Centroid

Draft: written, not yet checked by a person Operation · centroid

One point per feature, at its geometric centre or, if asked, at a point guaranteed to lie inside it.

Pitfall. The centroid of a concave or multipart polygon can fall outside it; use inside for labels and joins back to the polygon.

Pitfall. Centroids of data in a geographic CRS are computed on degrees and are slightly off the true centre on the ground.

About this operation
Also called	Feature To Point, centroids, point on surface, representative point, label point
Group	`vector/geometry`
Checks	`geometry.valid`
Pitfalls	2 listed above
Columbus tool	Not stated
Notes	Not stated
Status	active
Broader	Not stated
Narrower	Not stated
Related	`convex-hull`

Parameters

Name	Type	Required	Default	Values	Note
`inside`	boolean	no	`no`	Not stated	Use a point guaranteed to lie inside each polygon (point on surface) instead of the centroid.

Inputs and outputs

Direction	Name	Type	Geometry	Note
Input	`layer`	vector	Not stated	Not stated
Output	`output`	vector	point	Not stated

In each engine

Engine	Tool	Parameters	Differences
`apache-sedona`	`sedona.sql.ST_Centroid`	`inside` always no. Use ST_PointOnSurface for a point inside the polygon. `layer` → `geometry`	A Spark SQL function: arguments are columns or column names, run over a Spark DataFrame.
`apache-sedona`	`sedona.sql.ST_PointOnSurface`	`inside` always yes. Use ST_Centroid for the geometric centroid. `layer` → `geometry`	A Spark SQL function: arguments are columns or column names, run over a Spark DataFrame.
`geopandas`	`geopandas.GeoSeries.centroid`	`inside` always no. Use GeoSeries.representative_point() for a point inside the polygon.	A property of the GeoSeries, not a method; returns a GeoSeries without the other columns. Warns when the CRS is geographic.
`geopandas`	`geopandas.GeoSeries.representative_point`	`inside` always yes. Use GeoSeries.centroid for the geometric centroid.	Returns a GeoSeries without the other columns.
`postgis`	`st_centroid`	`inside` always no. Use ST_PointOnSurface for a point inside the polygon. `layer` → `#1`	On geography it computes a spheroidal centroid (use_spheroid).
`postgis`	`st_pointonsurface`	`inside` always yes. Use ST_Centroid for the geometric centroid. `layer` → `#1`	None noted
`qgis`	`native:centroids`	`inside` always no. Use native:pointonsurface for a point inside the polygon. `layer` → `INPUT` `output` → `OUTPUT`	ALL_PARTS gives one point per part of multipart features.
`qgis`	`native:pointonsurface`	`inside` always yes. Use native:centroids for the geometric centroid. `layer` → `INPUT` `output` → `OUTPUT`	ALL_PARTS gives one point per part of multipart features.
`sf`	`sf::st_centroid`	`inside` always no. Use st_point_on_surface for a point inside the polygon. `layer` → `x`	of_largest_polygon = TRUE gives the centroid of each feature's largest part.
`sf`	`sf::st_point_on_surface`	`inside` always yes. Use st_centroid for the geometric centroid. `layer` → `x`	None noted
`shapely`	`shapely.centroid`	`inside` always no. Use shapely.point_on_surface for a point inside the polygon. `layer` → `geometry`	None noted
`shapely`	`shapely.point_on_surface`	`inside` always yes. Use shapely.centroid for the geometric centroid. `layer` → `geometry`	None noted
`terra`	`terra::centroids`	`inside` → `inside` `layer` → `x`	None noted
`turf-js`	`turf.centroid`	`inside` always no. Use turf.pointOnFeature for a point inside the polygon. `layer` → `geojson`	The mean of all vertices, not the area-weighted centroid (turf.centerOfMass gives that). Turf works on GeoJSON in longitude/latitude (WGS 84); data in another CRS must be reprojected first.
`turf-js`	`turf.pointOnFeature`	`inside` always yes. Use turf.centroid for the geometric centroid. `layer` → `geojson`	Turf works on GeoJSON in longitude/latitude (WGS 84); data in another CRS must be reprojected first.
`whiteboxtools`	`CentroidVector`	`inside` always no `layer` → `input` `output` → `output`	None noted