From agricultural data to traceable evidence.

A reliable assessment starts with an identifiable field. The proposed registry connects a stable field identifier and mapped boundary to area, location, the participating farm or organization and appropriate permissions. Crop and activity records belong to a particular season; field identity persists as those seasonal records change.

• Record field identifier, boundary, area and organization reference.
• Associate crop, growing season, applicable practices and consent with the relevant period.
• Flag missing boundaries, overlapping parcels, duplicates and incomplete registration before assessment.

The public field passport is a sample. Account records and exports are available separately to authorized users.

Suitable satellite datasets can add observation dates, boundary overlays, vegetation or land-surface indicators and change observations to a field record. Weather and environmental datasets can supply context where relevant. Each observation needs spatial alignment and a documented processing method.

• Check acquisition dates against the crop stage and reporting window.
• Record dataset, resolution, processing and spatial coverage.
• Identify cloud cover, revisit gaps and mixed pixels near parcel edges.

Remote sensing cannot independently prove every agricultural practice. Dataset availability and authorized integrations are program-dependent.

Observations, photographs, activity logs, documents and instrument measurements should retain their collection method, source and time. An authorized sensor integration also needs device identity, units and calibration context. Records connect to the correct field, season and activity rather than becoming an unstructured attachment archive.

• Flag incomplete metadata, duplicate submissions and inconsistent timestamps.
• Keep original source records alongside transformations.
• Request additional evidence when location, timing or interpretation is ambiguous.

Photo and sensor integrations are proposed; this website does not claim that every described collection channel is connected.

Field → Season → Activity → Evidence → Claim → Assessment → Review → Report. A field can have many seasons, a season many activities and an activity multiple source records. A claim may require independent corroboration. An assessment records the criteria and methodology version applied, and a report keeps links to the decision and its sources.

• Trace a review decision back to the evidence used.
• Keep confidence factors and missing evidence alongside the assessment.
• Preserve relationships as source records or methods change.

A graph expresses provenance and relationships; it does not guarantee that a claim is true.

Potential AI functions support reviewers rather than replace them. Imagery classification takes suitable dated imagery and produces proposed categories; validation needs labeled reference data and checks appropriate to the crop and conditions. Pattern detection compares authorized records and surfaces anomalies for inspection.

• Document extraction: authorized documents → proposed structured fields → source comparison and human confirmation.
• Evidence-gap assistance: claim requirements and available records → missing-item suggestions → rules and reviewer checks.
• Summarization: assessed records → draft report narrative → source-linked editorial review.

These are potential applications, not claims of deployed or independently validated AI. No model score alone establishes an environmental claim.

Proposed validation checks whether evidence is present, the record belongs to the correct field, observations fall inside the reporting period, units and ranges are appropriate and the required methodology version is applied. Conflicts should be surfaced, not silently discarded.

• Illustrative states: Submitted → Validation → Needs Review → Assessed → Approved or Rejected.
• Authorized reviewers inspect sources, request missing information and record a rationale.
• Material decision changes should retain an actor, timestamp and prior decision.

These example states describe the intended workflow, not an assertion that the operational workspace implements it today.

Intended outputs include field assessments, program summaries, missing-evidence lists, exception queues and source-linked evidence packages for authorized reviewers. Output formats and external integrations depend on agreed requirements and available implementation.

• Available workspace exports contain the records accessible to the signed-in account.
• Proposed assessment reports include scope, period, methodology, coverage and limitations.
• Integration scope, authorization and data sharing are agreed before connection.