What Is dMRV? Digital MRV for Carbon and Restoration Projects
dMRV is digital measurement, reporting and verification for carbon and restoration projects. It is only as strong as the field data underneath.
dMRV (digital measurement, reporting and verification) is the practice of showing a project's results with digital data that can be traced back to its sources: satellite and drone imagery, field records, photos and the models that turn them into numbers. For a carbon project, it means showing a registry and its auditors how many tonnes the trees removed, in a way they can check. For other restoration projects, it means the same for tree cover or survival, for a funder.
Most conversations about dMRV start with the satellites. We think it should start with the field data, because that is where most of the errors get in.
What does MRV mean?
MRV stands for measurement (or monitoring), reporting and verification. The phrase comes from climate policy: commitments had to be measurable, reportable and verifiable. It now covers how any project proves what it did.
- Measurement: collecting the data. Trees counted, plots revisited, canopy mapped.
- Reporting: turning the data into results someone else can read: hectares restored, survival rate, tonnes of carbon.
- Verification: an independent party checking that the results follow from the data.
In carbon projects, verification is formal: a third-party body audits the project against a standard before credits are issued. In most restoration projects, the funder plays that role.
What makes MRV digital?
The steps stay the same. What changes is how the data moves between them.
In a paper-based chain, field sheets become spreadsheets, spreadsheets become a report, and the report is checked by sampling. Each step is a copy, and each copy can drift from the one before.
In a digital chain, records are captured once, with their location, time and author, and every result in a report can be traced back to them. Remote sensing adds a second view over the whole area. Results can be refreshed when new data arrives instead of rebuilt for each report.
What data does dMRV use in a restoration project?
| Layer | Examples | What it cannot tell you on its own |
|---|---|---|
| Satellite imagery | Tree cover, land-cover change, deforestation alerts | Seedlings, species, what happened on the ground last week |
| Drone imagery | Canopy, crown counts, plot boundaries | Anything under the canopy; areas the drone did not fly |
| Field records | Plantings, survival counts, tree measurements, maintenance | Anything outside the plots someone visited |
| Photos | Geotagged evidence of work and condition | Whether the photo is from this site, unless its location is checked |
| Models | Allometric equations that turn tree size into biomass and carbon | Their own error, unless it is carried through |
Every layer has an error, and good dMRV carries those errors through instead of hiding them. A carbon developer we spoke with counts both GPS offset (about 3 metres) and tree-diameter measurement (about 5%) as error sources in its uncertainty model.
What data does an ARR carbon project have to keep?
For afforestation, reforestation and revegetation (ARR) projects under Verra's VM0047 methodology, version 1.1, the field side is specific:
- Sample plots. Area-based projects estimate biomass from field plots, measuring trees (diameter, and height where the allometric equation needs it) and scaling up by project area.
- A plot database. Every project plot and control plot has a unique ID, a location, a size and a layout.
- A planting census for dispersed planting, such as agroforestry: every planting unit counted at the start, then a sample revisited.
- A monitoring plan that names the sampling design, the field procedures, the quality checks and who handles the data. Records are kept electronically until at least two years after the last crediting period.
- A geolocation file of the project boundary, sent to the registry at registration and again when boundaries change. Verra asks for KML; our explainer on WKT vs GeoJSON covers the formats.
Uncertainty costs credits. VM0047 deducts for sampling uncertainty above a threshold, and expects measurement and boundary errors to be caught by quality checks and corrected before verification.
Is dMRV only for carbon credits?
No. The term grew up around carbon, where every tonne has to be verified before it is sold. But plenty of restoration organizations monitor for accounting and transparency, not credits: to show a funder what their money did, or to know which sites are struggling.
Those projects do not need carbon models or registry audits. They do need what makes the chain trustworthy: knowing which site a number came from, who recorded it, and when.
Where does dMRV usually break?
At the move from the physical world to the digital one. Bernat Fortet, who founded Restoration Scope, puts it as the step from atoms to bits: hundreds or thousands of people doing physical work, and every one of their records a chance for human error.
The usual failures are small and ordinary:
- Locations that are wrong or missing. Coordinates in the wrong UTM zone, latitude and longitude swapped, GPS points recorded before the phone settled. Our explainers on UTM coordinates and phone GPS accuracy cover the common cases.
- Photos that lost their evidence. Geotagged photos forwarded through a chat app arrive with no location. See what EXIF location data is.
- Sites and plots without stable IDs, so a revisit cannot be matched to the first visit.
- Shapes passed around as loose files. Boundaries in KML, shapefiles and spreadsheet cells, edited in one place and not the others. WKT vs GeoJSON explains the formats.
No satellite fixes these. A biomass model built on plots in the wrong place is precise and wrong.
How do you start with dMRV without buying a platform?
- Give every site, plot and planting a stable ID, and use it everywhere: field forms, spreadsheets, maps and reports.
- Record locations with their accuracy, and decide what is good enough before the season. Check your phones with a free GPS accuracy test.
- Keep photos with their metadata. Collect originals, not forwarded copies, and check them on a geotagged photo map before they go into a report.
- Keep one current version of each boundary, in a format other tools can read, such as GeoJSON.
- Write down how each reported number is calculated, so someone else could repeat it from the records.
That is most of the work, and it pays off whether or not you ever sell a credit. It is also the part we work on at Restoration Scope: the field records, photos and maps that everything else in the chain is built on. To compare tools for ARR projects, see our guide to ARR carbon monitoring software. For why we think operations come before reporting, see Reporting vs. Operations.
Frequently asked questions
What does MRV stand for?
Measurement (or monitoring), reporting and verification. It comes from climate policy, where results had to be measurable, reportable and verifiable, and now describes how any project shows what it achieved.
What is the difference between MRV and dMRV?
The steps are the same. dMRV does them with digital data, such as satellite and drone imagery, field apps and connected records, so results can be updated more often and traced back to their sources instead of being assembled by hand for each report.
Do you need dMRV if you don't sell carbon credits?
Not the carbon machinery. But the same habits help any project that reports to funders. Give sites and plots stable IDs, record locations with their accuracy, keep photos with their metadata, and keep records in one place.
What field data does an ARR carbon project need?
Under Verra's VM0047 methodology, area-based projects measure sample plots in the field, keep a plot database with a unique ID, location and size for every plot, and archive the data electronically. Census-based projects count every planting unit at the start, then sample. The project boundary goes to the registry as a geolocation file.
Can satellites replace field monitoring?
Not for young restoration. Satellites see canopy and land cover over large areas, but they cannot count seedlings under 1 metre, tell species apart reliably or check what happened on a specific planting day. Field data and remote sensing check each other.
