Open audit · written for the producer

How your score was calculated — nothing hidden

Jethro, this page shows every number, source and rule we used to decide that your plot is cleared for export to Europe. If a machine or a person made a judgement, it is written here so you — or an independent auditor — can check it and, if needed, challenge it.

Deforestation test✔ Deforestation-free0.00 ha lost after 31 Dec 2020

Legality6 / 6 checksall documents present & valid

Final risk score8.9 / 100Negligible → cleared for export

How to read this page. There are three parts: Part A — the deforestation test (is any forest lost?). Part B — the legality checklist (is the plot legal?). Part C — the risk score (one number that combines everything). Each part shows the exact inputs and rules.

Part A — The deforestation test

We take the exact GPS boundary of your plot and lay it over satellite pictures of the land. We compare today against the legal reference date, 31 December 2020. If any forest inside your boundary was cleared after that date, it counts against you. Here is your data:

Variable	Meaning (plain language)	Your value
A — plot area	Size of your mapped boundary	2.31 ha (23,100 m²)
baseline	The legal cut-off image we compare against	31 Dec 2020 · Planet NICFI
latest	Most recent clear satellite pass	09 Jul 2026 · Sentinel-2
L_raw — loss signal	Pixels flagged as possible forest loss after cut-off, inside your plot	0.03 ha (300 m²)
MMU	Minimum Mapping Unit — signals smaller than this are treated as noise, not clearing	0.05 ha
buffer	GPS edge tolerance — signals within this distance of the boundary may be a neighbour or a footpath	10 m
L_verified — confirmed loss	Loss remaining after the rules and a field check	0.00 ha

The rule we applied, step by step

# 1. A raw signal of 0.03 ha appeared on the eastern edge of your plot. L_raw = 0.03 ha # 2. Is it bigger than the noise threshold (MMU)? 0.03 ha < MMU (0.05 ha) → below threshold — not automatic clearing # 3. Is it on the plot edge (could be a neighbour / path)? distance to boundary = 6 m < buffer (10 m) → edge signal — needs a human check # 4. Field agent visited on 07 Jul 2026 and photographed the spot. finding = "shade-tree pruning, cocoa canopy intact — NOT clearing" # 5. Final confirmed loss after review: L_verified = 0.00 ha # Deforestation-free share of the plot: DFI = (A − L_verified) / A = (23,100 − 0) / 23,100 = 100.00% ✔ COMPLIANT

Why we didn't just fail you automatically. A cheap system would have marked that 0.03 ha as "deforestation" and rejected your cocoa. We don't. Small edge signals are common (pruning, paths, cloud shadow), so the rule sends them to a human field check before any decision. That is the difference between a fair audit and a blunt filter.

Part B — The legality checklist

Being deforestation-free is not enough for Europe — the plot must also be legally produced. Each item below is either present (✔) or missing (✗). You scored 6 out of 6.

✔
Land-use right — attestation from the commune of Dondon. Weight in legality score: high
✔
Producer identity verified — national ID matched.
✔
Cooperative membership — FECCANO card #2211.
✔
No protected-area overlap — your boundary was checked against the national parks & reserves map. No overlap.
✔
No embargo or environmental infraction — checked against the public infraction registry.
✔
Free, informed consent — you signed to allow your data to be used for this declaration (FPIC principle).

Legality completeness = documents present ÷ documents required = 6 ÷ 6 = 100%. A missing high-weight document (e.g. land-use right) raises risk far more than a missing optional one (e.g. tax record).

Part C — The final risk score

Finally we combine everything into one number from 0 to 100, where lower is safer. Five factors each get a sub-score (0 = perfect, 100 = worst) and a fixed weight. The weights never change between producers — they are published here so the score can't be gamed.

# Weighted risk model (weights are fixed & public) Risk = 0.40·Deforestation + 0.25·Legality + 0.15·Traceability + 0.10·DataQuality + 0.10·CountryBenchmark

Factor	What it measures	Weight	Your sub-score	Contribution
Deforestation	Confirmed forest loss after cut-off. You: 0 ha, but one signal was reviewed → small watch value.	40%	2	0.80
Legality	Missing or weak documents. You: 6/6 complete.	25%	5	1.25
Traceability	How cleanly beans link plot→bag→lot. You: QR bags, 1 intermediary.	15%	8	1.20
Data quality	Capture method (walked vs circle estimate), GPS precision, image freshness & cloud cover. You: perimeter-walked, ±3 m, <30 days, 4% cloud.	10%	6	0.60
Country benchmark	EU country-risk class for the origin (a baseline, not about you personally).	10%	50	5.00
Final risk score				8.85 ≈ 8.9

0 – 24

Negligible

Cleared — DDS can be issued

← you are here (8.9)

25 – 49

Standard

Must fix issues before export

50 – 100

High

Excluded until resolved

Why your score isn't zero. Two things keep it slightly above zero, and both are honest: (1) one deforestation signal had to be reviewed, so we keep a small "watch" value of 2; and (2) the origin country carries a fixed baseline of 50 on the benchmark factor — that is about the country, not about you. Even so, your weighted total is 8.9, well inside "negligible".

Part D — Volume plausibility (a separate anti-fraud gate)

Passing the risk score is not the last check. Even a perfectly clean plot cannot declare more cocoa than the land can physically grow. This blocks the most common fraud: registering one small clean plot and pushing cocoa from an unmapped (possibly deforested) plot through it.

# Land can only yield so much. Ceiling = area × maximum plausible yield. Ceiling = 2.31 ha × 560 kg/ha = 1,294 kg / year Exported = 512 kg → 512 / 1,294 = 40% ✔ PLAUSIBLE # If Exported > Ceiling → plot is auto-frozen and excluded from the lot.

Worked example of a failure. Producer Rony Baptiste (#067) declared 118% of his 2.6 ha capacity — impossible from his land alone — so the system froze the plot and removed it from the lot before export. Honest producers are never penalised; only impossible volumes are stopped.

How other outcomes look (for fairness)

So you can see the rule treats everyone the same, here are three real cases from the same lot:

Producer	Deforestation	Legality	Score	Outcome
Jethro Similien (you)	0 ha	6/6	8.9	Cleared
Fabiola Casseus	0 ha	5/6 (doc pending)	21.5	Fix & re-submit
Rony Baptiste	0.7 ha cleared 2022	6/6	78.0	Excluded from lot

Data sources & dates (provenance)

Plot boundaryGPS survey · 04 Jul 2026 · ±3 m

Cut-off baselinePlanet NICFI · Dec 2020

Change detectionSentinel-2 + GFW/Hansen v1.12

Legality layersNational parks & infraction registry

Field verificationAgent visit · 07 Jul 2026

Record hash0x7c4e…d92a

Your rights

◉
Right to see every number used — this page.
◉
Right to challenge — if you think the satellite or a check is wrong, request a re-survey; a human will visit.
◉
Right to your data — your plot and documents belong to you and your cooperative.
◉
No black box — the weights are fixed and public; the same rule is applied to every producer.

See the EU statement this feeds →

Why European buyers value this page. An EUDR audit can demand to see how a "deforestation-free" claim was reached. Most systems can only show a green tick. TerraProva shows the whole reasoning — inputs, thresholds, the human review, and the hash that proves none of it was changed afterwards. That is what turns a claim into defensible evidence.