Compliance guide

EUDR geolocation requirements explained

The exact geolocation rules in EUDR: six-decimal WGS84 coordinates, when a point is enough and when you need a polygon, the 4-hectare rule, accepted file formats, and the geometry errors that get a DDS rejected.

8 min read · updated July 2026 · not legal advice

EUDR dossier · plotvera ART. 2(28)
Article 2(28): points up to 4 hectares, perimeter polygons above
Article 2(28): points up to 4 hectares, perimeter polygons above EPSG:4326

Geolocation is the load-bearing innovation of EUDR. Previous regimes asked where goods came from at country or concession level; Regulation (EU) 2023/1115 demands the coordinates of every plot of land that produced the commodity, embedded in the due diligence statement itself. It is also where most first filings fail - not on policy questions but on geometry mechanics. This guide is the complete ruleset.

The rule in the regulation's own terms

Article 2 defines geolocation as the location of a plot described by latitude and longitude coordinates with at least six decimal digits, and requires:

  • Plots of 4 hectares or less: may be described by a single point (one latitude/longitude pair).
  • Plots over 4 hectares: must be described by polygons - enough points to delineate the perimeter of the plot.
  • Cattle: geolocation refers to the establishments where the animals were kept, for which a point suffices.

A "plot of land" means land within a single real-estate property with homogeneous enough conditions to assess deforestation risk - in practice, the farm field, the harvest block, the plantation parcel. You cannot substitute an administrative region, a cooperative's office pin, or a mill location for the production plot.

Why six decimal places

Six decimal places of a degree is roughly 11 centimetres at the equator. The point is not survey-grade accuracy of your GPS fix - it is that the declared coordinate must not be truncated into ambiguity. Five decimals (~1.1 m) technically fails the letter of the definition; two decimals (~1.1 km) fails its purpose, since satellite screening of a point that might be anywhere in a square kilometre is meaningless. Collect at full receiver precision and never round.

Coordinate system and format

  • CRS: WGS84 (EPSG:4326), coordinates as decimal degrees. Files in projected systems - UTM metres are the classic case - must be reprojected, not relabelled.
  • Order: GeoJSON uses [longitude, latitude]. Swapped-axis files are the most common single error in supplier data; a coffee farm at "latitude 74" is in the Arctic Ocean.
  • Format: TRACES accepts geometry as GeoJSON, in a specific profile with its own property rules and silent failure modes (the GeoJSON format guide documents it fully). Supplier data arrives as GeoJSON, KML from Google Earth, WKT from databases, or shapefile exports - all convertible, all needing validation after conversion.

Polygon quality rules

Beyond the 4-hectare threshold, polygons must be geometrically valid:

  • Closed rings - first and last vertex identical.
  • No self-intersection - a figure-eight boundary is invalid; these appear constantly in hand-walked GPS traces.
  • Sensible winding and no duplicate consecutive vertices - tolerated by some tools, rejected by strict parsers.
  • Realistic area - if the polygon's computed area is 40 ha and the supplier's declared farm is 4 ha, one of them is wrong; both get scrutiny.
  • No absurdities - plots at (0,0), in the sea, or overlapping a city centre indicate placeholder or corrupted data. Authorities run exactly these sanity checks; run them first.

Points versus polygons in practice

SituationWhat you declare
Smallholder coffee/cocoa plot ≤ 4 haSingle point, 6 decimals
Soy field, forestry harvest block > 4 haPolygon delineating the perimeter
Cattle establishment (any size)Point for the establishment
Supplier offers only a village centroidNot acceptable - request plot-level data

Even where a point is legally sufficient, a polygon is strictly better evidence: it defines exactly which pixels of satellite history answer for your consignment. With only a point, screening must buffer an assumed area around it, which widens the zone in which someone else's clearing can flag your plot. Commodity guides cover the data reality per chain: coffee and cocoa (smallholder points), soy (large polygons), rubber and palm oil (mixed at scale).

Dates travel with coordinates

The DDS pairs geolocation with the time or range of production. This is what makes the 2020 cutoff testable: the claim is not "this plot is forest-free" but "this commodity was produced on this plot after 31 December 2020 without deforestation having occurred after that date". Keep harvest or production dates in the same dataset as the geometry from day one; retrofitting dates onto a bare coordinate list is painful.

Scale limits: how much geometry fits in a filing

Real consignments carry real data volumes - a rubber or cocoa shipment can aggregate tens of thousands of smallholder points. Plan for system limits: geometry files have size ceilings both in TRACES and in screening services, so bloated files (needless vertex density, embedded styling from GIS exports, seven-figure coordinate precision padded with trailing zeros) are worth cleaning before they hit a hard limit. Keep properties minimal - an identifier per plot plus production dates - and split very large supply sheds across statements if needed. The plot identifier matters more than it looks: it is the join key between your supplier records, screening verdicts and the filed statement when anyone asks you to reconstruct the chain two years later.

Data collection tactics that work

Two companion guides go deeper here: the smallholder collection playbook for point-capture campaigns at farm-gate scale, and the plot polygon guide for the five routes to boundary geometry.

  1. Make plot data a contractual delivery item per lot, in a named format ("GeoJSON, WGS84, plot-level") - not a goodwill request.
  2. Validate the day data arrives; every error found early is a supplier conversation instead of a TRACES-NT filing crisis.
  3. For suppliers without GIS capacity, a smartphone walk of the boundary or a drawn polygon over recent imagery both beat nothing - accuracy expectations scale with plot size.
  4. Version your plot files. Screening verdicts attach to a specific geometry; when a supplier corrects a boundary, keep both and note why. Five-year retention applies to the evidence trail, not just the final answer.

Once geometry is clean, screening and country benchmarking determine how much further diligence the consignment needs - and the deadline guide tells you how much runway you have to get there.

Deadlines: 30 Dec 2026 · 30 Jun 2027

Screen your first 3 plots free

Upload a GeoJSON, WKT or KML file, or draw a plot on the map, and get geometry validation plus a satellite deforestation verdict against the 31 December 2020 cutoff. No card, no sales call.

Related guides