Erase

Keep only the parts of input features that fall outside an erase mask. The geometric inverse of clip.

basic

An erase step keeps the parts of each input feature that fall outside an erase mask. Features entirely covered by the mask drop out; features that cross the mask boundary are cut.

This is the geometric inverse of Clip. Hand it a parcel layer and a flood zone, get back the parcels outside the flood, with parcels on the edge trimmed at the boundary.

Inputs

  • Source layer. Any geometry type (points, lines, or polygons).
  • Erase mask. Another data layer you can read. Prefer a polygon layer as the mask. The portal unions that layer's geometries and differences the source against the result. Geo-boundary items and pasted GeoJSON are not a separate erase input on this path; put the mask in a data layer first.

Output

A new layer with the source's schema. Each row's geometry is ST_Difference(source.geometry, mask). Rows whose geometry is fully covered by the mask drop out; rows that never touch the mask come through unchanged.

How to use it on its own

From Analyze (guided)

  1. Open a data layer and switch to the Analyze tab.
  2. Click Erase this layer.
  3. Pick the erase-mask data layer.
  4. Click Create derived layer.

That posts a single-step erase recipe to the portal. The job runs on PostGIS on the server. Expect a short server job, not an instant browser result. Not gated on the optional server-heavy analysis tier.

From the new-item wizard

  1. New-item wizard → Derived layer.
  2. Pick a Source.
  3. Add an Erase step.
  4. Pick the erase-mask data layer.
  5. Create the item.

What it is not

  • Not an in-browser DuckDB erase. Meter-accurate difference needs PostGIS on the server.
  • Not a paint-over editor that deletes features in place. Erase creates a new derived layer; the source stays untouched.
  • Not Spatial Analyst Erase licensing. Same idea, free on your PostGIS.