Soil Carbon MRV: How Regen Practice Becomes a Verifiable Credit

The Soil Core: What Direct Sampling Measures

The most defensible MRV tier is the physical soil core. A steel tube is driven into the ground at defined depth intervals, extracting a cylinder of earth that represents a known volume of the soil column. The sample is oven-dried at 105 degrees Celsius to remove all moisture, then combusted at high temperature using dry combustion (ISO 10694:1995) to determine total organic carbon concentration in grams per kilogram of soil. A 0-30 centimetre profile captures the biologically active zone where most management-induced carbon change occurs over three to five year timescales. A 0-100 centimetre profile adds the stabilised fraction that accumulates over decades and is required to support any permanence claim under Verra VM0042 version 2.0 (2022).

Two corrections are non-negotiable before a concentration reading becomes a carbon stock. Bulk density converts the mass-based concentration measurement into a volumetric stock figure in tonnes of carbon per hectare. Compacted soils and well-structured aerated soils at the same carbon concentration hold fundamentally different quantities of carbon per hectare; without the bulk density correction, the numbers cannot be compared across fields, years, or management histories. Cation exchange capacity correction accounts for how mineral surface charge affects the stabilisation and retention of organic matter, ensuring that the stock change reported reflects management-driven accumulation rather than differences in clay mineralogy between sampling depths or dates.

Composite sampling distributes the cost of spatial variability. A single grab sample from one location in a field is not statistically representative of the parcel. The Climate Action Reserve Soil Enrichment Protocol v1.1 (2021) requires a minimum of twelve to fifteen soil cores per management stratum, composited into a single laboratory sample. For a fifty-hectare project site stratified into three management zones, that is between thirty-six and forty-five cores at baseline and again at each remeasurement event. Remeasurement intervals are fixed at three to five years under most active protocols; shorter intervals produce a carbon stock signal that cannot reliably exceed the analytical noise floor for most management systems.

Laboratory cost for the full composite sampling regime runs from one hundred to five hundred US dollars per composite sample depending on the depth profile and analytical panel required. At three to five composites per stratum across a typical three-stratum project site, full baseline sampling costs between nine hundred and seventy-five hundred US dollars in direct laboratory fees before project management, site access, and logistics are added.

COMET-Farm and the Model Layer

Direct sampling anchors the measurement at the point of the core. Process models extend it through time and across spatial gaps between sample points. COMET-Farm is the most widely used free simulation tool for agricultural carbon accounting in North America. Developed by Colorado State University in partnership with the USDA Natural Resources Conservation Service (operational since 2014), it runs on the DayCent biogeochemical model, which simulates soil organic carbon as a function of climate data, soil texture and depth, decomposition kinetics, and management inputs including tillage intensity, crop rotation, cover-crop species, grazing pressure, and organic matter applications (Del Grosso et al., 2001, Ecological Modelling). An operator enters a management history for a defined field parcel; the model returns a time-series estimate of carbon stock change. The web interface is free, though calibrating outputs against field-specific measured parameters requires technical interpretation.

The Cool Farm Tool, originally developed by Unilever and the University of Aberdeen and now managed by the Cool Farm Alliance (Hillier et al., 2011, Soil Use and Management), offers farm-scale greenhouse gas and carbon accounting built on the RothC soil carbon turnover model. Platform subscriptions run from zero to five hundred US dollars per year depending on organisation tier. Yasso, a model from the Finnish Environment Institute based on the litter decomposition research of Tuomi et al. (2011, Ecological Modelling), is used less frequently in commercial MRV programmes but is technically appropriate for agroforestry and silvopasture systems with substantial woody litter inputs, where its decomposition pathway representation is more accurate than DayCent's simplified residue module.

The ceiling on model-based MRV is uncertainty. COMET-Farm carries a published uncertainty range of approximately plus or minus twenty to forty percent at individual field scale when management inputs are drawn from operator records rather than directly measured soil parameters (Colorado State University COMET-Farm Technical Documentation, 2023). That uncertainty is reducible through calibration against direct-sample data, but model-only MRV is consistently penalised by conservative crediting deductions in active protocols, meaning that the higher the model uncertainty, the fewer credits per tonne sequestered the project can claim.

From Orbit: Remote Sensing as a Proxy Layer

Sentinel-2's multispectral imager detects spectral reflectance in the shortwave infrared bands (1,450 to 2,500 nanometres) where the organic matter content of bare mineral soil produces a measurable absorption signal. The European Space Agency Copernicus programme has made this data free at ten-metre spatial resolution with a five-day revisit cycle since 2015. Castaldi et al. (2019, Remote Sensing of Environment) demonstrated partial least-squares regression mapping of topsoil organic carbon from Sentinel-2 bare-soil composite imagery at county scale, achieving R-squared values of 0.51 to 0.68 depending on soil mineralogy and clay content. This correlation is operationally useful for stratifying a project area into zones of differing soil carbon density before sampling, reducing the number of cores needed to achieve statistical precision. It is not credit-grade measurement accuracy for absolute stock estimation.

Three structural constraints define what remote sensing can and cannot contribute. First, passive optical sensor penetration depth is approximately zero to five centimetres. A verification protocol requiring a zero to thirty centimetre carbon stock figure cannot substitute satellite reflectance at that depth. Second, the bare-soil spectral algorithm requires cloud-free observations on dates when the soil surface is exposed, a condition many agricultural parcels meet for only a fraction of the annual calendar. Third, NDVI, the normalised difference vegetation index widely cited as an indirect soil carbon proxy, measures green canopy biomass, not carbon stocks. NDVI correlates with carbon inputs to the soil over time, because more biomass produces more litter, but it does not measure existing organic carbon stocks at a point in time.

