Sentinel-2 and satellite monitoring: how the EU checks an EUDR statement

The EUDR requires a due diligence statement, but its enforcement also relies on a means of verification independent of operators' statements alone: satellite observation. This guide explains how this monitoring works in practice, without going into the technical aspects of image analysis, which remain the remit of the authorities.

Sentinel-2, the programme behind the check

Sentinel-2 is part of the European Copernicus programme, a set of Earth-observation satellites whose images are public and free. Two twin satellites cover all land surfaces with a repeat pass roughly every five days, at a resolution of 10 metres for the finest spectral bands — enough to detect a change in forest cover over an area of a few hundred square metres, but not to distinguish an individual tree.

What the check actually cross-checks

The competent authority does not settle for an isolated image: it compares the state of the declared plot at the cut-off date of 31 December 2020 with its state at a more recent date, using the same geographical coordinates as those supplied in the statement. A discrepancy between the declared use (intact forest, pre-existing agricultural plot) and what the time series of images shows triggers a more thorough examination, not automatically a penalty — the satellite image is an indication, not irrefutable proof on its own, particularly in the case of cloud cover or ambiguous vegetation cover.

What an operator can check for itself, beforehand

Before filing a statement, you can consult the Sentinel-2 imagery history of a plot free of charge through several public viewers (the official Copernicus Browser, or third-party tools that republish the same data). This prior check lets you spot a potential inconsistency yourself — a plot whose statement asserts continuous use while the image shows recent clearing — before an inspection reveals it.

The limits of the method

• The resolution does not make it possible to detect partial forest degradation (selective felling of trees without complete deforestation) as easily as outright deforestation.
• Cloud cover can delay the availability of a usable image over some tropical areas, where the repeat pass every five days does not guarantee a clear image in each cycle.
• Distinguishing between types of vegetation (natural forest vs. an agricultural plantation already established before 2020) requires finer analysis than a simple comparison of green cover, and relies on additional data layers (national forest land registry, land-use history).

Why the precision of the initial record matters more than it seems

A coordinate offset by a few hundred metres from the actual plot can, in a satellite cross-check, match the statement to a different neighbouring area from the one actually farmed — with a risk of a false positive for non-compliance that has nothing to do with actual practice on the ground. A GPS reading taken directly on the plot rather than estimated from a map reduces this risk of offset.

Common mistakes

• Thinking the satellite image is infallible — it remains an indication cross-checked with other sources, not absolute and definitive proof on its own.
• Never checking your own plot before filing, when the viewing tool is public and free.
• Recording an approximate coordinate in the belief that the margin of error has no bearing on a remote check.

Checklist

1. Consult the plot's Sentinel-2 history before filing a statement.
2. Record the coordinates directly in the field, not from an approximate map.
3. Keep a record of this prior check, useful if you are later asked for justification.
4. Do not be alarmed by an isolated discrepancy — it triggers an examination, not automatically a penalty.