Soil carbon is a category where remote sensing promises more than it can yet deliver. Satellite-based monitoring of soil organic carbon, paired with machine-learning models trained on soil samples, has improved enough that it now plays a meaningful role in soil carbon project workflows. But the gap between “contributes useful information” and “replaces direct sampling” remains real, and the contested space between those two positions is where vendor marketing and protocol requirements often diverge.
This piece works through what satellite-based soil carbon MRV genuinely accomplishes, where it still requires sampling, and how procurement teams should evaluate vendor claims.
What the underlying physics actually allows
Soil organic carbon doesn't have a clean satellite signature. Unlike atmospheric methane (with specific absorption bands) or surface vegetation (with characteristic spectral patterns), soil carbon content must be inferred from proxies — soil color, surface moisture, vegetation patterns, and other observable variables — combined with ground-truth samples that calibrate the inference.
This means satellite soil carbon estimation is a modelling exercise: optical or radar inputs go into trained models that produce SOC estimates. The estimates' accuracy depends on the quality of training data, the appropriateness of the model for local conditions, and the robustness of the inference under conditions that differ from the training set.
Two practical consequences follow. First, satellite-derived SOC is most reliable where training samples are dense and representative — typically well-studied agricultural regions in the US, EU, and parts of Australia. It is less reliable in regions where soil sampling has been sparse or methodologically inconsistent. Second, satellite estimates lag in time — the model's training data ages, and SOC dynamics differ across years, climates, and management practices.
The vendor landscape
Several companies offer satellite-augmented soil carbon services, with meaningful differences in approach.
CIBO uses simulation-based modelling that integrates remote sensing, climate data, soil profile data, and management records to produce field-level SOC estimates and trajectories. The approach emphasizes mechanistic process modelling rather than purely empirical inference.
Indigo Carbon, operating within Indigo Ag's broader carbon program, uses remote sensing alongside soil sampling within a vertically integrated program structure. The remote sensing is a component of the workflow rather than a standalone product.
Yard Stick, more recently, has focused on a hybrid approach: in-field measurement combined with remote sensing to scale verification across larger acreage than direct sampling alone could cover.
These vendors are not interchangeable. Their approaches reflect different bets on what protocol verifiers will accept, what farmers will adopt, and how the regulatory and market environment will evolve.
What protocol verifiers actually require
Soil carbon credit protocols — including Verra, Gold Standard, Climate Action Reserve, and CDM — have not converged on accepting satellite-only MRV for high-integrity credits. The major protocols require direct soil sampling on defined sampling schedules and densities, with remote sensing serving as a supplementary verification layer rather than a replacement.
This boundary is sometimes blurred in vendor marketing. “Satellite-based MRV” in vendor materials often means “a workflow that includes satellite analytics alongside required physical sampling” rather than “satellite data replacing sampling.” The distinction matters for what credit buyers should expect from any specific project.
The protocols' caution reflects real uncertainty about satellite-only accuracy at the levels of precision required for tradeable credits. SOC changes of agricultural relevance are small relative to baseline SOC stocks and natural variability. Satellite-derived estimates that get within 20-30% of true values are useful for many purposes; for carbon credit issuance, the precision requirement is much tighter.
Where satellite-augmented MRV genuinely helps
Despite the protocol limits, satellite-based MRV plays valuable roles in soil carbon workflows.
Sample stratification. Where physical sampling is required, satellite analytics can guide where samples are taken — identifying within-field variability that informs efficient sampling design. A 100-acre field is rarely homogeneous; satellite-derived variability maps help allocate sampling resources where they'll yield the most information.
Management practice verification. Satellite imagery confirms whether claimed management practices — cover crops, no-till, residue retention — are actually occurring on enrolled fields. This is a different question from SOC quantification, and it's one where remote sensing is genuinely reliable.
Risk identification. Substantial SOC loss events — tillage of long-undisturbed soil, conversion of cover-cropped fields, fire — can be flagged through satellite monitoring even when their precise impact on SOC requires sampling to quantify. For project risk management, this matters.
Cost reduction on supplementary measurement. When satellite analytics reduces the sampling density required by a protocol — while still meeting protocol requirements — the cost savings can be substantial. Projects move from prohibitively expensive direct measurement to economically viable hybrid approaches.
What buyers should require
For procurement teams evaluating soil carbon credits, three questions separate credible offerings from less credible ones.
First, what protocol does the project follow, and what does the protocol actually require for SOC verification? If the answer is “a proprietary methodology accepted by the issuing vendor's own registry,” the verification is weaker than projects under established multi-party protocols with independent verification.
Second, what is the role of physical sampling in the MRV workflow? Projects that rely overwhelmingly on satellite estimation with minimal sampling face higher technical and credibility risk than projects with appropriate sampling design.
Third, what's the uncertainty quantification? Credible providers disclose confidence intervals on their SOC estimates rather than presenting point values. Uncertainty grows with longer prediction horizons and conditions further from training data; estimates without uncertainty are less trustworthy than estimates with disclosed bounds.
The honest framing
Satellite-based soil carbon MRV is a genuinely valuable tool, used appropriately. It scales verification activities that would be cost-prohibitive with sampling alone. It provides ongoing monitoring between sampling events. It identifies anomalies that warrant investigation. It supports more efficient sampling design.
It does not yet replace direct soil sampling for high-integrity credit issuance, and the protocols that govern credible credit issuance are unlikely to accept satellite-only MRV in the near term. Vendor positioning that suggests otherwise is overstating the technology's current capabilities.
For buyers, the implication is to read project methodologies carefully rather than relying on technology labels. Projects with credible hybrid approaches — appropriate sampling design augmented by satellite analytics — can produce high-quality credits. Projects relying disproportionately on remote sensing alone are buying technology hype rather than verification rigor.