Why Buffer Distance Depends on the CRS

A GIS buffer distance only has meaning when you know how the software measures distance in the coordinate reference system. A value of 500 might mean 500 metres in one projected CRS, 500 feet in another, or 500 degrees if an operation interprets the raw units of a geographic CRS.

Even when a projected CRS uses metres, that does not guarantee that a 1,000-metre distance on the map represents exactly 1,000 metres on the ground everywhere. Every map projection introduces distortion, and the amount and type of distortion depend on the projection and location.

This is why choosing a buffer distance is not just a matter of typing a number into a GIS tool. You need to know what units the operation expects, how those coordinates represent the Earth, and whether the CRS is suitable for measuring distance in the area being analysed.

The buffer distance is measured somehow

A buffer creates an area within a chosen distance of a feature.

If you ask for:

buffer distance = 1000

the GIS still needs to know what 1000 represents.

Possible interpretations include:

• 1,000 metres;

• 1,000 feet;

• 1,000 US survey feet;

• 1,000 degrees;

• 1,000 arbitrary map units.

That information comes from the spatial reference and, in some systems, from the particular geometry or geography calculation being used.

QGIS's introductory documentation on vector buffering describes buffer distances as being defined in map units according to the CRS, with possible units such as metres, feet, or decimal degrees.

PostGIS makes the rule particularly explicit for its geometry type: ST_Buffer interprets the distance using the units of the geometry's spatial reference system. For its geography type, by contrast, the buffer distance is given in metres and PostGIS handles the geodetic input differently.

So before asking whether 1000 is a sensible buffer distance, first ask:

One thousand what?

Geographic coordinates are angular coordinates

A geographic coordinate reference system typically describes position using longitude and latitude.

For example, coordinates might look like:

longitude = 11.52
latitude  = 3.87

These are angular values, commonly expressed in degrees.

Degrees are useful for describing locations on the Earth, but a degree is not a fixed ground distance in the way a metre is.

One degree of latitude corresponds to roughly similar north-south distances across much of the Earth, although not exactly the same everywhere. One degree of longitude changes much more dramatically: its ground distance decreases as you move towards the poles.

Near the equator, one degree of longitude covers roughly 111 kilometres. At higher latitudes it covers less, eventually approaching zero at the poles.

That means a simple planar rule such as:

buffer by 0.1 coordinate units

has no globally constant interpretation in metres when those units are longitude and latitude degrees.

This is one reason buffering raw longitude/latitude geometry as though degrees were ordinary Cartesian distance units can produce misleading results.

The broader distinction between geographic and projected coordinate systems is covered in Geobble's CRS material, including EPSG:4326 vs EPSG:3857.

A projected CRS gives you planar coordinates

A projected coordinate reference system transforms locations from the curved Earth onto a plane.

Coordinates might then look like:

Easting  = 552,840 m
Northing = 428,190 m

If the CRS uses metres and a conventional planar buffer algorithm interprets the geometry in that CRS, entering:

1000

means 1,000 coordinate-system metres.

This makes projected coordinates much more convenient for many distance-based operations.

But there is an important qualification:

Projected metres are not automatically distortion-free ground metres.

Projection determines how positions on the curved Earth are represented on the plane. That transformation cannot preserve every spatial property perfectly across an unlimited area.

A projection may preserve some properties well while allowing distance or scale to become increasingly distorted elsewhere.

So the correct question is not simply:

Does this CRS use metres?

It is:

Does this CRS represent distance adequately in the area and at the scale of my analysis?

Metres in the CRS do not guarantee accurate distance

Consider two projected CRSs that both use metres.

A point buffered by 10,000 map units will receive a 10,000-metre planar buffer in either one if the tool performs ordinary Cartesian geometry buffering.

But the two projections may have different scale distortion at that location.

As a result, those 10,000 map metres may correspond differently to actual distance on the Earth's surface.

This is why the coordinate unit alone is not enough to judge a CRS.

A suitable local or regional projected CRS is usually designed so that distortion remains acceptably small within its intended area of use. Take the same CRS far outside that area and its measurements may become much less appropriate.

The EPSG dataset therefore records not only the mathematical definition of a CRS but also its scope and area of use. Those are operationally important when the CRS is being used for measurement rather than merely display.

Why Web Mercator deserves particular caution

Web maps commonly use EPSG:3857, also called Web Mercator or WGS 84 / Pseudo-Mercator.

Its coordinates are expressed in metres, which makes it tempting to assume that it is a good CRS for any calculation involving metres.

It is not.

The EPSG registry identifies the scope of EPSG:3857 as web mapping and visualisation. It also documents scale differences relative to an ellipsoidal Mercator implementation.

The practical problem is that Web Mercator scale distortion increases with latitude. A map-space buffer that appears to extend a fixed number of projected metres is therefore not a generally reliable model of a fixed ground distance across the world.

That does not make EPSG:3857 a bad CRS. It makes it a CRS designed for a different primary purpose.

If the analysis requires defensible physical distances, choose the measurement method and CRS for that purpose rather than using Web Mercator simply because the basemap happens to be displayed in it.

