How to Check Invalid or Impossible Coordinates

Validate numbers, geography and production meaning

EUDR coordinate checking needs three levels: whether the numbers and file are valid, whether the location is geographically plausible, and whether it actually identifies the relevant production place. A coordinate can pass the first two checks while still pointing to a supplier's office rather than the plot where the commodity was grown.

Use automated checks to identify candidates for investigation, then compare them with source evidence. Do not move points to the nearest plausible farm or correct a polygon solely to obtain a successful upload. A correction changes evidence and should be supported, documented and reviewed before the location contributes to a product decision.

Check numerical ranges and representation

Latitude must fall between minus 90 and plus 90 degrees, and longitude between minus 180 and plus 180 degrees. Values outside those ranges are invalid in the intended geographic coordinate representation. Check that the numbers are genuinely numeric and that decimal separators, signs and exported fields have not changed during processing.

A spreadsheet can convert a coordinate into text, drop a minus sign or interpret a comma as a field separator. Inspect the original and exported values rather than assuming the map software caused the error. The EUDR definition requires at least six decimal digits, but the number of digits does not establish that the underlying position is accurate.

Inspect axis order and reference system

The Commission's current GeoJSON specification uses longitude before latitude and WGS 84 decimal degrees. A table labelled latitude, longitude may need its fields deliberately reordered when creating GeoJSON. If both values fall within valid ranges after a swap, a simple range validator may miss the error entirely.

Projected easting and northing values need transformation from their actual reference system. Assigning the WGS 84 label to an untransformed file can produce an incorrect location without fixing the data. Retain the source system and transformation method so a reviewer can investigate systematic shifts instead of guessing how the file was prepared.

Compare the location with the declared country

Plot the geometry against the declared country and suitable geographic context. A point in an ocean, another country or an implausible industrial district deserves investigation. Country boundaries and coastal details can themselves have limitations, so an automated mismatch should open a review rather than immediately determine the substantive legal result.

For example, a farm near a border may appear just across a coarse boundary layer. Check the source, underlying coordinate and relevant local evidence. Another record may have a genuine country-code error. Both need correction for the right reason, not an automatic movement of the location to the country stated on the invoice.

Look for suspicious repetition

Repeated coordinates assigned to many unrelated producers can indicate that a collection centre or district midpoint was copied into farm fields. Repetition can also have legitimate explanations, such as shared establishments or duplicate supplier records. Investigate the pattern before deleting data or treating it as fraud.

Compare the producer identities, plot identifiers, collection method and source files. A cooperative can help distinguish shared management from an import error. The supplier handover guide explains why producer, plot and shipment identifiers should be kept separately. That structure makes suspicious merges and repetitions easier to diagnose.

Validate polygon structure and area

Check that polygon rings close, do not cross themselves and represent a genuine area rather than a line. Inspect unexpected overlaps and sharp spikes. The current Information System documentation identifies unsupported shapes, including polygons with holes, so local GIS validity alone is not enough to establish compatibility.

Compare the calculated geometry area with the reported production area where appropriate. A large difference can reveal an incorrect unit, a boundary mistake or inclusion of unrelated land. Do not replace a required polygon with a point to bypass the problem. Resolve the geometry while preserving the production plot that the evidence needs to describe.

Compare against the source production evidence

A geographically plausible location still needs to identify the correct plot or establishment. Check recognised land references, producer confirmation, field observations or other reliable evidence. For cattle, the record must fit the establishment history, rather than merely a plausible livestock site near the supplier's address.

Where the original evidence is insufficient, ask for clarification or additional collection. A mapping analyst should not infer the exact production plot from a broad village location simply because nearby cultivated land is visible. The unresolved identification should remain visible in the due diligence file until it can be supportably resolved.

Record corrections and affected decisions

Keep the original geometry, corrected version, evidence, reason, reviewer and date. Identify which lots and production periods relied on the earlier record. If the correction changes the assessed land, revisit the relevant land-history and legality analysis, the risk conclusion and any required regulatory follow-up.

The geolocation requirements guide places validation within the full assessment. Use Articles 2(28), 9 and 10 and the official GeoJSON error guidance. A validated file is a better evidence input, but it remains necessary to establish that the locations actually relate to the product and meet the substantive conditions.

Set clear ownership for errors discovered during import testing. The technical team may diagnose a malformed file, but a supplier or land reviewer may need to confirm the intended boundary. Keeping those responsibilities explicit prevents a software workaround from becoming an unsupported statement about where a commodity was produced.