OpenCanopy

EU Deforestation Regulation · forest cutoff 31 Dec 2020

Sustainable agriculture
on ethically sourced land.

OpenCanopy is an independent check a government can run when a plantation asks to be approved. Give it one submitted plot. It returns the satellite record back to the cutoff, and whether that record supports approval.

Follow one plot Skip to the case file
Plot
KS-19 · fictional applicant
Location
0.7800° S, 110.0800° E
Record
Sentinel-2 · Hansen GFC v1.12
Scroll
14 Sep 2019 Sentinel-2 L2A · cloud 3%

The submission

A plantation in West Kalimantan asks to be approved.

It sends one thing a reviewer can check: the location of the plot. Every other claim on the form is the applicant’s own word.

14 Sep 2019

Before the cutoff, it was forest.

Unbroken canopy across the whole box. A river winds through the upper left. Mean NDVI 0.88.

31 Dec 2020

The date the law looks back to.

Forest turned into farmland after this day cannot supply the EU market. The duty starts in December 2026. The forest that counts is the forest of 2020.

Hansen loss year · 2023

Then, in one year, about 880 hectares.

Each red cell is a 30-metre pixel the Hansen Global Forest Change map records as lost. Almost all of the post-2020 loss on this plot falls in 2023.

8 Aug 2026

Today the ground has rows, roads and blocks.

A vision model reads the current image and finds rectangular plots in a grid, straight access roads, and planting rows. The loss map ends in 2024, so this image shows what the land is now. It is not a 2026 detection.

Cross-check

Two regions marked. Only one holds up.

  • ConfirmedLarge lower region. 43% lost after 2020.
  • InconclusiveSmall corner. 14% lost, not enough to confirm. A person reviews it.

What OpenCanopy stands behind

This plot shows post-2020 tree-cover loss consistent with plantation conversion.

The record does not support approval. The government decides.

Read the full case file

Method

OpenCanopy's record through history.

  1. 01

    Two pictures

    The clearest Sentinel-2 scene before the cutoff, and the clearest one now. Same box, same season where possible.

  2. 02

    The loss year

    Hansen Global Forest Change marks the year each 30 m pixel lost its trees, through 2024.

  3. 03

    What it became

    A vision model reads the current image and outlines regions that look converted: blocks, roads, rows.

  4. 04

    The cross-check

    A region is confirmed only when the loss map shows substantial clearing inside it after 2020. Anything else stays inconclusive.

Limits

The purpose of a check.

Says

  • This plot shows post-2020 tree-cover loss consistent with plantation conversion.
  • This region is inconclusive and needs a person.
  • No post-2020 loss and no conversion in the imagery: consistent with the applicant’s claim.

Never says

  • Non-compliant, illegal, or banned. Those are findings for an authority.
  • A supplier score. A check is about one piece of ground, not a company.
  • Detected in 2026. The loss map ends in 2024. Newer images show what the land is now.
  • An enhanced image. Every picture is the satellite scene as delivered.

Public by design

People who live with that forest can read the same pack.

Approval rests on the images and the loss map, not on the applicant’s description of the land. Every check carries its scene IDs, dates and cloud cover, so anyone can pull the same data and see the same thing.

Sources

Everything on this page comes from open data.

Imagery
Copernicus Sentinel-2 L2A, via Microsoft Planetary Computer
Loss year
Hansen Global Forest Change, GFC-2024-v1.12 (loss through 2024)
Forest in 2020
JRC Global Forest Cover 2020, v4
Vision model
Perceptron Mk1.5. Asked only whether converted land looks planted.
Applicant
Plot KS-19 is fictional. The coordinates and the landscape are real.