9.3 Clip, Erase, and Dissolve

Key takeaways

• Clip keeps the part of a layer that overlaps another; erase removes it.
• Dissolve merges adjacent features sharing an attribute value.
• Together they prepare data for cartography and modelling.

Introduction

Three vector operations appear in almost every workflow: clip, erase, and dissolve. They're conceptually simple but deceptively powerful. This short lesson covers each, their subtleties, and how they interact.

Clip

Clip cuts the input layer by a polygon mask, keeping only the parts that overlap.

SQL:

1SELECT ST_Intersection(r.geom, b.geom) AS geom, r.attributes
2FROM rivers r, boundary b
3WHERE ST_Intersects(r.geom, b.geom);

QGIS: Vector → Geoprocessing Tools → Clip. GeoPandas: gpd.clip(roads, city_boundary).

Use cases:

• Reduce a continental dataset to the area of interest.
• Produce a map-ready layer that doesn't extend beyond the study region.

Note: clipping doesn't split features at clip boundaries into multiple rows unless you explicitly want that — the attribute table remains the same, only geometry is trimmed.

Erase (difference)

Erase removes the parts of the input that overlap the mask.

SQL:

1SELECT ST_Difference(l.geom, p.all_protected) AS geom
2FROM landuse l
3CROSS JOIN (SELECT ST_Union(geom) AS all_protected FROM protected) p;

Use cases:

• Identify developable land (all - protected).
• Produce the "rest of the world" outside a study area.

Dissolve

Dissolve merges adjacent or overlapping features that share an attribute value, producing fewer, larger features.

SQL:

1SELECT class, ST_Union(geom) AS geom
2FROM landuse
3GROUP BY class;

GeoPandas:

dissolved = landuse.dissolve(by='class', aggfunc='sum')

Use cases:

• Simplify a cadastre by owner (all parcels owned by one entity become one multipolygon).
• Merge census blocks into districts by aggregating attributes.
• Remove internal boundaries within a country.

Dissolved features typically carry aggregated attributes (sum of population, count of members, average of density). Decide aggregation functions intentionally.

Attribute handling

When dissolving:

• Sum / count — straightforward for numeric attributes.
• Average — may need weighting (area-weighted average).
• Mode / first — pick one value for text attributes.
• Drop — for attributes that don't aggregate meaningfully.

Combining the three

A common pipeline: clip → dissolve → erase → display.

Example — "forests inside our study area, merged by species, minus protected zones":

1WITH forests AS (
2  SELECT species, ST_Intersection(geom, (SELECT geom FROM study_area)) AS geom
3  FROM forests_layer
4  WHERE ST_Intersects(geom, (SELECT geom FROM study_area))
5),
6merged AS (
7  SELECT species, ST_Union(geom) AS geom
8  FROM forests
9  GROUP BY species
10),
11unprotected AS (
12  SELECT m.species,
13         ST_Difference(m.geom, (SELECT ST_Union(geom) FROM protected)) AS geom
14  FROM merged m
15)
16SELECT * FROM unprotected WHERE NOT ST_IsEmpty(geom);

Four lines of logic, written one step at a time, debuggable at each.

Pitfalls

• Dissolve with invalid geometries silently produces wrong results. Validate first.
• Clip without preserving attributes — make sure you select the attributes you need.
• Erase of huge layers — can be slow without pre-union and indexes.
• Dissolve doesn't always merge adjacent features if they don't touch exactly (slivers, snapping issues).

In GeoPandas

1# Clip
2clipped = gpd.clip(roads, boundary)
3[object Object]
4[object Object]
5[object Object]
6

Self-check exercises

Summary

• Clip keeps the overlapping portion; erase removes it.
• Dissolve merges features by attribute, aggregating the rest.
• A typical pipeline chains clip → dissolve → erase; work step by step.
• Validate and snap before large operations to avoid surprises.

Further reading

• QGIS Documentation — geoprocessing tools reference.
• GeoPandas Documentation — overlay, clip, dissolve.
• Geocomputation with R — vector operations chapter.
• PostGIS cheat sheets for common spatial SQL patterns.