Carbon report methodology
In force since 15 October 2026
This page gathers what every BioAurea carbon report cites in a single line: what it measures, where the range comes from, how roots and soil are counted, and how a figure can be traced. It applies to the three reports of the Carbon Explorer: by vereda, by coordinate and by polygon. The reports themselves are currently issued in Spanish.
What a BioAurea report is not: it is not a field measurement nor a certification; it does not deduct permanence buffers nor assess additionality. Carbon pools and factors follow the IPCC 2006 and 2019 guidelines. It is not a validation or registration under any carbon standard.
1. What each report measures
| Report | What the number is |
|---|---|
| Vereda | Average of the whole vereda over a 500 m grid, calibrated with the National Forest Inventory. For a specific plot, query by coordinate or polygon. |
| Coordinate | Live estimate of the surroundings of the coordinate, a circle of 250 m radius (about 20 ha); it is not the vereda average. The 10 m view only distributes that value inside the circle; it does not change it. |
| Polygon | Average of the veredas the polygon crosses, weighted by the area it covers in each one. It is not an estimate of the polygon itself. |
In the coordinate report, tonnes are only published if the person querying gives the area of their plot: the 250 m circle is the surroundings being evaluated, not the plot. Without an area, the report gives densities per hectare.
2. The four carbon pools
- Above-ground biomass: trunks, branches and leaves.
- Roots: below-ground biomass, derived from the above-ground biomass (see section 4).
- Soil: organic carbon in the top 30 cm.
- Dead organic matter: fallen wood, branches and fine debris, and standing dead trees.
Total: above-ground biomass + roots + soil (0 to 30 cm) + dead organic matter. The report gives the carbon in vegetation (above-ground biomass + roots) first, then soil, then the total, because each has a different error. Carbon is converted to CO2 equivalent by multiplying by 3.667 (44/12, the ratio between the masses of CO2 and carbon). Above-ground biomass, soil and dead organic matter are estimated separately and calibrated with field plots that were not used to train the model.
3. The 90 % range
The range was calibrated with National Forest Inventory field plots the model did not see: there, 90 out of 100 measured values fell inside a range like this one. In very dense mountain forest and outside Colombia the real coverage is lower.
The margin comes from the model error on inventory plots it was not trained on; it is fixed per carbon pool and value level and is not recomputed per query. The method is split conformal prediction, 90 %. If a pool has no valid calibration, the report says the range is not available; it never replaces it with an approximate margin.
The margin of the total is not the sum of the margins: biomass and roots are added because they move together; soil and dead organic matter are combined in quadrature because their errors are independent. That is why the total is proportionally narrower than each pool on its own.
A wide range means the model has little information on that terrain. To commit volume, use the low end. The model ranks plots better than it measures a single one.
In the vereda report the biomass margin is the general inventory one, not the forest or non forest one. In the polygon report the vereda margin is applied to the average; it is not a range of the polygon itself.
The range is the error of today's estimate. It is not a permanence buffer: that buffer covers the future risk of fire, logging or pests, it is a design decision of each project and the report does not deduct it.
4. Roots
Roots are computed from above-ground biomass with the Cairns relation adjusted to the Colombian inventory. It is not a fixed share: it decreases as biomass increases, so it changes from one plot to another. Each report states the percentage of above-ground biomass applied in that case.
5. Soil
In Colombia the soil value is expressed on the scale of the National Forest Inventory. That is the scale of the modern inventory plots; legacy soil profiles (IGAC, SISLAC) were measured with another protocol and give different values for the same site, so they should not be compared directly with the report figure. Outside Colombia the direct model output is served, without national calibration, and the margin is larger.
The historical soil series is not publishable yet: until it is regenerated with the full set of variables, the vereda report chart shows above-ground biomass only.
6. Vereda time series
The chart shows above-ground biomass, one value per year: the last available date of each year, not an average, so every point can be checked against the report for that date. A vertical line marks the engine change: before it, the series was recomputed with the previous engine; from it on, it is the current engine.
7. How a vereda compares
The vereda report gives its percentile of carbon in vegetation among the veredas of its department and municipality, and the department mean. The percentile is computed on the latest published snapshot of the vereda map (not on the report date) and is omitted when the area has fewer than ten veredas. Means are not weighted by area: every vereda weighs the same.
The vegetation class is the national quintile of above-ground biomass among the veredas of the country; it is a BioAurea convention, not a standard. The thresholds travel in the API response, together with the date of the snapshot they were computed on.
8. Polygon
The area is the real geometry of the vertices. The carbon is the average of the veredas the polygon crosses, weighted by the surface in each one (a grid of points inside the polygon, each assigned to its vereda), from the latest published snapshot. It is not a satellite extraction of the polygon.
9. Date comparison at a point
Each date is an independent estimate with its own margin. A change is only declared when the difference exceeds the combined 90 % margin; otherwise the difference is not conclusive. The combined margin is the square root of the sum of the squares of both margins. The reading of the cycle phase (growth, mortality, emission) is an interpretation, not a measurement, and is folded away.
10. Traceability
- Identifier: every report carries one with the BA prefix and ten characters. It is deterministic: the same query (place, effective date and model version) returns the same identifier. It lets a report be cited in a claim.
- Effective date: the observation closest to the requested date; the deviation in days is stated.
- Engine revision: the date of the last change to a model or a calibration. If a figure changes because a model changed, this date changes too.