Agroforestry carbon MRV: measuring trees one at a time

Most carbon methodologies count plots. Agroforestry counts trees — each with its own species, age, height, girth and location. That single difference reshapes what the field survey and the verification workflow have to do.

Agroforestry increases carbon sequestration by integrating trees into farming systems. Alongside the carbon, it improves soil health, supports biodiversity and builds climate resilience — and it generates carbon credits that give farmers a direct reason to plant and maintain trees.

Those credits let businesses offset emissions while supporting reforestation. But the measurement problem is unlike any other nature-based methodology, because the unit of account is not a field. It is a tree.

Why agroforestry MRV is measured per tree

In a rice or biochar programme, the record is a plot or a production batch. In agroforestry — and in afforestation, reforestation and revegetation (ARR) projects generally — biomass, and therefore sequestered carbon, is estimated from the dimensions of individual trees. Species, age, height and girth (DBH) all feed the allometric calculation, and they change every season.

That makes the field survey the hard part. A project with a hundred thousand trees needs a hundred thousand accurate, located, photographed, re-visitable records — collected by field teams often working well beyond mobile coverage.

The methodologies now say so explicitly. Verra’s ARR methodology VM0047 splits projects into an area-based approach, which pairs remote sensing with field sampling against a dynamic baseline, and a census-based approach for dispersed plantings that keep the existing land use. Agroforestry is the second case, and the census-based route asks for a complete census of planting units — every tree, not a sample. Our Agroforestry dMRV app is built around that requirement.

What the field app captures

• Per-tree measurement — species, age, height and girth recorded against each individual plant.
• Location per plant — every tree carries its own coordinates, not just a plot boundary, so re-visits find the same tree.
• Automated and manual measurement — height and girth captured either way, so field teams are not blocked when conditions defeat the automated route.
• Photographic evidence — photo and location captured per plant, giving verifiers something to check against.
• Works offline — full functionality without a connection, which is the normal condition in plantation landscapes.

Verification, built into the workflow

Collection is only half of it. The monitoring dashboard is where the data becomes evidence:

• Live survey monitoring — progress and incoming data visible from the desktop while teams are still in the field.
• Validation of everything collected — all mobile app data checked rather than sampled.
• Two-step verification — a second review pass targeting complete accuracy in the record that goes forward, before anything reaches the validation and verification body (VVB).
• Two-way communication — validators and ground teams resolve queries directly, so corrections happen while the team is still on site.
• Integrated analytics — survey effort optimised, saving time and field resource.

The effect is that errors are caught while they are still cheap to fix. A discrepancy found during the survey costs a message; the same discrepancy found at verification can cost a season. What the VVB eventually receives is the whole evidence pack — per-tree records, photographs, GPS trails and plot boundaries as shapefile, KML or GeoJSON — rather than a summary it has to take on trust.

From tree dimensions to tonnes of CO₂e

What a surveyor records is not carbon. It is species, age, height and girth — and girth taken at breast height is what a methodology means when it asks for DBH, the diameter at breast height. Dimensions, in other words. The methodology supplies the conversion from those dimensions to biomass — a species-wise allometric equation that turns DBH and height into above-ground biomass (AGB), with a root-to-shoot ratio for the below-ground part (BGB) — and from biomass to carbon dioxide equivalent. The field survey supplies the inputs, and any error in them propagates all the way to the tonnage claimed.

How much of that estimate survives contact with an auditor depends on things the survey design decides long before the calculation: how strata are drawn, whether a sample is representative of them, and what uncertainty deduction the resulting spread forces on the claim. A census sidesteps the sampling argument entirely — there is nothing to extrapolate from — but it only does so if the census is genuinely complete, which is a QA/QC problem rather than a statistical one.

A well-run agroforestry system sequesters in the range of 5–15 tonnes of CO₂e per hectare per year. That headline rate is worth reading carefully, because it is a per-hectare summary of numbers that were never collected per hectare: it is the sum of individual trees, divided by the area they stand on. The range is wide because species mix, planting density and stand age differ enormously between projects — and the only thing that explains where a given project sits within that band is its own tree record.

The second survey is the one that decides the credit

A first survey produces an inventory: what is standing, where, and how big. No credit follows from it. Credit follows from growth — the difference between two measurements of the same tree, taken in two different monitoring periods. That increment, aggregated across a stand, is what a forester would call mean annual increment, and it is the only thing a crediting period actually pays for.

Growth is only half of what the second visit establishes. The other half is survival. A tree that has died between monitoring periods has to come out of the claim, and only a record that can be reconciled tree by tree can show whether it did. Satellite canopy and NDVI change over the same parcels is a useful cross-check between field visits, but it cannot tell a project which individual stems it has lost.

That reframes the whole exercise. A survey that cannot be repeated against the same individual plants has not produced a baseline; it has produced a snapshot. If the second survey cannot be reconciled tree by tree with the first, a project does not have a growth increment to claim. It has two inventories that happen to cover the same land.

This is why per-plant coordinates matter more than they first appear. A plot boundary tells a returning field team which field to walk into. It does not tell them which of four hundred trees inside it is the one measured at 2.1 metres eighteen months ago.

What makes a tree record re-visitable

• Its own location, not the plot’s — coordinates captured per plant, so a re-visit is a navigation problem rather than a search.
• A photograph taken at capture — the fastest way for a returning surveyor, or a verifier, to confirm this is the same tree.
• Species and age carried forward — recorded once and reused, rather than re-judged by whoever happens to walk the plot next season.
• Captured offline, at the tree — a record written where the measurement happens, not reconstructed from a notebook after the team is back in signal.
• Validated before the team leaves — a missing DBH is a two-minute fix during the survey and a lost plant at verification.

None of this is exotic. It is simply the difference between a dataset that supports a second monitoring period and one that quietly cannot.

Related reading

The Agroforestry dMRV app page covers the platform itself, and Apps shows the wider family of field applications. For how agroforestry sits alongside biochar, AWD, DSR and mangrove blue carbon, see DMRV methods.

• The 5M approach to running a digital MRV programme
• The field and verification failures that cost credits
• What a biochar dMRV system has to capture
• How agroforestry carbon credits are generated and verified