GeoJSON not UTF-8: accented producer names and the EUDR system

What the Information System does

The GeoJSON standard requires UTF-8 (RFC 7946, §12). The Information System's documentation does not describe what it does with other encodings; in practice a byte that is not valid UTF-8 either stops the file being parsed or arrives as a replacement character in the producer's name.

A broken example

{
  "type": "FeatureCollection",
  "features": [
    {
      "type": "Feature",
      "properties": {
        "ProducerName": "Coop�rative",
        "ProducerCountry": "CI",
        "ProductionPlace": "Parcelle 1",
        "Area": 1.2287
      },
      "geometry": {
        "type": "Polygon",
        "coordinates": [
          [
            [-5.123456, 6.123456],
            [-5.122456, 6.123456],
            [-5.122456, 6.124456],
            [-5.123456, 6.124456],
            [-5.123456, 6.123456]
          ]
        ]
      }
    }
  ]
}

What the checker reports for this file:

1. The file contains 1 character that could not be decoded as UTF-8. Producer names with accents are the usual casualty.

Fix it in EUDRplot

1. Open the file in the checker. The checker reads the bytes before decoding them, so it can tell a UTF-8 file from one saved in another encoding.
2. Re-save the file as UTF-8 at the source. In the program that made it, choose UTF-8 when saving or exporting. In Excel, use CSV UTF-8 rather than plain CSV. Then open the new file in the checker.
3. Check the names. Look at a producer name with an accent (é, ã, ñ, ç) in the editor. If it reads correctly, the encoding is right.

Fix it in QGIS

1. Tell QGIS the real encoding. Layer › Add Layer › Add Vector Layer, choose the file, and set Encoding under Options to the one it was saved in, usually windows-1252 for files from Western European Windows machines.
2. Check the attribute table. Open the attribute table and confirm the accented names now read correctly.
3. Export as GeoJSON. Right-click › Export › Save Features As…, format GeoJSON, CRS EPSG:4326. QGIS writes GeoJSON as UTF-8.