The EU Deforestation Regulation is written as a duty on the operator who places goods on the EU market. But the evidence it demands lives on the ground where the commodity was grown — and for most exporters that ground is not one estate they control, but hundreds or thousands of supplier plots they have never set foot on. The hard part of EUDR was never checking one plot carefully. It's standing behind a whole supply base, at scale, without the budget to walk any of it.
The obligation flows downhill
The legal duty sits with the EU importer. But it cannot be discharged without a geolocated, deforestation-free record for every plot that fed the consignment — and that requirement flows straight back down the chain. A South African timber, pulp or agricultural exporter selling into Europe will be asked, as a condition of sale, to produce plot geolocation and deforestation-free evidence for its entire supply base, out-growers and third-party suppliers included. You end up carrying the evidentiary burden for land you don't own and can't inspect.
As of October 2026, EUDR due-diligence obligations apply from 30 December 2026 for large and medium operators and 30 June 2027 for micro and small enterprises, with the deforestation cut-off fixed at 31 December 2020. (The timeline has been amended before — confirm against the current EU text, Regulation (EU) 2023/1115, before you rely on it.) The practical effect is the same either way: well before the deadline, your buyers will want the supply-base evidence in hand.
Why plot-by-plot fieldwork doesn't scale
A supply base of hundreds or thousands of plots cannot be verified by sending someone to each one. The cost and the calendar rule it out — and in any case, most of what EUDR asks is a question about the past: was this ground forest on 31 December 2020, and has it been cleared since? No present-day site visit can answer that. The only record that reaches back to the cut-off, across every plot at once, is the open satellite archive, which has imaged most of the planet every few days for a decade. That makes supply-base verification a remote-sensing problem before it is ever a fieldwork problem.
Screen first, scrutinise what's flagged
The workable method is risk-based triage — the same logic a good auditor uses when there is too much to check by hand. Run every plot in the base against the satellite archive from the 2020 baseline forward. The large majority — stable forest or plantation on a normal rotation — clear quickly: no post-2020 conversion, nothing to escalate. A minority show a change that matters: a step loss with no replanting, or a conversion to cropland, bare ground or built surface inside the plot boundary. Those are the plots you pull out for close reading. Instead of spreading thin attention across the whole base, you concentrate expensive scrutiny — and any fieldwork — exactly where the signal says it's needed.
A screen done properly returns, for every plot in the base:
- a clean-or-flagged status against the 31 December 2020 cut-off;
- for flagged plots, the date and nature of the change — enough to tell a harvest cycle from a genuine conversion;
- the pre-2021 baseline attached to each plot, so the record is anchored to what was there before the cut-off, not asserted after the fact;
- the plots where cloud, small size or boundary error leave the answer genuinely uncertain — named as such, not quietly guessed.
The triage is only as good as two things underneath it
A supply-base screen rests on two foundations, and a weakness in either turns the whole exercise into false comfort or false alarm.
The first is geometry. Every clean-or-flagged result is a spatial query run inside a plot boundary; if the boundary is offset, in the wrong reference system, or sweeping in roads and unplanted bush, the screen is reading the wrong ground. A clean result on a mislocated polygon is reassurance about the field next door; a flag on one is a false alarm you'll waste real money chasing. Verifying the boundaries comes before screening them, not after — it's the geolocation half of the problem.
The second is the harvest-versus-conversion distinction, applied consistently across the base. Commercial plantations on a normal rotation drop their canopy dramatically and routinely — and to a naive change-detector, every felling looks like deforestation. Screen a timber supply base without accounting for that, and you drown in false positives from land that was never converted at all. Telling the sawtooth of a harvest cycle from the step change of a true conversion is what keeps the flagged list short enough to act on.
What screening is, and isn't
Done well, satellite screening is the thing that makes an EUDR supply base tractable: it turns an impossible plot-by-plot audit into a short, ranked list of plots that actually warrant attention, each one anchored to a verifiable baseline. What it is not is a substitute for the rest of the file. It reduces fieldwork sharply; it does not abolish it — some flagged plots will still need a ground check to resolve. And it speaks only to the land-cover question. The legal-ownership, supply-chain and benefit-sharing records the due-diligence statement also requires are a separate exercise that no satellite can collect.
Where the line sits
TerraX produces and signs the spatial evidence — a screened supply base with each plot's status, the change date and nature where flagged, and the pre-2021 baseline attached, with its uncertainty stated. It does not issue an EUDR determination, collect the legal and supply-chain records the statement also needs, or guarantee acceptance by a competent authority. That determination, and its consequences, remain yours. What we hand over is a triaged, evidenced supply base a due-diligence file can be built on — and a much shorter list of plots to worry about.
The supply base is where EUDR stops being one careful check and becomes a problem of scale. Screening it against the archive is how you find the few plots that matter before the deadline finds them for you. It's part of our EUDR forestry compliance service — if you're facing a supplier base you have to stand behind, tell us roughly how many plots and where, and we'll start by screening it.