The project CRS and the analysis CRS are not necessarily the same thing

Modern GIS applications can display several layers with different coordinate reference systems together.

A layer stored in WGS 84 longitude/latitude can appear perfectly aligned with a Web Mercator basemap because the GIS transforms coordinates for display.

That visual alignment does not tell you which coordinate system a particular processing algorithm will use for its calculations.

Depending on the application and operation, processing may use:

• the input layer's CRS;

• a specified output or analysis CRS;

• the project CRS;

• an internally chosen projected CRS;

• geodesic or ellipsoidal calculations.

This is why changing the map display projection should not be assumed to change the mathematical behaviour of every buffer operation.

Likewise, seeing two layers align on screen does not prove that a numeric buffer distance will be interpreted in the units you intended.

The CRS used for display and the reference system used for analysis are related concerns, but they are not always the same setting.

Assigning a CRS is not the same as reprojecting before buffering

Suppose a longitude/latitude dataset is incorrectly labelled as though its coordinates were already in a metre-based projected CRS.

Changing the CRS label does not convert:

11.52°, 3.87°

into projected metre coordinates.

It merely changes how the software interprets those numbers.

If the data really need to be represented in another CRS, the coordinates must be transformed.

This distinction is especially important before a distance-based operation. Incorrectly assigning a projected CRS just to make a buffer tool display metres can move the data conceptually to the wrong places and invalidate the result.

Assigning a CRS vs Reprojecting Data covers that distinction in detail.

For buffering, the practical rule is simple:

Do not change a CRS label to obtain convenient units. Transform the geometry when a different coordinate representation is genuinely required.

A suitable local projection can make planar buffering straightforward

Suppose you want a 2-kilometre buffer around facilities in a relatively compact study area.

A reasonable workflow may be:

1. identify a projected CRS appropriate to the study area;

2. transform the input geometry into that CRS if necessary;

3. confirm that its coordinate unit is metres;

4. create a buffer with distance 2000;

5. retain or transform the result as needed for later use or publication.

The important phrase is appropriate to the study area.

A locally suitable UTM zone, national grid, regional CRS, or other low-distortion projection may be appropriate depending on the geography and required precision.

No single projected CRS is universally best for buffering.

A system suited to Cameroon is not automatically suitable for Canada, and a projection designed for one UTM zone becomes increasingly inappropriate as the analysis extends far beyond that zone.

What if the study area crosses projection zones?

Local projections become more complicated when the analysis covers a large area.

Suppose features extend across:

• several UTM zones;

• an entire continent;

• both sides of the antimeridian;

• high polar latitudes;

• the whole world.

There may no longer be one simple local projected CRS that keeps distance distortion small everywhere.

Using one UTM zone for a continent-wide buffer, for example, extends that projection well outside the region for which it was intended.

Possible alternatives include:

• processing subsets in appropriate regional projections;

• choosing a projection designed for the larger area and acceptable distortion requirements;

• using a geodesic/geography-aware distance implementation;

• reconsidering whether a uniform buffer is the appropriate model for such a broad analysis.

The correct solution depends on the purpose and required accuracy.

The lesson is not “always use UTM”. It is match the distance method to the geographic extent.

Some software can buffer geographic data intelligently

Not every GIS buffer on longitude/latitude data simply treats degrees as Cartesian units.

Spatial systems can provide geodesic or geography-aware operations.

PostGIS is a useful example.

For geometry, ST_Buffer works in the coordinate units of the geometry's spatial reference system.

For geography, the distance is expressed in metres. PostGIS then chooses a planar spatial reference system suited to the geography's bounding box, performs the buffer calculation there, and transforms the result back to WGS 84.

That makes many ordinary metre-based geographic buffers convenient.

But the PostGIS documentation also warns that this approach may not behave as desired for geometries much larger than a UTM zone or objects crossing the dateline.

So geography-aware buffering reduces the need for users to select a projection manually in some workflows, but it does not abolish the underlying geometric problem.

Large extents and difficult geographic configurations still require judgement.

Planar and geodesic distance are different models

Imagine two points on the Earth's surface.

A planar calculation measures distance after representing those locations in a flat coordinate system.

A geodesic calculation considers their separation on a curved Earth model, commonly an ellipsoid.

For a small region represented in a suitable local projection, planar and geodesic results may be sufficiently close for the intended use.

Across large distances, the difference can become much more important.

This leads to a useful distinction:

CRS choice matters because it affects the planar geometry. Distance model matters because not every workflow needs to be planar in the first place.

A GIS system that offers a geodesic buffer may therefore produce an appropriate fixed-ground-distance result from geographic input without requiring the user to permanently reproject the source layer.

But that does not mean planar buffering is wrong. It means the analytical model should be chosen deliberately.

Why buffers can look different after reprojection

Suppose you create a planar buffer in one projected CRS and then transform the resulting polygon into another CRS.

Its appearance may change.

A shape that looked circular around a point in the analysis projection may look stretched or otherwise distorted when displayed in another projection.

That does not necessarily mean the buffer became incorrect during reprojection.

