How Copernicus Satellite Data Powers EUDR Due Diligence

What is the Copernicus programme?

Copernicus is the European Union's Earth observation programme, coordinated by the European Commission and implemented in partnership with the European Space Agency (ESA), EU Member States, and EU agencies. It is the largest Earth observation programme in the world, providing continuous, free, and open data about our planet from a constellation of dedicated satellites and contributing missions.

For EUDR compliance, Copernicus is particularly significant because the European Commission has explicitly referenced it as a data source for verifying deforestation-free supply chains. The programme's Sentinel satellite family provides the imagery needed to detect land-use changes at the scale required by the regulation — covering every production plot, in every country, with consistent and repeatable measurements.

Unlike commercial satellite providers that charge per image or per square kilometre, Copernicus data is freely available under an open licence. This means any operator, compliance platform, or competent authority can access the same imagery — creating a level playing field for verification.

Sentinel-2: the workhorse for vegetation monitoring

The Sentinel-2 mission consists of two identical satellites — Sentinel-2A and Sentinel-2B — orbiting the Earth in a sun-synchronous orbit at 786 km altitude. Together, they provide global coverage with a revisit time of approximately 5 days at the equator. Here are the specifications that matter for EUDR due diligence:

• Spatial resolution: 10 metres in four key bands (blue, green, red, and near-infrared). This means each pixel in the image represents a 10m × 10m area on the ground — sufficient to detect deforestation at the plot level, even for smallholder farms.
• 13 spectral bands: Sentinel-2 captures light across 13 spectral bands, from visible light (443 nm) through near-infrared (865 nm) to shortwave infrared (2190 nm). The near-infrared and shortwave infrared bands are particularly valuable for vegetation analysis because healthy vegetation reflects strongly in these wavelengths.
• 5-day revisit time: With both satellites operating, any point on Earth is imaged every 5 days. This frequent revisit enables time-series analysis and helps overcome cloud cover — a persistent challenge in tropical regions where EUDR-regulated commodities are produced.
• Radiometric resolution: 12-bit quantisation provides 4,096 intensity levels per band, enabling detection of subtle changes in vegetation health that might indicate early-stage degradation before full deforestation occurs.
• Level-2A products: ESA provides atmospherically corrected surface reflectance data (Level-2A), which removes the effects of atmospheric scattering and absorption. This is essential for comparing images taken on different dates under different atmospheric conditions.

NDVI explained

The Normalized Difference Vegetation Index (NDVI) is the primary metric used to assess vegetation health from satellite imagery. It exploits a fundamental property of plant biology: healthy green vegetation absorbs most visible red light for photosynthesis while strongly reflecting near-infrared (NIR) light. Bare soil, water, and built-up areas do not show this pattern.

The formula is straightforward:

NDVI = (NIR − Red) / (NIR + Red)

NDVI values range from −1 to +1:

• 0.6 to 0.9 — Dense, healthy forest canopy. This is what you expect to see on a plot that has not been deforested.
• 0.3 to 0.6 — Moderate vegetation, such as cropland, grassland, or degraded forest.
• 0.1 to 0.3 — Sparse vegetation or bare soil.
• Below 0.1 — Water, rock, sand, or built-up areas.

A significant drop in NDVI between the baseline period and the current period is a strong indicator of deforestation or forest degradation. For example, if a plot showed NDVI values of 0.75 in mid-2020 but now shows 0.25, that suggests the forest cover has been removed and replaced with bare soil or low vegetation — exactly the kind of change the EUDR is designed to detect.

In addition to NDVI, the Normalized Difference Moisture Index (NDMI) uses the shortwave infrared band to detect changes in vegetation water content. NDMI is particularly useful for identifying forest degradation — situations where the canopy is thinned or stressed but not completely removed.

How change detection works

EUDR compliance verification is fundamentally a change detection problem: has the land cover on a specific plot changed between the cutoff date (31 December 2020) and the present? Here is how satellite-based change detection works in practice:

Establishing the 2020 baseline

The first step is creating a reliable baseline image of each production plot from around the cutoff date. Rather than using a single satellite image from 31 December 2020 (which might be cloudy or affected by seasonal variation), analysts typically create a composite image from multiple acquisitions during a defined baseline window — commonly June to August 2020 for tropical regions, when cloud cover tends to be lower.

