Layer 01 · Data ingestion

Measurement, without the lock-in.

Every certificate starts with a real number, and Nset is MRV-agnostic. Performance enters the rail as primary Tier 3 data through our measurement partner Hyphen, or through any provider you already use — normalized into one canonical schema, so the legal instrument never depends on a single black-box model.

01 Two ways in

Primary data, or bring your own.

Stream A · Primary data · Hyphen aMRV

Tier 3, via Hyphen’s automated MRV

For the highest-resolution number, Nset natively integrates Hyphen aMRV deployed in your program to capture primary Tier 3 data at the source — eddy-covariance flux towers reading CO₂, CH₄, and N₂O continuously, upscaled across your boundaries.

Physics-based measurement, not modeled estimates

Continuous, quantified uncertainty attached to every reading

Aligned to GHG Protocol, ISO 14064, and IPCC Tier 3

Stream B · BYOM · MRV adapter

Bring-your-own measurement

Already working with a remote-sensing MRV provider? Keep them. A provider-agnostic adapter ingests their outputs into the rail’s canonical schema, so the instrument is built on your evidence — without vendor lock-in.

Any provider submits into one canonical MRV object

No rewriting the instrument per provider

No dependence on any single black-box model

02 How Hyphen aMRV measures

Physics at the tower. Coverage across the shed.

Hyphen’s atmospheric MRV (aMRV) reads the land continuously with eddy-covariance flux towers, then extrapolates that measured signal across the entire supply shed — validated by a second, independent tower, so every upscaled value carries a quantified uncertainty bar.

Hardware · Eddy covariance

Flux towers reading the land directly

Eddy-covariance towers measure the CO₂, CH₄, and N₂O exchanged between land and atmosphere continuously — a physics-based method, not an assumption.

Sampling rate

10–20 Hz, continuous

Spatial upscaling

footprint → landscape

Tower footprint

Upscaled, full supply shed

Tower-derived fluxes link to satellite and environmental layers, extending direct measurement across every acre — not just the area under the mast.

Uncertainty quantification

90% confidence

Estimate ± band

Independent tower points

<10% of independently measured values fall outside the 90% confidence interval — error is reported with every estimate, so the claim is defensible.

Optional · N₂O soil probes

Automated soil-flux chambers isolate nitrous oxide directly — deployed as continuous monitoring paired with a tower, or a 1–2 year campaign to refine emission factors.

Scales without redesign

A continuous Tier 3 monitoring network across the landscape.

Add towers, sites, and regions without rebuilding workflows — the same architecture grows from a single farm to a whole sourcing footprint.

24/7 /365

Continuous flux, not an annual snapshot.

Tier 3

Direct-measurement inventory — above default and country factors.

03 Your baseline

Your baseline is your inventory — the reference you measure progress against.

Your base year serves as a like-for-like reference point against which to measure progress. Each reporting period is compared back to that base year — so the credibility of every reduction claim rests on the base year being built from real data, not averages. Measurement is what makes the reference hold.

Base year · the reference

A like-for-like point to measure against

Your base year is the starting inventory you compare every reporting period against — not a forecast or a business-as-usual scenario, but your own footprint, held steady enough to make progress visible.

Reductions · observed

Measured data makes the comparison real

Supplier-specific measurement shows whether emissions changed in the supply chain, not just in the factor library. When practices change and emissions fall, it counts as progress — not a model update.

Eddy covariance

· physics-based

UN-endorsed

· supports NDC reporting

GHG Protocol

· Land Sector, aligned

ISO 14064

· quantification

04 The BYOM adapter

One evidence layer, many providers.

Remote-sensing outputs vary in format, resolution, model logic, and confidence intervals. The adapter normalizes them into a canonical object so the rail — and the legal instrument — stays provider-independent.

Source

MRV provider

Remote-sensing or field provider produces observations.

Adapter

Normalized evidence

Mapped into the canonical MRV object.

Logic

Verifier / registry

Project and methodology logic applied.

Settle

Issuance

Instrument payout, issuance, settlement.

05 The canonical MRV object

Every observation, the same shape.

Providers submit into one schema, not vendor-specific payloads. Each field is what makes a claim auditable rather than a black box.

project_boundary

Spatial boundary of the project or supply shed

coordinates · timestamp

Where and when the observation was acquired

sensor · source

Which instrument or dataset produced it

methodology_mapping

Reference to the methodology it satisfies

model_version

Exact version of the model that produced the estimate

confidence_interval

Reported interval around the estimate

uncertainty_treatment

How uncertainty is handled and discounted

qa_flags

Quality-assurance status and exceptions

supporting_artifacts

Underlying imagery or derived outputs, where available

provider

The submitting provider, for provenance

A canonical object lets the rail ingest outputs from many remote-sensing MRV providers — established or new — without rewriting the instrument each time.

06 What the adapter guarantees

Four properties that keep the instrument honest.

01 · Normalization

One object, not many payloads

Providers vary in format, resolution, model logic, and confidence intervals. The rail needs a canonical MRV object, not vendor-specific payloads — so downstream logic never has to know who produced the data.

02 · Provenance

Every claim carries its source

Provider, sensor or data source, acquisition and processing dates, model version, methodology reference, spatial boundary, confidence interval, and whether underlying imagery or only derived outputs are available.

03 · Version control

Real change vs. model drift

An estimate can change because the land changed, the dataset changed, or the model changed. The rail distinguishes them — so a certificate never reprices on model drift rather than real-world performance.

04 · Dispute & override

Rules for when sources disagree

Defined logic for what happens when Provider A and Provider B disagree, a project developer disputes a value, or a registry or verifier rejects the data — resolved by rule, not by argument.

Built for governance

Modular, interoperable infrastructure — with attention to governance, transaction integrity, information security, data interoperability, and digital MRV. The point is a legal instrument that does not inherit the failure mode of any single model.

Frequently asked

Measurement, answered.

Is Nset tied to a single measurement provider?

No. Nset is MRV-agnostic. Performance can enter as primary Tier 3 data through our measurement partner Hyphen, or through BYOM, where any provider submits observations into a canonical schema. The legal instrument never depends on a single black-box model.

What is Bring-Your-Own Measurement?

A provider-agnostic MRV adapter. Each provider submits observations into Nset’s canonical MRV object, so the rail can ingest many providers’ outputs without rewriting the instrument and without vendor lock-in.

How does Nset stop model drift from repricing an instrument?

Every MRV claim carries version metadata. When an estimate changes, the rail distinguishes whether the real world changed, the dataset changed, or the model changed — so a certificate reprices on genuine performance, not on model drift.

Next layer

02 · Allocation engine

Put the rails under your supply.

Not sure where your program’s data stands against the conformance bar? Get your readiness review.

Get your readiness review