GREET Model vs Primary Data: How Greentruth's Data-Quality Hierarchy Works

The GREET model vs primary data question is the most common technical question buyers and MRV teams ask before adopting a verified-emissions-token program. Should the CI score come from a modeled default like Argonne's R&D GREET? Or should it come from measured primary data at the producer site? The honest answer is: both, in a defined hierarchy, with each tier of data recorded on the token and each tier auditable under ISO 14064-3 reasonable assurance. This page lays out how the trade-off actually works inside the QET methodology.

GREET model vs primary data, in one paragraph. Greentruth handles the trade-off between modeled defaults and primary measured data as a four-tier data-quality hierarchy: site-specific primary → process-specific primary → peer-reviewed secondary → industry-average secondary (where the R&D GREET 2025 defaults live). Each QET records the tier the underlying data sits in as a token attribute. Higher tiers strengthen defensibility under reasonable assurance; lower tiers remain auditable. CA-GREET 3.0 applies only inside the QET-LCFS extension for California Low Carbon Fuel Standard submissions.

What a QET is in the first place

The trade-off is older than QETs. Every lifecycle CI calculation has to source its inputs from somewhere — and the choice between modeled defaults and measured primary data has consequences the industry has been working through for decades.

• Modeled defaults. Lifecycle models like Argonne National Laboratory's R&D GREET publish standardized emission factors for energy pathways — compression, transmission, flaring, fugitives, processing, distribution. They are reproducible, auditable against a public dataset, and refreshed on a defined cadence. Their limitation: they describe an average producer or operator, not the specific producer or operator behind the gas the buyer is consuming.
• Primary measured data. Continuous-monitoring sensors, flyovers, satellite measurements, third-party inspection — measured values from the specific physical asset the buyer's supply chain depends on. Their strength: producer- and pathway-specific, not averaged. Their limitation: they require operational measurement programs that not every producer has at every site, and the data needs to survive a verifier's reasonable-assurance review.

The trade-off is real. Either pure approach has known failure modes. The QET methodology resolves it as a hierarchy — not a choice — with every token recording where in the hierarchy its data lives.

How the pipeline pathway model uses GREET defaults at scale

The QET-NG methodology v2.3 defines four tiers of input data, ordered by defensibility:

• Site-specific primary. Measurements at the specific producer site contributing to the buyer's supply. The strongest tier. Examples: continuous methane monitoring at a specific pad, source-level direct measurement under OGMP 2.0 Level 4 or Level 5, calibrated meter data with documented audit history.
• Process-specific primary. Measurements at a comparable process at the same operator, where site-specific data isn't available. Strong but one tier down. Example: an operator's source-level measurement program covering most sites with statistical extrapolation to a specific site whose direct measurement is gapped.
• Peer-reviewed secondary. Published peer-reviewed data for the relevant production or processing pattern. Useful where primary data is unavailable. Example: published peer-reviewed studies of a specific basin's methane intensity.
• Industry-average secondary. Industry-average factors — including the R&D GREET 2025 defaults. The baseline tier, still auditable but the most generic.

The hierarchy is not a "use only the top tier" rule. It is a layered defensibility model: the buyer's audit team and the framework verifier both want to know, for each input, which tier it came from. A site can run on Tier 1 data for production and Tier 4 modeled defaults for distribution and still be fully audit-defensible — as long as the tier is recorded on the token.

How biogenic emissions interact with the hierarchy

For the full QET-NG methodology

R&D GREET (Greenhouse gases, Regulated Emissions, and Energy use in Technologies) is Argonne National Laboratory's lifecycle modeling dataset. It is the public reference dataset that anchors Tier 4 (industry-average secondary) in the QET-NG methodology's data-quality hierarchy.

Specifics that matter for an MRV team or verifier:

• R&D GREET 2025 is the version in force. Each QET records the exact dataset version used at issuance, so framework exports can be reconstructed against the underlying data lineage.
• The dataset is refreshed annually. Greentruth's methodology pins the refresh cycle: the new version is incorporated within the first 30 business days of each year, and tokens issued after the cutover use the new version.
• It covers the right scope. Compression, transmission, flaring, fugitives, facility venting, pipeline fugitives, pipeline venting — all under the lifecycle boundary the QET methodology contemplates.
• Standardized GWP factors. IPCC AR5 GWP100 (CH₄ = 28, N₂O = 265) — the same factors the QET methodology applies across multi-pollutant CI.

