/polygon resolution and suitability for the pressure pathway.

Buffer rationale. Ecological, hydrological, legal or risk basis; not a universal distance rule.

Confidentiality. Sensitive species, communities, borrower sites and commercial locations may require restricted access and aggregation.

Stable evidence ID linked to a DIRO, metric, target or claim.

Publisher/data owner and source contact where relevant.

Document, dataset or model name, version, release and extraction dates.

Geographic coverage, geometry, resolution and coordinate system.

Population, value-chain stage, time period and boundary.

Definitions, units, transformations and calculation method.

Licence, commercial-use and redistribution conditions.

Assumptions, proxy rationale, uncertainty and limitations.

Confidentiality/access class, retention period and reviewer.

Specialist input is most important where screening could change a material credit, investment, operating or claim decision; where a legal or ecological designation is uncertain; or where the organisation proposes to describe restoration, net gain, no impact or nature-positive outcomes. Validation may involve ecologists, hydrologists, geospatial specialists, engagement specialists, legal counsel, model-risk teams or data owners.

Stabilise the register. Inventory current sources, assign IDs and classify evidence.

Prioritise material gaps. Focus on gaps that could change locations, DIROs, metrics, targets or claims.

Improve through counterparties and targeted validation. Request site/activity data and commission specialist work for high-consequence decisions.

Embed controls. Automate metadata, access, versioning, lineage, review, retention and reassessment while preserving judgement.

A consumer-products group has 4,000 direct suppliers but exact operating coordinates for only 18%. It maps suppliers by commodity, country and spend, then uses sector and country proxies to prioritise 300 suppliers. It requests coordinates and production data from 80 priority suppliers, applies IBAT and Aqueduct overlays to verified locations, and escalates 22 matches for local evidence. The disclosure reports coverage and confidence bands rather than implying that all suppliers received the same site-level assessment.

Calling self-reported data verified. Primary describes provenance, not assurance.

Using one confidence score without dimensions. The nature of the weakness becomes invisible.

Losing dataset versions during spatial joins. The result cannot be reproduced or updated.

Treating an address geocode as a facility footprint. The point may not represent the nature interface.

Publishing exact sensitive locations unnecessarily. Transparency should not create ecological, community or security harm.

Keeping proxies indefinitely. Every material proxy needs an owner and improvement trigger.

Figure 2. Show confidence dimensions and the improvement roadmap.

Prerequisite: TNFD Explained - Recommendations, LEAP, DIROs and how to start.

Apply next: TNFD LEAP Approach - a step-by-step guide.

Evidence: Geospatial Data for TNFD - coordinates, maps, data quality and confidentiality.

Advanced: TNFD metrics, targets and the limits of composite biodiversity scores.