The first projection was used to define the distance relationship. The second changes how the resulting geography is displayed on a plane.

This is the same fundamental reason that world maps can make the same geographic feature look different under different projections.

The important question is where and how the distance criterion was defined—not whether the final polygon looks perfectly circular in every map projection.

A circular-looking buffer is not proof of correct distance

The reverse is also true.

A buffer may look beautifully circular on screen and still represent the wrong ground distance.

If the operation treated longitude/latitude degrees as Cartesian units, or used a projection with substantial scale distortion in the study area, visual symmetry tells you very little about measurement accuracy.

Maps display coordinate geometry. They do not automatically reveal how map units relate to ground distance.

A useful buffer check therefore includes numbers, not just appearance.

Ask:

• What CRS did the operation use?

• What are its coordinate units?

• What distance value was entered?

• Is the CRS appropriate for the location?

• Is the operation planar or geodesic?

• Does an independent distance check confirm the expected radius?

How to diagnose a suspicious buffer

Suppose you intended a 5-kilometre buffer, but the result is obviously too large or too small.

Start with the units.

If the coordinates look like longitude and latitude

Values such as:

11.5, 3.8

suggest geographic coordinates in degrees.

Check whether the buffer operation is interpreting the distance in degrees or providing geodesic handling.

If the coordinates are large projected numbers

Values such as:

552000, 428000

may indicate projected coordinates, but do not assume the units are metres. Inspect the actual CRS definition.

If the units are metres but the result is still questionable

Check whether the projection is suitable for the study location.

A metre-based CRS can still have substantial scale distortion outside its intended area.

If the map looks right but measurements are wrong

Check whether on-the-fly reprojection is hiding a mismatch between the display CRS and the coordinate system used by processing.

If the dataset's CRS is missing or uncertain

Do not experiment by assigning different CRSs until the buffer looks plausible.

First identify the actual coordinate system from the dataset's metadata, coordinate values, extent, and provenance.

A plausible-looking buffer is not proof that the underlying interpretation is correct.

Choosing a CRS is also choosing where distortion is acceptable

Every projection is a compromise.

For buffer analysis, distance or scale distortion is especially relevant because the operation is explicitly defined by distance.

A useful projected CRS should keep that distortion within an acceptable range for:

• the geographic area;

• the buffer distance;

• the purpose of the analysis;

• the required precision.

A simple exploratory map may tolerate more distortion than cadastral, engineering, scientific, or regulatory analysis.

This is why there is no universal threshold saying that a particular CRS is always accurate enough.

The acceptable error follows from the use case.

The buffer distance should be documented with its measurement context

A saved layer named:

facility_5km_buffers

looks self-explanatory.

It is not.

For reproducibility, useful metadata may need to record:

• buffer distance: 5;

• unit: kilometres;

• input CRS;

• CRS or distance model used for calculation;

• whether the operation was planar or geodesic;

• whether features were reprojected;

• whether buffers were dissolved;

• relevant software or algorithm where behaviour matters.

That information becomes especially important when the result is exported into another CRS after analysis.

The final layer may contain longitude/latitude coordinates, even though its 5-kilometre geometry was originally constructed using a projected metre-based CRS.

Without provenance, a later user cannot tell how the distance was established.

A practical rule for buffer analysis

For a relatively small local or regional study area, a good default is often:

Use a projected CRS designed for that area and appropriate for distance measurement, then buffer using its known linear units.

For larger, cross-zone, global, or otherwise difficult geographies:

Consider a geodesic or geography-aware operation, or divide the analysis into regions with appropriate projections.

And in either case:

Do not infer suitability merely from the fact that the CRS uses metres.

Those rules are more reliable than shortcuts such as “always use UTM”, “never buffer WGS 84”, or “Web Mercator uses metres, so it is fine”.

Each of those shortcuts ignores an important part of the problem.

The CRS determines what the number means—and the projection affects how well it measures

A buffer begins with a deceptively simple instruction:

distance = 1000

But that number sits inside a spatial model.

The coordinate reference system helps determine the units in which planar geometry is expressed. The projection determines how the curved Earth has been represented on that plane. The software determines whether it uses those planar coordinates directly or performs a geography-aware calculation.

That is why buffer distance depends on more than the value typed into the tool.

Before trusting a buffer, establish:

What are the units?

What CRS or distance model is being used?

Is it appropriate for this location and extent?

How much measurement distortion is acceptable for the intended use?

Once those questions are answered, the buffer distance becomes meaningful rather than merely numerical.

References

1. QGIS Documentation — Vector Spatial Analysis (Buffers). Explains that buffer distances are expressed in map units according to the CRS and identifies metres, feet, and decimal degrees as possible coordinate units.

2. PostGIS — ST_Buffer. Documents that geometry buffers use the units of the spatial reference system while geography buffers use metres, and explains the internal planar handling and limitations of geography buffering.

3. EPSG — WGS 84 / Pseudo-Mercator (EPSG:3857). Defines Web Mercator as a projected CRS intended for web mapping and visualisation and documents its scale limitations, supporting the distinction between a metre-based display CRS and a CRS chosen for accurate measurement.