EUDR Methodology

Picterra is committed to supporting compliance with the EU Deforestation Regulation (EUDR) by providing robust, scalable tools to help organizations assess and minimize the risk of deforestation and forest degradation in their supply chains, while guiding their potential mitigation measures.

Methodology overview

Our deforestation risk assessment methodology is designed to categorize agricultural sourcing plots based on the risk of recent deforestation, using a transparent, reproducible, and spatially explicit process.

We assess the following EUDR-regulated commodities:

Picterra’s analysis combines authoritative global forest layers datasets, national forest inventories, as well as higher spatial resolution Earth Observation satellite imagery to unveil ambiguous cases where needed, to determine deforestation risk for each plot.

Picterra’s methodology builds on the convergence of evidence (FAO/GIZ WHISP “What’s in this plot”) approach, integrating, intersecting and analyzing multiple layers for undisturbed forest cover in 2020, plantation maps in 2020 and deforestation alerts in order to maximize information while enabling simple, traceable risk categorization.

Data conformity

Picterra Tracer ensures data accuracy through automated verification and correction processes.

Any issues detected are clearly flagged in the user interface and reflected in the attribute table, accompanied by detailed warnings and error indicators to support resolution. 

Depending on the severity of the issue, Picterra Tracer may prevent the creation of EUDR reports to ensure compliance.

Icon 

Category 

Conformity(EUDR)

Action

Info

Conform

No action is required. This message is for informational purposes only and indicates that an automatic correction has been applied to the plot.

Warning

Conform

No action is required for analysis, but Picterra recommends reviewing this plot with the supplier for verification.

Error

Not Conform

This plot can be analyzed, but the results may be unreliable or incompatible with methodology requirements (e.g., non-EUDR compliant). Picterra recommends verifying with the supplier.

Critical Error

Not conform

This plot cannot be analyzed due to errors that obstruct the analysis process.

Deforestation risk assessment

Risk classification approach 

Our approach to identifying deforestation risk is based on three key steps:

Authoritative mapping layers.

We use recognized global and national layers datasets that depict land cover, land use, and deforestation activity.

Note: For a detailed list of mapping layers used in this assessment, existing customers can refer to the technical documentation provided during onboarding or request it directly from their assigned Customer Success Contact Point

Plot-level risk classification.

These layers are combined and intersected with farm plot boundaries — using polygons for plots larger than 4 ha, or GPS coordinates for plots smaller than or equal to 4 ha — to assess each farm plot and assign a deforestation risk rating based on observed indicators.

Resolving uncertainty.

In cases of ambiguity—particularly for plots marked as high risk—we enable verification of the deforestation assessment using very high-resolution satellite imagery across time and customized machine learning models.

Plots are analyzed and assigned one of three deforestation risk levels related to the presence of a negligeable or non-negligeable deforestation risk according to EUDR:

High risk - non-negligeable risk

The plot shows strong signs of recent deforestation, either with large deforested areas or alerts confirmed by multiple data sources.

Low risk - negligeable risk

The plot contains a limited forest area, but there aren’t enough alerts to confirm recent deforestation. These areas should still be monitored over time, as they may be at risk in the future.

Very low risk - negligeable risk

The plot does not contain forest or is mostly made up of established plantations. These are considered safe for sourcing.

The assigned deforestation risk levels are automatically recorded in the attribute table under the designated attribute name, using one of the accepted values: high, low, or very low.

MethodologyAttribute nameAttribute values
EUDRDeforestation risk  high,    low,    very low

Deforestation risk analysis – workflow

The following schematic outlines a structured sequence of the classification logic and criteria used to evaluate deforestation risk:

Deforestation Analysis Methodology

Convergence of evidences approach (c.f. FAO WHISP methodology)

Global & regional data

Local detections

Global forest layers:

Different global layers of native and protected forest state in 2020.

Local plantation layer:

Country specific layers of forest plantations and agricultural plantations in 2020.

Global tree cover loss alerts:

Different sources of tree cover losses information from 2021 to now (near-real time).

Global plantation layers:

Different global layers of forest plantation and agricultural plantations state in 2020.

Proprietary plantation layer:

Picterra detection model discriminating tree plantation from native forest.

Satellite imagery:

10-30m: Sentinel-1/2, Landsat 5-8 3m: Planetscope 50-30cm: Skysat, Pléiades Néo; Worldview III

1. Global & regional data

Global forest layers:

Different global layers of native and protected forest state in 2020.

Global plantation layers:

Different global layers of forest plantation and agricultural plantations state in 2020.

2. Local detections

Local plantation layer:

Country specific layers of forest plantations and agricultural plantations in 2020.

Proprietary plantation layer:

Picterra detection model discriminating tree plantation from native forest.

3. Alerts

Global tree cover loss alerts:

Different sources of tree cover losses information from 2021 to now (near-real time).

