Satellite evidence for EUDR due diligence in southern Africa

EUDR compliance timeline and evidence

What the regulation actually requires

Before a relevant product is placed on the EU market or exported from it, the operator has to hold the geolocation of the plots where it was produced — a polygon for plots over four hectares, a point for smaller ones — and be able to show that those plots were not deforested or degraded after 31 December 2020. For timber and charcoal, that means a defensible statement about land cover on each plot across a period that is already history, frequently on land the reporting company never managed.

You cannot fly a drone into 2020. The archive of open satellite imagery — Sentinel and Landsat, stretching back well before the cut-off — is often the only evidence that reaches back far enough to answer the question the regulation asks.

Why the country of origin changes the job

The EU grades every producing country as low, standard or high risk, and the grade sets how much due diligence an operator must do.

South Africa, Eswatini and Lesotho are graded low risk. Operators still have to collect plot geolocation and supporting information, but a formal risk assessment is not required unless there is a risk of mixing with, or circumvention through, other origins.

Mozambique, Zimbabwe, Zambia, Namibia, Botswana, Malawi, Tanzania and Angola are graded standard risk. Full due diligence applies, including a risk assessment that has to conclude the risk is negligible. That assessment is where plot-level satellite evidence carries the most weight.

The grade follows the country where the wood was grown, not the port it left from. Material that crosses a border in the region before shipping carries its origin with it.

When the free screening says “review”

Free screening tools now exist, built on the EU’s own 2020 forest cover map and on FAO’s open tools. Use them: for most plots they are enough.

They struggle where public maps disagree with each other — savanna and woodland, bush-thickened rangeland, plantation mosaics, and small plots at the edge of a 10 m pixel. Those plots come back flagged for review, and someone then has to make the call.

That call is the service. TerraX re-reads the flagged plot from the imagery archive and writes down the reasoning, so the decision in your due-diligence file rests on evidence rather than on whichever map happened to be consulted.

What plot-level evidence looks like in practice

It begins with the plot geometry: the actual boundary of each source plot, as a coordinate polygon where the plot size requires one. Against that geometry, the imagery archive is read to characterise land cover through time — forest, plantation, cleared ground — at and around the 2020 cut-off and onwards. The deliverable is a per-plot evidence record: the geometry, the land-cover trajectory, and the imagery basis for each conclusion, in a form that slots into a due-diligence file.

Establishing a pre-2021 baseline from the archive

The pivotal question is the state of each plot on 31 December 2020. That is reconstructed from imagery captured before the cut-off — building a baseline of what was standing then, against which everything after is compared. Because the archive is dense, the baseline is not a single cloudy scene but a considered read across the surrounding seasons.

Trap one: a harvest cycle is not a conversion

This is where remote sensing earns its fee, and where a naive reading does real damage. A commercial plantation on a normal rotation shows up in a time series as dramatic, repeated loss of canopy — because that is what harvesting looks like from orbit. It is not deforestation. Genuine deforestation is a change of land use: forest or plantation becoming something that is no longer forest and is not replanted.

Confuse the two and the evidence fails in both directions. Read a harvest as a conversion and you flag compliant timber as non-compliant. Wave through a real conversion as "just a harvest" and you certify a breach. Telling them apart is a judgement about the pattern over time — cycle versus permanent change — and it is exactly the interpretation TerraX is accountable for.

There is a case between the two. For wood, converting natural or naturally regenerating forest into plantation after the cut-off counts as forest degradation, even though the land still carries trees. A time series has to be read for what replaced what, not only for whether canopy came back.

Trap two: was it forest at all?

The regulation defines forest by thresholds: more than half a hectare, trees taller than five metres, and canopy cover above ten percent, or trees able to reach those thresholds where they stand.

Across southern African savanna, bush-thickened rangeland often sits close to those lines. Whether a thinned block was forest on 31 December 2020 is a measurable question — canopy cover and height before the cut-off — not something to assert.

Measured carefully, it protects a legitimate product from being flagged. It also stops a genuine clearing being waved through as bush control.

Who this is for

• EU importers and operators sourcing wood or charcoal from southern Africa, particularly from standard-risk countries.
• Certification bodies and auditors who need an independent read on plots a screening tool has flagged.
• South African traders and processors handling material grown elsewhere in the region.
• Plantation companies that want harvest cycles documented as harvest cycles before someone reads them as loss.

Where this sits alongside FSC and PEFC

Satellite evidence does not replace chain-of-custody certification. FSC and PEFC track the wood through the supply chain; the EUDR evidence layer establishes the land-use history of the plots the wood came from. They answer different questions and belong side by side in the same due-diligence file.