Clip vs Intersect vs Union

Clip, Intersect, and Union are GIS overlay operations, but they preserve different parts of the input geography. Use Clip when you want to trim one layer to a boundary. Use Intersect when you want only the geography shared by two layers, usually with information from both. Use an overlay Union when you want to preserve the complete area covered by either polygon layer and divide it according to the boundaries of both.

A useful way to remember the distinction is to ask what should survive the operation. Clip keeps part of one input. Intersect keeps what the inputs share. Union keeps the complete coverage of both.

That sounds simple, but Clip and Intersect can sometimes produce geometries that look identical, while the word union itself has more than one meaning across GIS software. Understanding what each result represents is therefore more useful than memorising the tool names.

Clip, Intersect, and Union at a glance

Suppose one layer contains land parcels and another contains a flood zone.

• Operation: Clip; What survives?: The parts of the input features inside the clipping boundary; What is the second layer doing?: Defining where the input should be kept; Typical question: Which parts of these parcels lie inside my study area?

• Operation: Intersect; What survives?: Only geography shared by the inputs; What is the second layer doing?: Contributing to both the spatial result and, commonly, its attributes; Typical question: Which portions of these parcels overlap the flood zone?

• Operation: Union; What survives?: Overlapping and non-overlapping portions of both polygon layers; What is the second layer doing?: Contributing its boundaries and information to the complete overlay; Typical question: How do these two polygon systems divide the entire area they cover?

This describes the common layer-overlay meaning of the operations. Exact field handling and available geometry combinations can differ between GIS applications, so the geographic question should remain the primary guide.

Clip trims one layer using another

Clip is appropriate when you already know which dataset you want to keep but only need the part that falls within a particular boundary.

Imagine that you have a national roads layer but your study covers only one municipality. The road features are the information you care about; the municipal boundary simply tells the GIS which portions to retain.

Roads completely outside the municipality disappear from the result. Roads completely inside it remain. A road crossing the municipal boundary is cut at that boundary, and only the section inside survives.

The important point is that the two inputs do not play equivalent roles. There is an input layer and a clipping layer.

If you reverse them, you ask a different question.

Clipping roads by a municipality means, roughly, “keep the roads inside this area”. Clipping municipality polygons using some other polygon layer instead means that those municipality geometries are now the features being trimmed.

Clip is consequently well suited to extracting data for a study area, reducing a larger dataset to a region of interest, or removing portions that are irrelevant to later processing.

Intersect keeps the geography that inputs share

Intersect asks a different question:

Where do these datasets overlap?

Suppose a parcel overlaps a flood zone. The intersection is the part of that parcel lying within the flood zone. The portion belonging only to the parcel is excluded, as is any flood-zone area that does not overlap the parcel.

In common GIS layer-overlay tools, the result can also receive attributes from both inputs. A resulting polygon might therefore retain the parcel identifier while also carrying the flood classification of the zone with which it overlaps.

This is one reason why Clip and Intersect should not be treated as interchangeable even when their maps initially look the same.

Consider one parcel and one flood-zone polygon. Clipping the parcel with the flood zone and intersecting those two layers may produce the same visible piece of geometry. The purpose of the two results is nevertheless different.

With Clip, the flood zone acts primarily as a boundary that determines which part of the parcel survives.

With Intersect, the overlap between parcel and flood zone is itself the subject of the result.

That distinction matters especially when attributes from the overlay layer will be used afterwards. If you need to know not merely which parcel geometry falls inside an area but also which zone it overlaps, which category that zone belongs to, or how multiple classifications combine, Intersect is generally the more natural operation.

Intersect is not the same as Intersects

There is another terminology trap worth separating from the overlay operation.

An intersection operation can construct new geometry representing the space shared by its inputs. An intersects predicate simply tests whether two geometries have any point in common and can return a true-or-false result.

For example, a spatial database can ask whether a road intersects a district without creating the section of road inside that district. If the actual shared geometry is required, a geometric intersection operation is needed instead.

The distinction becomes important when moving between desktop GIS tools, spatial databases, and programming libraries, because similar names may refer either to testing a spatial relationship or to constructing a new geometry.

Union preserves the complete overlay

Intersect throws away everything outside the shared area. An overlay Union does the opposite: it preserves the overlapping and non-overlapping parts covered by the input polygon layers.

Return to the parcels and flood zone.

An overlay Union can retain:

• parcel areas outside the flood zone;

• parcel areas inside the flood zone;

• flood-zone areas not covered by parcels.

Where boundaries from the two inputs cross, the output is divided so that those differences remain represented.

The result is therefore more than the outline of both datasets combined. It is a partition of their complete coverage according to the boundaries contributed by each layer.

This makes Union useful when the differences between two polygon systems matter.

Suppose, for example, that two datasets divide the same territory using different classifications. If you only intersect them, portions represented by one dataset but not the other disappear. A Union instead lets you retain those non-overlapping areas while also identifying the places where both classifications apply.

That completeness can make the output considerably more complex than either input. Every relevant boundary crossing can create additional pieces, so a Union between detailed polygon datasets may contain many more features than the original layers.

Why “Union” does not always mean the same operation

The word union is overloaded in geospatial software.

In a layer-overlay context, Union commonly means preserving all overlapping and non-overlapping portions of two polygon layers, splitting features along their shared boundaries and retaining the information required to describe those pieces.

A geometry function called union can mean something different.

