A digital MRV system is four layers stacked on each other, and the credibility of the whole is set by the weakest one. Most attention goes to the top and bottom, which include impressive imagery and polished reports, while verification concentrates on the two in the middle.
The Four Layers
Layer 1 — Remote sensing. Continuous coverage of the whole project area. Optical for land cover and disturbance, radar for structure and cloud penetration, LiDAR for canopy height where available. This layer sees everything and measures no carbon.
Layer 2 — Field plots. Stratified ground measurement. The only layer where carbon stock is observed rather than inferred. Small in area, decisive in influence.
Layer 3 — The statistical engine. Allometric conversion, stratification logic, sample-size determination, uncertainty propagation. Where the numbers actually become numbers.
Layer 4 — Reporting. Registry-ready outputs, retained inputs, traceable derivation.
| Layer | Covers | Measures | Typical weakness |
|---|---|---|---|
| Remote sensing | All area, continuously | Proxies only | Saturation, cloud, calibration range |
| Field plots | Sampled locations | Carbon stock directly | Too few, poorly stratified |
| Statistical engine | All calculations | — | Opaque, uncertainty unpropagated |
| Reporting | Outputs | — | Inputs not retained |
Why the Middle Layers Decide Everything
A verifier does not assess a map. They assess a number and its derivation.
That puts Layers 2 and 3 under direct scrutiny: how were plots selected, how many, on what basis, which equations, what uncertainty, propagated how. A system strong at Layers 1 and 4 and thin in the middle produces confident-looking outputs that cannot be defended when questioned.
This is also where systems differ most and disclose least. Sensor specifications are published; sampling design and uncertainty methodology frequently are not.
How the Layers Talk to Each Other
The critical interface is between Layers 1 and 2, and it runs in both directions.
Ground to sensor: field plots calibrate the relationship between sensor values and biomass. Without them, remote estimates have no anchor.
Sensor to ground: remote monitoring detects change between inventories, which targets where field effort should go next. A disturbance flagged in Layer 1 becomes a re-measurement priority in Layer 2.
Systems that treat these as separate workflows, such as a mapping team and a field team producing parallel outputs, lose most of the value. The integration is the product.
What Good Looks Like
A defensible stack has a few observable properties.
The calibration range is stated, and estimates outside it are flagged rather than silently extrapolated. Validation plots are held out from model fitting, so reported error describes prediction rather than memorisation. Uncertainty is propagated through every conversion rather than attached at the end. Raw field records are retained and exportable. And the sampling design is documented well enough that someone else could repeat it.
None of those require a particular vendor. All of them are checkable.
Where Projects Usually Cut Corners
Three patterns recur.
Buying coverage instead of calibration, meaning extensive imagery with a thin, convenience located ground sample.
Reporting model fit statistics from training data as if they were accuracy.
Treating uncertainty as a disclosure obligation rather than a design parameter, so it is calculated once at the end instead of driving decisions at the start.
Frequently Asked Questions
dMRVThe statistical engine and the field sample. They determine whether a number survives verification.
Image credit
Hero image: Global tree canopy map by NASA Earth Observatory (Jesse Allen and Robert Simmon, based on data from Michael Lefsky). Public domain, via Wikimedia Commons.
Sources
1. IPCC 2006 Guidelines Vol. 4 (AFOLU) — https://www.ipcc-nggip.iges.or.jp/public/2006gl/vol4.html (accessed 16 Sep 2026)
2. Verra methodologies and monitoring requirements — https://verra.org/methodologies/ (accessed 16 Sep 2026)
3. ICVCM Core Carbon Principles — https://icvcm.org/core-carbon-principles/ (accessed 16 Sep 2026)
Move beyond estimates
Verifiers test sampling design, uncertainty and whether a number traces back to the field. TREEO dMRV captures that evidence in real time, in one auditable chain.
