EUDR geolocation: coordinates, polygons and the four-hectare rule

What the regulation asks for

Geolocation of all plots of land where the relevant commodities were produced, expressed as latitude and longitude to at least six decimal places. Six decimals is roughly 0.1 m at the equator. The requirement is a precision format, not a precision guarantee, but it rules out the rounded coordinates that supply chains have historically passed around.

Whether you provide a point or a polygon depends on plot size and commodity.

• Four hectares or less A single point One latitude/longitude pair to six decimal places is sufficient, for commodities other than cattle. This is what makes smallholder collection tractable at all, a phone GPS reading taken standing in the plot.
• More than four hectares A polygon The plot boundary, given as at least six points describing the perimeter. Required for every commodity other than cattle.
• Cattle Every establishment Geolocation of all establishments where the animals were kept. Cattle move, and the obligation follows the animal across every holding in its life, not just the last one.

Four hectares is per plot, not per farm. A grower with six separate two-hectare parcels supplies six point coordinates, not one polygon around all of them. Conversely, one nine-hectare field is one polygon. The unit is the plot of land where the commodity was produced, and drawing the boundary generously to "cover" adjacent land is actively harmful. It pulls in tree-cover loss that has nothing to do with you.

The format

The information system accepts GeoJSON. A point plot and a polygon plot look like this, note that GeoJSON orders coordinates longitude first, which is the single most common formatting error in submissions.

Three things that reject or distort submissions:

• Reversed coordinate order. GeoJSON is [longitude, latitude]. Most GPS apps and spreadsheets present latitude first. A plot in Honduras that lands in the Indian Ocean has been swapped.
• Unclosed polygons. The first and last coordinate pair must be identical. Six distinct corners means seven entries in the ring.
• Wrong datum or projection. Coordinates must be WGS 84. National grid references and UTM coordinates need converting first, and a silent projection mismatch shifts plots by hundreds of metres.

Collecting it, in practice

The technical requirement is trivial. Getting the data out of a supply chain that runs through cooperatives, intermediaries and thousands of smallholders is not. Four routes, roughly in order of reliability.

1. 1.Direct capture by the producer. A phone app, walked around the plot or stood in the middle of it. Most accurate, hardest to organise, and dependent on the season. You can only map what is accessible when the crop is there.
2. 2.Cooperative-held registries. Many cooperatives already hold member plot data for internal or certification purposes. Quality varies enormously and the boundaries are often years old, but it is the fastest starting point.
3. 3.Existing certification data. Some schemes hold plot polygons. Useful as a base layer; not sufficient on its own, and not a substitute for your own verification.
4. 4.Cadastral or land-registry records. Reliable where they exist and are digitised, which in many origin countries means not at all.

Start with the plots, not the paperwork. Every other part of EUDR compliance can be done at a desk in December. Geolocation cannot. It depends on other people, in other countries, often within a specific window in the harvest cycle. If you are going to be late for December 2026, this is what will make you late.

Plots you cannot get

You will have some. The honest options are narrow, and one of them is not an option.

• Segregate. Keep unmapped supply out of EU-bound lots. Commercially painful, legally clean, and usually the right answer for a first season.
• Substitute the supply. Move EU-bound volume to producers you have mapped, and route unmapped volume to markets without the requirement.
• Invest in the mapping. Fund the collection directly. For strategic origins this is cheaper than losing the supply, and it is exactly the kind of risk mitigation Article 11 contemplates.
• Not an option: approximating. Dropping a pin on a village, using a cooperative's office coordinates, or generating a plausible polygon is submitting false information in a due-diligence statement. It is also trivially detectable, a screen against the forest baseline shows immediately when a hundred "plots" sit on the same building.

Mixing, and why plot data has to stay attached

Commodities get bulked. A container of green coffee may hold lots from four hundred plots; a pulp shipment may draw on dozens of concessions. Bulking lots from many plots into one consignment is normal and allowed, but the statement covers the whole consignment, so every plot contributing to it has to be mapped and screened, and one unmapped plot contaminates the batch it enters.

The practical implication is that plot identity has to survive aggregation. If your systems lose the link between a plot and the lot it went into at the point of bulking, you cannot reconstruct it later, and you will end up screening far more land than you needed to.

Related

• Collecting plot data from smallholders and cooperatives
• The 2020 cutoff and what screening actually tells you
• Filing a due-diligence statement