No active verification protocol accepts remote sensing as a stand-alone evidence tier for carbon credit issuance. Verra VM0042 v2.0 (2022) and the Gold Standard SOC Methodology (2021) both treat remote sensing as a stratification and change-detection tool only. Its value is in the hybrid stack, not as a replacement for ground-truth measurement.

Hybrid Stacks and the Active Protocols

Hybrid MRV pairs ground-truth sampling with spatial interpolation and temporal extrapolation. The combination reduces the cost of full direct-sampling coverage while preserving the measurement floor required for market-credible credit issuance. Verra VM0042 version 2.0 (2022) formalises this through a tiered evidence structure. Tier 1 applies IPCC default emission factors, which carry high uncertainty and generate conservative crediting. Tier 2 applies empirically calibrated simulation models such as COMET-Farm, with moderate uncertainty and moderate credit generation. Tier 3 applies direct measurement with model cross-validation, carrying the lowest uncertainty and the highest credit issuance per tonne sequestered. Most commercially active soil carbon projects operate at Tier 2 with Tier 3 validation sampling at baseline and each remeasurement event.

The Gold Standard SOC Methodology, revised in 2021, accepts hybrid approaches where direct baseline samples anchor the model calibration and a statistical subset of direct cores validates model extrapolations at remeasurement. The Climate Action Reserve Soil Enrichment Protocol v1.1 (2021) requires direct sampling at baseline and each remeasurement period but allows process model interpolation between sample points across the project area. Emerging implementations layer machine-learning spatial interpolation over these foundations, training algorithms on national soil survey databases such as SoilGrids from ISRIC or the USDA Web Soil Survey to weight the extrapolation of ground-truth samples across the parcel using terrain, soil-type, and remote-sensing covariates. The output is a wall-to-wall soil organic carbon map updated at each remeasurement, with uncertainty bounds propagated through the full analytical chain.

This is the architecture the Farm Intelligence pillar identifies as the verification layer for every mechanism pillar that makes a carbon claim. Mycorrhizal-fungi networks, cover-crop biomass additions, compost applications, and rotational-grazing residue cycling all accumulate in the same soil carbon pool. MRV is how the operator makes that accumulation legible at the resolution a market requires.

The Economics and the Honest Counter

Verification cost in the voluntary carbon market for soil carbon projects runs from fifteen to forty-five US dollars per tonne of CO2 equivalent issued (Ecosystem Marketplace Insights Brief, 2024). Three cost components combine to produce that range. Laboratory fees for baseline and remeasurement composite sampling run from one thousand to twenty-five thousand US dollars per project per vintage depending on project area and stratum count. Third-party verification audits under Verra or Gold Standard accreditation cost between five thousand and twenty thousand US dollars per verification event. Verra's registry charges zero-point-two US dollars per tonne of CO2 equivalent for credit issuance. At five hundred tonnes CO2 equivalent per year project scale, the combined total runs between seven thousand five hundred and twenty-two thousand five hundred US dollars, a per-tonne verification cost of fifteen to forty-five dollars.

The voluntary market prices standard soil carbon credits at twenty to forty US dollars per tonne, with premium projects carrying documented additionality, third-party permanence insurance, and measured co-benefits reaching fifty to eighty US dollars (Ecosystem Marketplace 2024). The arithmetic at small scale is thin: a fifty-hectare cover-cropping project sequestering 0.3 tonnes CO2 equivalent per hectare per year generates fifteen credits annually. At thirty dollars per tonne, that is four hundred and fifty US dollars in revenue. Minimum verification costs are seven thousand five hundred dollars. The economics close only above approximately two hundred to five hundred hectares under conservative assumptions, or where co-benefit premiums drive the credit price to sixty dollars or above.

The additionality objection is the most substantive structural challenge to soil carbon markets at scale. Additionality requires that the carbon sequestration would not have occurred without the financial incentive of the credit payment. For cover crops and reduced tillage, the risk is significant: regional adoption was already rising before most credit programmes launched in the US Corn Belt and Midwest. A 2022 review in Nature Climate Change (West et al.) documented systematic over-crediting in US soil carbon projects where field-level management was improving independently of credit programme enrolment, implying that a material fraction of issued credits represented sequestration that would have occurred in the absence of the payment. Gold Standard, Verra, and the Climate Action Reserve each operationalise additionality tests differently, but none has a fully reliable method for distinguishing financial additionality in a sector where agronomic adoption is accelerating for independent reasons.

Permanence compounds the challenge. Soil organic carbon is not geologically stable. A drought year, a single deep tillage event for pest management, or a land-use conversion reverses in months what accumulated over a decade. Active protocols address this through buffer pools and conservative permanence deductions that require operators to over-sequester to account for projected reversal risk, but a twenty-year permanence obligation is a genuine constraint on management flexibility that most operators underestimate at programme entry. The biochar-based carbon credit market, surveyed in detail at Biochar Carbon Credits 2026, illustrates a more durable permanence profile: biochar's one-hundred-year carbon stability removes the reversal risk that shadows soil organic carbon. The two pathways are complementary: biochar locks permanently; soil organic carbon builds and must be actively maintained.

The soil held the carbon before the market asked. MRV did not create the sequestration. It made the number legible enough to hold.