Satellite imagery:

10-30m: Sentinel-1/2, Landsat 5-8 3m: Planetscope 50-30cm: Skysat, Pléiades Néo; Worldview III

The deforestation analysis starts by sorting farm plots into two main categories based on how the land is used:

  • Likely to contain forest: areas with a lot of dense forest cover.
  • Unlikely to contain forest: areas used for agriculture, mainly plantations, with only small patches of forest.

Then, each plot is assigned a risk level based on detected signs of deforestation, and indicated with the alerts.

Plot risk reclassification

Picterra Tracer supports per-plot deforestation risk reclassification, enabling users to override the automatically computed risk level to confirm or correct it ie in case the automated deforestation analysis produced ambiguous or false-positive results or editional evidence is available from the field audits. 

Reclassification can be applied following complimentary deforestation level 2 assessment.

Deforestation risk level 2 assessment

To conduct deforestation level 2 risk assessment, two types of evidence are accepted:

Certified verification evidence

When certification documents or field audits confirm that the actual deforestation risk differs from the initial Picterra Tracer assessment.

The results of Level 2 assessment must be manually entered into the attribute table under the designated attribute name, using one of the accepted values: ‘high’, ‘low’, or ‘very low’.

MethodologyAttribute nameAttribute values
EUDRDeforestation risk level 2   high,     low,     very low

 

Additionally, a spreadsheet can be uploaded containing the relevant plot IDs along with a column titled “Deforestation risk level 2 description”, where the supporting evidence for each reclassification can be documented.

Once a plot is assigned a “Deforestation risk level 2” value, Picterra Tracer will override the automatically computed risk with this manually provided classification.

This process ensures traceability and gives users the flexibility to validate exceptional cases without compromising consistency across large-scale datasets.

Level 2 assessment through remote sensing

Picterra offers a Deforestation risk level 2 analysis advanced assessment that leverages very high-resolution satellite imagery (50 cm or 30 cm) from two time points (prior to the cut-off date and the most recent), and machine learning deforestation models to precisely verify the deforestation status.

More information about this process can be provided by the Customer Success Contact Point assigned to the customer account.

Reclassification workflow 

There are two ways to execute per-plot deforestation risk reclassification once the Deforestation level 2 results are available:

  • Pre-analysis (pre-tagging)
  • Post-analysis (using “parent analysis flow” to avoid changing other results)

Pre- analysis

If you have verified evidence of the deforestation risk for a given plot (e.g. results from a ground audit), you can assign the “Deforestation risk level 2” attribute before running the analysis.

This value will then be applied directly to the plot’s “Deforestation risk” field, along with a note indicating that the classification was based on a Level 2 assessment — no further action is needed.

Post-analysis with parent analysis flow

If you have already completed an analysis and a plot was flagged as “high risk,” but you have since conducted a Level 2 deforestation risk assessment that justifies reclassifying the plot to “low” or “very low” risk, you can update the classification in two ways:

Update the “Deforestation risk level 2” attribute and use the “parent analysis” flow

This method allows you to reuse the results from the previous analysis while selectively overriding the risk level only for specific plots.

Why use the parent analysis flow?
Re-running an analysis may lead to changes in the results of other plots if the underlying methodology or global Earth observation layers have been updated.

The parent analysis flow avoids this by:

  • Copying all results from the original (parent) analysis without re-processing.
  • Overriding the “Deforestation risk” value only for plots that have a “Deforestation risk level 2” set.

This ensures stability of the original results while allowing selective updates based on verified evidence.

How to run a new analysis with a parent analysis:
To use the parent analysis option:

Note that in this mode, you can’t select plot ids anymore because this is going to reuse the plot ids from the parent analysis.

Submission to EU TRACES NT

Picterra Tracer supports export of risk assessments in formats compliant with EU TRACES NT platform specifications, including:

  • Point / MultiPoint (for plots ≤ 4ha using GPS coordinates)
  • Polygon / MultiPolygon (for plots > 4ha using geospatial boundaries)

The data structure follows the EU GeoJSON file schema (v1.4, Nov 2024):
See latest specification here

Protected areas

In the analysis result view, you will also be able to visualize the extent of protected areas.

Satellite map in Picterra Tracer showing deforestation alerts and protected areas for a cocoa supply chain analysis in Peru.

If one of your plots is within 1 km of a protected area, you should conduct an on-the-ground investigation to determine if the plot is actually inside a protected area.

The boundaries of protected areas come from the WDPA dataset.

Protected Area, Key Biodiversity Area, and Species data reproduced and incorporated under licence from the Integrated Biodiversity Assessment Tool (IBAT) (https://www.ibat-alliance.org/). IBAT is provided by BirdLife International, Conservation International, IUCN and UNEP-WCMC. Contact ibat@ibat-alliance.org for further information.