Argonne R&D GREET model

A common point of confusion: CA-GREET 3.0 is not a substitute for R&D GREET 2025. It is California's adaptation of the lifecycle model, modified for the regulatory context of California's Low Carbon Fuel Standard (LCFS). It applies in one specific place inside the QET methodology:

• The California variant is invoked only inside the QET-LCFS extension methodology — the companion methodology layered on top of the core QET methodology for fuel pathway holders submitting to CARB under the LCFS.
• Outside LCFS submissions, R&D GREET 2025 is the reference dataset. A buyer or producer that is not submitting to CARB does not need to engage with the California variant.
• The two are not in conflict. They are different versions of the same underlying lifecycle model, tuned for different regulatory contexts.

This distinction matters for the team's audit posture. A QET-RNG with the QET-LCFS extension carries the CA-GREET 3.0 modeling reference; a QET-NG outside an LCFS submission carries the R&D GREET 2025 reference. Each is correct in its context; mixing them up is a common verifier-side flag.

How LCFS compliance handles CA-GREET 3.0 specifically

How mass-balance chain-of-custody interacts with the modeling layer

CARB Low Carbon Fuel Standard

The hierarchy is only useful if it's queryable on a per-token basis. Every QET on the EarnDLT registry records, as a token attribute:

• The MRV tier of the underlying data (Tier 1 → Tier 4).
• The methodology version that produced the carbon intensity calculation.
• The reference dataset version used (R&D GREET 2025 for the core methodology; CA-GREET 3.0 only inside the QET-LCFS extension).
• The verifier of record that signed off under ISO 14064-3 reasonable assurance.

For the buyer running a Discovery search, the MRV tier is a filterable attribute. A buyer building a defensibility-led procurement program can filter on Tier 1 or Tier 2 specifically; a buyer optimizing for cost can accept Tier 4 defaults — and either decision is recorded on the token in a way that survives a verifier review.

For the foundational QET overview

The data-quality hierarchy gets tested differently under different assurance regimes. Limited assurance and reasonable assurance ask different questions about the same data.

This is the practical implication of the assurance escalation under SB 253, CSRD ESRS E1, and parallel regimes: industry-average factors that survived limited assurance get tested harder under reasonable assurance. The team that builds its procurement program around Tier 1 / Tier 2 data has the easier transition.

How ISO 14064-3 reasonable assurance underpins verification

How the GHG Protocol handles this

In practice, virtually no real-world QET is built on a single tier of input data. The methodology supports hybrid construction — Tier 1 measurements at some stages of the value chain, Tier 4 GREET defaults at others — and records the tier per stage, not just per token.

How this looks on a QET-NG covering a full natural-gas pathway:

• Production stage. Often Tier 1 or Tier 2 where the producer runs source-level measurement under OGMP 2.0 Level 4 or Level 5.
• Gathering and boosting stage. Often Tier 2 or Tier 3 depending on the operator's measurement program.
• Processing stage. Often Tier 3 or Tier 4.
• Transmission stage. Often Tier 4 (modeled defaults applied across EarnDLT's 32,000-segment Lower-48 network model), unless the pipeline operator has primary data feeding the segment.
• Storage and distribution stages. Often Tier 4 unless the operator has primary data.

The hybrid construction is not a compromise — it is the realistic shape of a CI calculation across the full natural-gas value chain. Different operators have different measurement programs at different stages, and the QET records the tier of each stage so the audit trail is reconstructible.

The downstream GasTrace product runs on this hybrid architecture: free default tier at every stage (Tier 4 baseline), with optional producer-specific QET upgrades at specific stages where the buyer wants to sharpen the input from Tier 4 to Tier 1 or Tier 2.

How retirement records preserve methodology version

How QETs align across reporting frameworks