VM0047 lets an afforestation, reforestation or revegetation project measure its progress against a dynamic benchmark. The benchmark is built from a remote-sensing indicator called the stocking index, or SI. If that index is weak, the whole mechanism can fail quietly: the project grows trees, and the paperwork says nothing happened.
This article collects what the public documents say about the SI. It also marks the places where they say nothing.
The stocking index in numbers
30 m
coarsest pixel Verra's provider form accepts for the index (30 m or finer)
Annual
minimum cadence across baseline and monitoring years
14 May 2025
date VM0047 v1.1 became active
What the SI is
According to a published webinar summary with Verra staff, a stocking index is any indicator sufficiently correlated with above-ground biomass. It is not a particular product. Vegetation indices, canopy height layers and modelled biomass layers can all qualify in principle.
Version 1.1, active since 14 May 2025, changed the limiting variable of the area-based approach from project size to planting density. Plots with more than 50 trees per hectare are in scope, according to the same webinar summary.
What the methodology text requires
The v1.0 methodology is concrete about process:
- The monitoring plan must give a full description of the stocking index and how it is derived. A pointer to a database is not enough. - A database must list project plots with unique IDs, locations, sizes and configuration, plus the SI time series from the start to the present. Control plots must reference the project plot they are matched to. - Control plots are matched to project plots on continuous covariates that include, at minimum, the historical and the initial SI. - The benchmark compares the SI slopes of project and control plots. Significance of the difference is tested with a Z test that uses the standard errors of the two slopes. If the control slope is not significant (P above 0.05) or is negative, it is set to zero in the benchmark equation.
A rating-agency review notes that, in practice, matching rests on essentially one factor, the SI, at three time points before the project starts, with no apparent handling of uncertainty in the matching step.
What Verra's provider form adds
Verra publishes an application form for data service providers that supply the SI. It is the most numeric document in the set:
| Requirement | Value |
|---|---|
| Spatial resolution | Pixel size of 30 m or finer, for all years provided |
| Cadence | At least annual, covering baseline and monitoring years |
| Season | Optical imagery limited to a consistent seasonal window each year |
| Evidence | Demonstrated correlation of the SI with above-ground biomass |
Verra also states that a provider completing vetting does not guarantee that a project or its data will be accepted. Projects and data remain independently reviewed by Verra and the validation and verification body.
Why a noisy index costs credits
A data provider (Space Intelligence) explains the mechanism plainly. If the SI is too noisy to separate project and control trends, the two slopes are statistically indistinguishable. The baseline is effectively set to 1, and a well-run project can claim zero removals.
The resolution arithmetic shows why small plots are exposed. In the webinar summary with Verra staff, a data provider notes that a 0.25 ha plot (50 m by 50 m) holds 25 pixels at 10 m but only 4 at 25 m. With so few pixels per plot, there is little averaging to dampen noise. Noise in the SI erodes the one signal the benchmark depends on.
Vendors disagree on which type of index is best. One provider argues that vegetation indices saturate in dense canopies and vary with season and atmosphere, that canopy height captures height but not density, and that modelled biomass captures both. Treat this as a practitioner view, not a rule. VM0047 does not rank the types.
What the public documents do not say
In the excerpts reviewed for this article, the following are not stated:
- a numeric correlation threshold between the SI and biomass; - a required calibration area, such as an ecoregion or buffer; - a minimum number of project or control plots, or a minimum plot size for the SI; - an uncertainty deduction schedule that is specific to the SI.
Absence from the excerpts is not proof of absence from the full v1.1 text, so check the current document. It does mean that "meets the threshold" is not a claim anyone can make today. The practical test is a documented correlation, using local field plots or published evidence, and a derivation a verifier can follow.
What this means in practice
1. Write the SI derivation as a methods annex: inputs, season, version, training data and calibration records. 2. Keep an unbroken annual series from before the start date, because the matching step needs history. 3. Choose a layer whose resolution fits your plots. Verra's floor is 30 m, but small plots favour finer pixels. 4. Quantify the noise before the project starts, for example by checking how stable the SI is on stable control land. 5. Remember that a provider listing is not an approval of your project's data.
For the choice between the two Verra methodologies, see VM0042 versus VM0047. For the other measurement route, see the VM0047 census track for coffee agroforestry.
Frequently Asked Questions
Not in the public documents reviewed. The provider form asks for demonstrated correlation but gives no number. Check the current v1.1 text.
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.



