Spatial join

A spatial join step attaches attributes from one layer (the "join") to features of another (the "target") based on a spatial relationship. The output has the target's geometry and the combined attributes.

This is the workhorse of "tell me which X each Y is in". Hydrants per parcel. Parcels per neighborhood. Points within a buffer.

Inputs

• Target layer. Features keep their geometry; their rows are in the output.
• Join layer. Features whose attributes get attached.
• Predicate (the relationship): - Intersects. Any overlap. Default for most cases. - Within. Target is entirely inside join. - Contains. Target entirely contains join. - Nearest (one-to-one). Pick the single closest join feature per target. - Within distance. Target is within N meters/feet of join.
• Field selection. Which join fields to attach to the output. By default all of them, prefixed join_.
• Cardinality: - One-to-one (nearest): one output row per target row. - One-to-many (default): one output row per matching pair. If a target intersects three join features, it produces three output rows.

Output

• Geometry: from the target.
• Attributes: target's fields + selected join fields.
• Row count: depends on cardinality (see above).

How to use it on its own

1. New-item wizard → Derived layer.
2. Pick Target.
3. Add a Spatial join step.
4. Pick the Join layer.
5. Choose the Predicate and Cardinality.
6. (Optional) Trim the join-field selection.
7. Save.

Examples

• Hydrants in each parcel. Target: parcels. Join: hydrants. Predicate: contains. Cardinality: one-to-many → one row per (parcel, hydrant) pair. Pair this with a group-by step to count hydrants per parcel.
• Nearest fire station per parcel. Target: parcels. Join: fire stations. Predicate: nearest. Cardinality: one-to-one. The output adds a join_station_id and a join_distance per parcel.
• Parcels in a neighborhood. Target: parcels. Join: neighborhoods. Predicate: within. Cardinality: one-to-one (each parcel is in exactly one neighborhood, in well-formed data).

Notes

• Geometry index helps. Both layers should have GIST indexes (data layers do by default). Without indexes, spatial joins on 100k-feature layers slow down sharply.
• Coordinate systems are reconciled automatically. Both layers are reprojected to a common SRS for the join; no manual step.
• NULL geometry in either side drops the row from the join output. The original target features with NULL geometry are still present elsewhere; they just don't get join attributes.