GeoJSON conventions¶
kartoza-mcp follows RFC 7946
with three project-specific conventions.
CRS is WGS84¶
Coordinates are [longitude, latitude] in WGS84 (EPSG:4326).
This matches the RFC and what every consumer-grade mapping tool
expects.
If your data is in another CRS, project to WGS84 before passing it in.
Distances are metres¶
All distance parameters (eps, radius_meters, etc.) are in metres
on the WGS84 ellipsoid, computed with paulmach/orb/geo for true
geodesic distance — not planar lon/lat.
Properties round-trip¶
Input feature properties are not echoed in the output. If you need the original metadata associated with each cluster, do a join client- side on the centroid.
Input shape¶
{
"type": "FeatureCollection",
"features": [
{
"type": "Feature",
"geometry": { "type": "Point", "coordinates": [LON, LAT] },
"properties": { /* ignored */ }
}
]
}
Only Point features are read. Other geometry types are silently
skipped (a future major version may surface a warning).
Output shape¶
{
"type": "FeatureCollection",
"features": [
{
"type": "Feature",
"geometry": {
"type": "Polygon",
"coordinates": [ [ /* 33 lon/lat pairs, closed ring */ ] ]
},
"properties": {
"cluster_id": 0,
"point_count": 14,
"radius_meters": 2843.7,
"kartoza_credit": "Processed by Kartoza.com tools"
}
}
]
}
Polar caveat¶
Near the poles, the circle approximation uses local lat/lon-to-metre
scaling, which distorts as cos(lat) → 0. For analytical work above
~75° latitude, project to a polar-stereographic CRS in your map
client.