← Back to workspace
Evidence bundle

CPO-0024Ahou Kobena

Gnamangui, Côte d’Ivoire · 6.80 ha · cocoa · 6.776000, -6.553000

Deforestation-free
85% confidence · 1.3 s · €0.0008
Deforestation-free

Plot CPO-0024 (Ahou Kobena, Gnamangui, CI) covers 6.80 ha under cocoa. This land was already in agricultural use before the EUDR cut-off: less than the 0.5 ha FAO forest minimum was mapped as forest on 31 December 2020 (0.00 ha). Nothing was deforested to produce this commodity, which is the situation of the overwhelming majority of smallholder plots.

What the satellites saw

Imagery evidence

Rendered server-side from the same pixels the verdict was computed from. Deliberately not smoothed: blocky pixels are honest about what a 10-metre sensor can resolve. The white outline is the boundary the cooperative submitted.

Green: mapped as forest on 31 Dec 2020. Red: tree-cover loss after the cut-off. Amber: loss before the cut-off, which EUDR does not penalise. White outline: the declared plot.
Forest baseline & detected loss

Green: mapped as forest on 31 Dec 2020. Red: tree-cover loss after the cut-off. Amber: loss before the cut-off, which EUDR does not penalise. White outline: the declared plot.

JRC Global Forest Cover 2020 + Global Forest Change (Hansen/UMD)

Cloud-free median of 1 Sentinel-2 acquisition from the period around 31 December 2020 — the state the law measures against.
Around the cut-off — 2021-04-12

Cloud-free median of 1 Sentinel-2 acquisition from the period around 31 December 2020 — the state the law measures against.

Sentinel-2 L2A (surface reflectance) — median of 1 scene (S2A_29NQH_20210412_0_L2A)

Cloud-free median of 1 recent Sentinel-2 acquisition over the same ground.
Today — 2026-03-12

Cloud-free median of 1 recent Sentinel-2 acquisition over the same ground.

Sentinel-2 L2A (surface reflectance) — median of 1 scene (S2B_29NQH_20260312_0_L2A)

Convergence of evidence

Three independent datasets

Each line is reported on its own terms. Where they disagree, the disagreement is shown rather than resolved by quietly preferring one source - because an auditor citing the other source is exactly the situation this has to survive.

ClearForest cover at the EUDR cut-off (31 Dec 2020)379 ms

Less than the 0.5 ha FAO minimum was mapped as forest on the cut-off date (0.00 ha). This land was already in agricultural use before EUDR applied.

Baseline forest (ha)0
Baseline forest (%)0
Plot area measured (ha)6.825
Grid resolution (m)10
Pixels sampled840

JRC Global Forest Cover 2020 · V4 (2026 release) · European Commission — Joint Research Centre · 10 m · CC BY 4.0

ClearAnnual tree-cover loss after the cut-off87 ms

No tree-cover loss was detected on baseline forest inside this plot after 31 December 2020.

Post-cut-off loss on forest (ha)0
Total loss recorded, all years (ha)0.344
Years with lossnone
Detection floor (ha)0.09

Global Forest Change (Hansen/UMD) · GFC-2024-v1.12 · University of Maryland GLAD / Google · 30 m · CC BY 4.0

ClearHigh-resolution canopy structure change since the cut-off781 ms

Canopy structure inside the plot is stable since the cut-off (NDMI +0.010, NDVI -0.009).

Canopy moisture change (NDMI)0.01
Greenness change (NDVI)-0.0092
Baseline scenes stacked1
Baseline clear coverage (%)100
Current scenes stacked1
Current clear coverage (%)100
Plot pixels compared696
Probative for deforestationno — land already agricultural at cut-off

Sentinel-2 L2A (surface reflectance) · Collection 1, AWS Open Data · ESA Copernicus / Element 84 Earth Search · 10 m · Copernicus open licence

Article 10(4)

Every criterion assessed and recorded

Article 10(4) requires the operator to demonstrate how the information was checked against each criterion and how the degree of risk was determined. That means recording the criteria that came back clear, not only the ones that fired.

ArticleCriterionOutcome
10(2)(a)
Country risk classification
Côte d'Ivoire: standard risk tier — simplified due diligence not available.
triggered
10(2)(b)
Presence of forests in the country of production
Côte d'Ivoire retains significant forest cover, so conversion risk is material and the geolocation analysis carries real weight.
triggered
10(2)(c)
Presence of Indigenous Peoples
No recognised Indigenous territories are recorded for Côte d'Ivoire in our country profile.
n/a
10(2)(d)
Consultation with Indigenous Peoples
Criterion (c) does not apply, so (d) does not arise.
n/a
10(2)(e)
Claims by Indigenous Peoples over the land
Criterion (c) does not apply, so (e) does not arise.
n/a
10(2)(f)
Prevalence of deforestation in the country
severe deforestation pressure recorded for Côte d'Ivoire.
triggered
10(2)(g)
Source and reliability of the information
4 unresolved issue(s) on the submitted geolocation.
triggered
10(2)(h)
Governance, corruption and enforcement concerns
CPI 45/100 for Côte d'Ivoire does not indicate elevated falsification risk.
cleared
10(2)(i)
Complexity of the supply chain
Direct sourcing with full plot-level traceability.
cleared
10(2)(j)
Risk of circumvention or mixing
Declared volume is consistent with the mapped area and no duplicate geometries were found.
cleared
10(2)(k)
Conclusions of Commission expert groups
No expert-group conclusion published in the Commission register currently bears on this commodity and origin in a way that changes the assessment.
cleared
10(2)(l)
Substantiated concerns and non-compliance history
No substantiated concerns under Article 31 and no recorded prior infringement in this chain.
cleared
10(2)(m)
Any other information pointing to a risk
No further adverse information identified; land tenure evidence is on file.
cleared
10(2)(n)
Complementary certification information
Covered by Fairtrade. Treated as complementary evidence only: Article 10(2)(n) admits it solely where it meets the Article 9 information requirements, and certification is never a substitute for geolocation.
cleared
Composite risk
37/ 100 · standard

