Geometry first, because it is the part that is exact
Area, centroid and county come out of your own polygon on an equal-area projection. There is no sampling and no model in that step. So it is the one figure in the whole service that carries no caveat. It is also the figure that most often disagrees with the customer’s own records. That disagreement is usually where a study starts earning its fee.
Forest loss, and the floor underneath it
For each plot we report whether tree cover has been lost since the EUDR cut-off date of 31 December 2020, and in which year. The product behind it is public, annual, and pinned to a specific version in the output. A buyer can re-run the same question years later and get the same answer.
The floor matters more than the verdict. A single isolated pixel is the dominant false-positive mode of any annual forest-change product. A detection has to be supported by more than one pixel, which puts our stated floor at 0.5 hectares. Anything smaller is flagged as marginal rather than reported as loss. Every verdict is printed with that floor attached. “No loss detected” is evidence of absence only down to the smallest thing the instrument can see.
What a detection does not tell you
It does not tell you what was cut. Every public forest-change product answers the question “did tree cover disappear here”. A farmer clearing a woodlot they planted, a cooperative replanting coffee, and somebody taking down primary forest all produce the same signal. So every detection we return says cause-unverified, and the next step is a visit, never a stronger claim. This is the part of EUDR screening that nobody can currently sell. We would rather say so than let a buyer discover it during an audit.
Cropland, with its dates
The cropland baseline tells you what the ground looked like in 2019 and 2020. It cannot tell you what happened in 2021 or since. It does not recognise perennial crops, so a mature coffee or tea plot will read as not cropland. Both limitations are printed with the answer. Used correctly, the baseline is a way of asking whether a plot was already under cultivation before the cut-off. That is one of the questions EUDR poses.
Minimum plot size
A 10 metre pixel covers 100 square metres. Strip one pixel of contamination along each edge, as you must, and a very small field has almost nothing left in the middle. We set a minimum plot size when scoping and price against it. We would rather do that than return a figure for a 0.25-hectare plot that is mostly hedge.