Methodology v0.1 Beta · 2026-07-27

From parcel question to reproducible decision record

This methodology describes controlled-beta decision support. It separates mathematical consistency, software verification, Earth observation, field validation, legal review, and registry acceptance because none is a substitute for the others.

1. Intended use and decision boundary

Viridis supports early parcel screening, intervention comparison, evidence organization, monitoring planning, and design-partner review. Outputs are not registry-issued credits, appraisals, surveys, legal opinions, field certifications, or guarantees of value or outcome.

2. Parcel intake and identity

Each engagement begins with the decision, authorized parcel identifier or boundary, jurisdiction, land-system classification, time horizon, and available evidence. Automated parcel matching is treated as provisional until the partner confirms identity and boundary. Conflicts are surfaced rather than silently resolved.

3. Ecological condition and connectivity

Condition combines land-cover class, geometry, patch structure, corridor contribution, and observed pressure under an explicitly versioned scoring configuration. Forest and grassland follow different ecological trajectories. Ecological priorities are decision aids; they do not establish ownership, permanence, or legal protection.

4. Carbon-accounting boundary

Illustrative carbon ranges distinguish stored stock from a conservative, potentially creditable quantity after additionality, leakage, reversal buffer, removals, and data-quality gates. Biomass removed from a parcel cannot also remain in credited stock. Soil-carbon trajectories saturate. Production crediting would require a named program, approved method, baseline, monitoring plan, validation, verification, issuance, and sale.

5. Monitoring and provenance

Every evidence-bearing result records source class, observation time, method version, and whether data are live, partner-supplied, synthetic, mock, stale, or unavailable. Satellite or aerial observations can detect patterns and change; they do not by themselves prove cause, title, contractual compliance, or on-the-ground condition.

6. Verification ladder

  1. Illustrative estimate: broad assumptions or synthetic/demo inputs.
  2. Parcel-matched: a boundary and source inventory are matched but not necessarily field-validated.
  3. Partner-reviewed: the organization confirms parcel identity, assumptions, and decision relevance.
  4. Observed: dated remote or field evidence supports specific measured claims.
  5. Externally accepted: a qualified independent body accepts the artifact for its stated purpose.

7. What each assurance actually establishes

  • Mathematical consistency: stated relationships follow from stated assumptions.
  • Software verification: code satisfies tested or formally specified properties for the tested version.
  • Satellite observation: a sensor-derived signal was observed under documented limits.
  • Field validation: an authorized observer checked specified conditions in place.
  • Registry acceptance: the named program accepted a project or unit under its own current rules.

8. Uncertainty, calibration, and review

Reports expose parameter ranges, missing data, model limitations, alternative explanations, and sensitivity where practical. Ecological weights, thresholds, buffer sizes, prices, carrying capacity, jurisdiction rules, and remote-sensing calibration remain explicit choices. High-consequence or contradictory cases require human review and, where relevant, qualified field, legal, or registry professionals.

9. Validation and reproducibility

Each report identifies the input inventory, source class, retrieval or observation date where applicable, scoring configuration, method version, material transformations, automated checks, reviewer, and change history. Automated regression tests establish tested software behavior; partner review confirms decision context; field or registry validation must be performed by the qualified party responsible for that external conclusion.

10. Known oracle and empirical limitations

Remote providers can be unavailable, stale, cloud-obscured, spatially coarse, misclassified, or outside their calibration domain. Parcel services can return incomplete or conflicting boundaries. Model coefficients, ecological response, biomass estimates, prices, and future program rules can change. Viridis labels unavailable or fallback data, preserves contradictions, and requires human review rather than silently promoting a weaker source.

11. Additionality, leakage, and permanence

These are project-level claims, not properties conferred by software. Beta reports identify the evidence that would be needed, apply conservative illustrative deductions where stated, and refuse to label an estimate as an issued credit. A recorded agreement and competent holder—not a digital signature alone—create enforceable land obligations.

12. Authoritative reference frame

These sources frame terminology and evidence expectations. Their citation does not imply endorsement, validation, or acceptance of a Viridis output.

13. Change log

v0.1 Beta — 2026-07-27: public versioning introduced; assurance types separated; intended use, parcel identity, uncertainty, carbon boundary, human review, provenance, additionality, leakage, permanence, and registry limits made explicit.