What Is dMRV? Digital Monitoring, Reporting and Verification for Nature-Based Carbon

A forest carbon project measures its trees, then does not look again for three to five years. In between, a fire can burn through it, a road can open up, a third of the seedlings can die and nobody finds out until the next inventory.

dMRV — digital measurement, reporting and verification, exists to close that gap. It combines satellite and aerial remote sensing with statistically designed field plots, propagates uncertainty explicitly, and produces an auditable data chain a registry or verifier can interrogate.

It is not a replacement for measurement. It is a replacement for guessing between measurements.

Traditional MRV runs on a cycle. A field crew inventories plots, a report is written, a verification body audits it, and credits issue. Then nothing happens for three to five years until the next cycle.

The problems with that cadence are practical. Between measurements, you have no visibility, meaning a fire, an encroachment, or a mortality event is only discovered at the next inventory, sometimes years later. Furthermore, the cost is concentrated and largely fixed, which is why small projects are frequently priced out entirely.

Cost benchmarks vary widely by project type and geography. According to analysis published by 1t.org and the World Economic Forum, MRV can run roughly USD 0.15–1.4 per tonne and account for up to a fifth of the total cost of a credit; separate published estimates put traditional field inventory for large REDD+ projects in the range of USD 5–15 per hectare per year.

What "digital" adds is not automation for its own sake. It is the ability to see change when it happens, and to show a verifier exactly how a headline number was produced.

A dMRV system that survives verification has four layers, and weakness in any one of them caps the credibility of the whole.

Layer 1 — Remote sensing. Optical satellite imagery for land cover and disturbance, radar for cloud-penetrating structural change, LiDAR where available for canopy height. This layer covers the entire project area continuously, but it does not measure carbon. It measures proxies.

Layer 2 — Field plots. Stratified sample plots where trees are actually measured: diameter, height, species. This is the only layer where carbon stock is observed rather than inferred. Its job is to calibrate and validate Layer 1.

Layer 3 — The statistical engine. Allometric equations converting measurements to biomass, stratification logic, sample-size determination, and uncertainty propagation through every conversion step. This is where most of the methodological risk lives, and where most published systems are least transparent.

Layer 4 — Reporting. Registry-ready outputs in the formats standards bodies expect, with the underlying data retained and traceable.

The common failure is a system strong in Layers 1 and 4, such as impressive maps and polished reports, but thin in Layers 2 and 3. A verifier specifically tests Layers 2 and 3.

Satellites measure reflectance and structure. Neither is carbon. The relationship between what a sensor records and what a tree contains is established by regression against ground measurements, and that regression is only valid within the range of conditions it was calibrated on.

A model trained on one forest type and applied to another, which has a different species mix, stand structure, and soil, produces confident numbers that are wrong. Verifiers know this, which is why methodologies specify ground-truth requirements rather than accepting remote estimates at face value.

The practical consequence: field sampling does not disappear under dMRV. It gets smaller, better targeted, and more statistically defensible.

The IPCC framework describes three tiers of emission factor precision. Tier 1 uses global default values, which are fast, free, and imprecise. Tier 2 applies country- or region-specific factors. Tier 3 uses measured, site-specific data and repeated inventories, delivering the lowest uncertainty.

dMRV is a route to Tier 3 for projects that could not otherwise afford it, because it reduces the cost of repeated measurement. But the tier is earned by the data, not by the software. A digital platform running on Tier 1 defaults is a Tier 1 system with a nicer interface.

Questions worth asking, in rough order of how much they reveal:

How is the sampling design determined, and from what measured variance? What uncertainty is reported, and how is it propagated through the allometric conversions? Can raw field data be exported and inspected by a third party? Which methodologies has output been accepted under, and by which verification bodies? What happens when the model and the field data disagree?

Vendors comfortable with those questions usually have good answers. Vendors who redirect to accuracy percentages without a published method are describing a claim, not a system.