The composite is built by selecting the best (least cloudy) pixel from each available image in the window, producing a cloud-free representation of the land surface during the baseline period. NDVI and NDMI values are calculated from this composite to establish the baseline vegetation state.

Current-period analysis

The same compositing process is applied to recent imagery — typically the most recent 3–6 months — to create a current-state representation of each plot. This accounts for seasonal variation and ensures that temporary changes (such as deciduous leaf drop) are not mistaken for deforestation.

Differencing and thresholding

The change detection algorithm compares the baseline and current NDVI/NDMI values for each pixel within the plot boundary. A significant negative change (drop in vegetation index) that exceeds a defined threshold is flagged as potential deforestation or degradation. The threshold is calibrated to balance sensitivity (detecting real changes) against specificity (avoiding false alarms from natural variation).

Cloud masking and seasonal compositing

Tropical regions — where most EUDR-regulated commodities are produced — present a major challenge for optical satellite monitoring: persistent cloud cover. In equatorial Africa, Southeast Asia, and parts of South America, cloud cover can exceed 80% during the wet season, making individual satellite images unusable.

Two techniques address this:

• Cloud masking: Sentinel-2 Level-2A products include a Scene Classification Layer (SCL) that identifies each pixel as cloud, cloud shadow, water, vegetation, bare soil, or other categories. Cloud-masked pixels are excluded from analysis, ensuring that only clear observations contribute to the vegetation indices.
• Seasonal compositing: By aggregating multiple images over a defined time window (e.g., 3–6 months), the compositing algorithm can find at least one cloud-free observation for most pixels, even in persistently cloudy regions. Median or best-pixel compositing methods are commonly used.

The combination of Sentinel-2's 5-day revisit time and seasonal compositing means that even in the cloudiest tropical regions, reliable vegetation assessments can be produced for most production plots. In rare cases where cloud cover is truly impenetrable for extended periods, supplementary data sources (such as Sentinel-1 radar, which penetrates clouds) can fill the gaps.

Commodity-specific thresholds

Not all land-use changes look the same from space. Converting dense tropical forest to a cattle ranch produces a very different spectral signature than converting secondary forest to a coffee plantation under shade trees. Effective EUDR monitoring systems use commodity-specific thresholds to account for these differences:

• Cattle and soya: These commodities typically involve complete forest clearance, producing large, abrupt drops in NDVI. Detection thresholds can be relatively aggressive because the signal is strong and unambiguous.
• Palm oil: Oil palm plantations have a distinctive spectral signature — they are green and photosynthetically active, but their canopy structure differs from natural forest. Detection relies on both NDVI magnitude and texture analysis to distinguish plantations from native forest.
• Coffee and cocoa: These crops are often grown under shade trees, meaning the canopy may still appear partially forested from space. Detection requires more sensitive thresholds and may incorporate NDMI (moisture index) to identify changes in canopy density and structure.
• Wood and rubber: Timber harvesting and rubber plantation establishment can range from selective logging (subtle canopy changes) to clear-cutting (dramatic changes). Multi-temporal analysis helps distinguish permanent land-use conversion from temporary disturbance followed by regrowth.

From satellite data to compliance verdict

The final step in the satellite analysis pipeline is translating raw change detection results into an actionable compliance verdict for each production plot. A typical classification system assigns one of three risk levels:

• LOW risk: No significant vegetation change detected between the baseline and current period. The plot appears to have maintained its forest cover (or was already non-forested before the cutoff date). This supports a finding that the product is deforestation-free.
• MEDIUM risk: Some vegetation change detected, but the signal is ambiguous — it could be natural variation, seasonal effects, or minor disturbance rather than deforestation. Further investigation is recommended, which may include higher-resolution imagery, field verification, or supplier engagement.
• HIGH risk: Significant vegetation loss detected that is consistent with deforestation or forest degradation after the cutoff date. This triggers enhanced due diligence requirements and may require the operator to seek alternative sourcing or obtain additional evidence that the change was not deforestation.

The satellite analysis report — including the baseline and current imagery, NDVI/NDMI values, change maps, and risk classification — becomes part of the operator's due diligence documentation. It provides the auditable, evidence-based foundation that competent authorities expect when reviewing EUDR compliance.

By combining freely available Copernicus data with automated analysis pipelines, operators can now verify deforestation-free status across thousands of production plots in a matter of hours — a task that would have been impossible through field visits alone. This is the technology that makes EUDR compliance at scale not just feasible, but practical.