Only “negligible” (below 10) clears Article 10(1). Anything above it means Article 11 mitigation is owed before the product may be placed on the market.

Findings
10(2)(g)4 unresolved data-quality issues on the geolocation+30

Article 10(2)(g) turns on "the source, reliability, validity, and links to other available documentation" of the Article 9 information. Outstanding: Coordinates carry only 3 decimal digits; Polygon required for a 6.80 ha plot; Coordinates carry only 3 decimal digits; Polygon required for a 6.80 ha plot.

→ Resolve every blocker and error before the statement is submitted.

10(2)(f)Severe national deforestation pressure+4

Côte d'Ivoire is experiencing severe rates of forest loss, much of it driven by the expansion of exactly this commodity. That raises the base rate of non-compliance for any plot in the country, independent of what this particular plot shows.

→ Maintain continuous monitoring rather than a one-time check.

10(2)(a)Côte d'Ivoire is standard risk — full due diligence applies+3

Côte d'Ivoire is not on the low-risk Annex to Reg. 2025/1093, so it defaults to standard risk. The Article 13 simplified route is unavailable and the full Article 10 and 11 obligations apply. Competent authorities must check 3% of operators sourcing from here annually.

→ Complete Article 10 assessment and retain the evidence for five years.

Provenance
Verification id
VER-A6959FCCBE3E
Evidence hash
a6959fccbe3e6997…
Plot geometry id
PLT-F3B5C4A8A836F375
Engine version
1.0.0
Verified at
2026-09-15 10:00:00Z
Capture method
gps point
Stated openly

Caveats and data-quality findings

Anything that weakens this verdict is recorded with it. A compliance tool that reports only its confident findings is a tool that will eventually be wrong in a way nobody saw coming.

Geolocation was supplied as a single point, so the plot was modelled as a 6.80 ha circle centred on it. The true field boundary may differ.
Coordinates resolve to at most 3 decimal digits · Art. 2(28)

Article 2(28) requires at least six. No ordinate on this plot carries more than 3, which puts the position within roughly ±111 m — for a 6.8 ha plot, potentially off the farm altogether. Trailing zeros are invisible in a number, so this is inferred from the whole geometry rather than from any single vertex.

→ Re-export from the source GPS records without rounding, or re-capture the point.

Polygon required for a 6.80 ha plot · Art. 2(28)

Above four hectares, Article 2(28) requires a polygon with enough points to describe the perimeter. A single point will be rejected.

→ Walk the perimeter with a GPS-enabled phone; this is typically a 20-minute task per plot.

Article 10(1)

Written risk assessment

Assembled deterministically from the measurements rather than generated freely, so the same evidence always produces the same assessment.

Plot CPO-0024 (Ahou Kobena, Gnamangui, CI) covers 6.80 ha under cocoa. This land was already in agricultural use before the EUDR cut-off: less than the 0.5 ha FAO forest minimum was mapped as forest on 31 December 2020 (0.00 ha). Nothing was deforested to produce this commodity, which is the situation of the overwhelming majority of smallholder plots.

The finding rests on 3 independent evidence lines: JRC Global Forest Cover 2020 (V4 (2026 release)); Global Forest Change (Hansen/UMD) (GFC-2024-v1.12); Sentinel-2 L2A (surface reflectance) (Collection 1, AWS Open Data). Less than the 0.5 ha FAO minimum was mapped as forest on the cut-off date (0.00 ha). This land was already in agricultural use before EUDR applied. No tree-cover loss was detected on baseline forest inside this plot after 31 December 2020. Canopy structure inside the plot is stable since the cut-off (NDMI +0.010, NDVI -0.009).

The Article 10(2) assessment raised 3 findings, giving a composite risk score of 37 (standard). The most material are: 4 unresolved data-quality issues on the geolocation [10(2)(g)]; Severe national deforestation pressure [10(2)(f)]; Côte d'Ivoire is standard risk — full due diligence applies [10(2)(a)].

The negligible-risk threshold in Article 10(1) is not met on the present evidence. Article 11 mitigation is owed before any product from this plot is placed on the market.

Confidence qualifiers: Plot shape was inferred from a point, not walked.

Methodological caveats recorded: Geolocation was supplied as a single point, so the plot was modelled as a 6.80 ha circle centred on it. The true field boundary may differ.