For example, a spatial database may calculate the mathematical point-set union of two geometries: everything occupied by either geometry is combined, and internal overlaps are removed. Two overlapping polygons may consequently become a single merged geometry rather than an attribute-rich overlay divided into several pieces.

Both operations are forms of union, but they serve different purposes.

This distinction is also why Union should not automatically be treated as another name for Dissolve, Merge, or Append. Those operations answer different structural questions about geometry or datasets and deserve separate treatment.

Clip and Intersect can produce the same shape

One of the most useful ways to understand these operations is to look at a case where the output map alone cannot tell you which one was used.

Let polygon A be a parcel and polygon B a flood zone.

The geometry of:

Clip A by B

can be identical to the geometry of:

Intersect A with B

because both may retain the part of A inside B.

The difference lies in what the operation means and what information the output is designed to carry.

Clip treats B as the boundary used to trim A. Intersect treats the spatial relationship between A and B as the result.

This is why choosing an overlay operation by looking only at the expected shape can be misleading. Think about the output records as well: what does each resulting feature represent, and which input information will you need afterwards?

A practical way to choose

Rather than memorising diagrams, begin with the sentence that describes your task.

If the sentence is:

“Keep my features, but only inside this boundary.”

Use Clip.

If it is:

“Keep only the places where these datasets overlap, and preserve the relationship between them.”

Use Intersect.

If it is:

“Keep everything represented by either polygon layer and preserve the boundaries created by both.”

Use an overlay Union.

A second question can help when Clip and Intersect both seem plausible:

Do I care about the second layer only as a boundary, or does its information belong in the result?

If it is mainly a boundary, Clip is usually the clearer expression of the task. If the relationship between both layers matters, Intersect is usually more appropriate.

What can go wrong with overlay operations?

Choosing the correct operation does not guarantee a sensible result. Overlay depends on the geometry supplied to it, and it often reveals problems that were difficult to see in the original layers.

Slightly different boundaries can create small fragments

Two boundaries can appear identical at normal map scale while differing slightly in their actual coordinates. Overlaying them may create narrow polygon fragments between the two versions.

These fragments are often called slivers. Their presence does not necessarily mean the overlay calculation failed; it may reveal that the inputs do not describe the supposedly shared boundary in exactly the same way.

Automatically deleting such pieces can therefore hide a data-quality problem rather than solve it.

One feature can become many

Clip and Intersect can split features at overlay boundaries, while Union deliberately preserves the subdivisions created by both polygon systems.

The number of output features can therefore be much larger than the number of input features.

For example, a single parcel crossing three different zones may become several separate records after an intersection. Those records are not automatically duplicates: each may represent a different geographic piece or relationship.

Feature counts should consequently be interpreted in terms of the operation rather than compared mechanically with the original row count.

Geometry validity matters

Malformed polygons can cause overlay operations to fail or produce unexpected results. Self-intersections and other invalid geometry are particularly relevant when polygon boundaries must be combined precisely.

Geometry repair is not automatically harmless, however. A repair operation can itself alter a feature, so it is worth understanding the validity problem before modifying the source geometry.

Coordinate reference systems still matter

Overlay operations compare coordinates. Those coordinates must describe the same locations consistently.

Modern GIS software can often transform layers between coordinate reference systems during display or processing, so different CRS labels do not automatically prevent an overlay. Problems arise when a CRS is missing, incorrectly assigned, transformed incorrectly, or otherwise interpreted in a way that causes geometries representing the same place to occupy different coordinate positions.

Reprojecting every dataset by habit is therefore less useful than confirming that the spatial references are correctly defined and that the operation is comparing geometries in a suitable common coordinate space.

Do you actually need an overlay?

Not every question involving two layers requires Clip, Intersect, or Union.

Suppose you have school points and district polygons and simply want to know which district contains each school. If you want to preserve the original points and attach a district identifier, a spatial join may express the task more directly.

Likewise, if you only need to select the schools that fall within one district, creating new intersection geometry may add nothing useful.

Overlay is most valuable when the resulting geometry itself matters: the portion of a road inside a zone, the part of a parcel affected by a restriction, or the pieces produced where two polygon classifications cross.

Choosing the simplest operation that represents the actual question usually produces data that is easier to understand afterwards.

The difference comes down to what you want to keep

Clip, Intersect, and Union all use spatial overlap, but they preserve different parts of the problem.

Clip keeps the relevant parts of one input inside a boundary. Intersect keeps only the shared geography and commonly brings together information from both inputs. Union, in its layer-overlay sense, preserves the complete polygon coverage of both layers and divides it according to their combined boundaries.

When two operations seem capable of producing the same visible result, do not choose by appearance alone. Ask what each output feature should represent, what information needs to survive, and whether geography outside the overlap still matters.

Once those questions are clear, the appropriate overlay operation usually becomes clear as well.

References

1. QGIS Documentation — Vector overlay. Documents the behaviour of QGIS Clip, Intersection, Union, Difference, and related vector-overlay operations, including which portions of the input layers are retained and how attributes are handled.

2. PostGIS — ST_Union. Defines geometric union as the point-set union of input geometries, with overlaps merged, and provides the technical basis for distinguishing geometric union from an attribute-preserving layer-overlay Union.

3. PostGIS — ST_Intersects. Defines ST_Intersects as a Boolean spatial relationship test, supporting the distinction between testing whether geometries intersect and constructing intersection geometry.