How Many Decimal Places Do Coordinates Need?

There is no single correct number of decimal places for every coordinate, because the right amount depends on the coordinate system, the accuracy of the source, the calculation being performed, and how the location will be communicated or published.

For decimal-degree latitude, five decimal places represent an increment of roughly one metre and six represent roughly a tenth of a metre. Those figures describe numerical resolution, not positional accuracy. A geocoded address can be stored to six decimals and still be tens of metres from the intended entrance. A surveyed point can be accurate to centimetres even if a public display deliberately rounds it.

The practical rule is therefore: preserve enough numeric precision for the workflow, but do not use displayed digits to claim more geographic certainty than the source supports.

What each decimal place means in geographic degrees

One degree of latitude corresponds to roughly 111 kilometres. Dividing that angle by powers of ten gives a useful order-of-magnitude guide:

Decimal places · Approximate latitude increment

0 — 111 km
1 — 11.1 km
2 — 1.11 km
3 — 111 m
4 — 11.1 m
5 — 1.11 m
6 — 0.111 m
7 — 0.011 m

These values are approximate and apply most directly to latitude. Longitude behaves differently because meridians converge towards the poles: near the equator, one degree of longitude is also roughly 111 kilometres, but at 60 degrees latitude, the east-west distance it represents is about half that. A fixed number of longitude decimals therefore does not correspond to one constant ground distance everywhere on Earth.

That variation is one reason decimal-place tables should be used as a scale guide rather than as an accuracy certificate.

Resolution is not accuracy

Suppose an address geocoder returns:

4.051237, 9.767868

Six decimal places make the coordinate numerically capable of distinguishing sub-metre increments in latitude, but the geocoder may have placed the address at a parcel centroid, a building centroid or an interpolated point along a road. The real uncertainty could therefore be several metres or much more.

The same issue appears with calculated coordinates. A polygon centroid can contain many decimal places because the mathematics was carried out with high numeric precision. That does not mean the source polygon was surveyed to the same precision.

If the distinction between those concepts is unclear, Coordinate Precision vs Coordinate Accuracy should be treated as the foundation. Decimal places describe how finely the number is expressed; accuracy describes how well the location corresponds to an appropriate reference.

Match the displayed precision to the source

Different sources justify different levels of apparent detail.

A high-quality survey tied to an appropriate reference frame may support centimetre-level coordinates. A phone location may vary by several metres depending on conditions. An address geocoder may identify a building, parcel, street segment or locality rather than a measured point. A place extracted from a narrative report may only be known within a neighbourhood or district.

Printing all of those locations to eight decimal places makes them look more comparable than they are.

A good display precision should answer a communication question: at what spatial scale is this position actually meaningful?

For a national overview of cities, three or four decimal places may already be more than enough. For engineering or survey work, decimal-degree display may not even be the most appropriate representation because the workflow is likely to use a projected or geodetic system designed for precise measurement.

Storage precision and display precision are different decisions

It is often sensible to keep more numeric precision internally than you show to a reader.

Repeatedly rounding coordinates during processing can introduce small shifts, especially when values are transformed between coordinate systems or used in calculations. Preserving the original numeric value reduces avoidable rounding loss and makes the source easier to audit.

The public or human-facing representation can still be rounded.

For example:

stored value:    4.05123741
published value: 4.05124

That does not mean the published value is more accurate. It means the presentation has been chosen at an appropriate level of detail while the canonical value remains available for computation.

The same principle is useful when exporting data. If downstream users need to reproduce an analysis, over-aggressive rounding in the export may be harmful even if the map label itself only needs a few decimals.

Projected coordinates need a different interpretation

The familiar decimal-degree table does not apply to coordinates measured in metres or feet.

If a projected CRS uses metres, a coordinate such as:

523421.27 m

has a numerical resolution of centimetres because the second decimal place represents hundredths of a metre. Once again, however, the source may only be accurate to one metre, ten metres or more.

The coordinate unit must therefore be known before decimal places mean anything. Three decimals can represent thousandths of a metre, thousandths of a foot, or thousandths of a degree depending on the coordinate system.

Do not choose precision from the shape of the number alone.

Decimal places do not substitute for an uncertainty field

If positional quality matters, report it directly when the data supports an estimate. A record such as:

latitude: 4.05123
longitude: 9.70452
horizontal_uncertainty_m: 15

communicates two different things: the coordinate values show how finely the position is written, whereas the uncertainty field describes something about the range of plausible positions. Adding more decimal places to the latitude and longitude would not provide the same information.

Darwin Core makes this separation explicit through distinct terms for coordinatePrecision and coordinateUncertaintyInMeters. The example comes from biodiversity data, but the principle is general: numeric resolution and positional uncertainty should not be collapsed into one field or inferred from one another.

Do not round too early in an analytical workflow

Presentation rounding and analytical rounding have different consequences.

Rounding a display coordinate from six decimals to five is usually harmless when the reader only needs an approximate point. Rounding the coordinates of every vertex in a detailed geometry before analysis can alter lengths, areas, topology or whether features touch.

Likewise, coordinates used as join keys should not be casually rounded unless the matching rule has been designed around that tolerance. Two independently measured points that represent the same place may differ slightly, while two different points may become identical after coarse rounding.

When precision affects geometry or matching, treat it as part of the method rather than as cosmetic formatting.

Privacy can require less precision than accuracy would allow

A technically accurate coordinate is not always appropriate to publish.

Locations can reveal homes, vulnerable people, sensitive facilities, endangered species or other information whose precise disclosure creates risk. In those cases, reducing displayed precision or generalising the geometry can be one layer of protection.

But simple decimal rounding should not be mistaken for a complete anonymisation method. The appropriate treatment depends on what can be inferred from the released data, other attributes and external datasets.

The broader concept of geolocation uncertainty is useful here because an intentionally generalised public location and an imprecisely observed location may both appear less exact, but for very different reasons. The metadata should preserve that distinction where it is safe to do so.

A practical way to choose decimal places

Instead of asking "How many decimals does GIS use?", work through five questions.

1. What are the coordinate units?

Degrees, metres and feet cannot be interpreted with the same decimal rule.

2. How was the location obtained?

Survey, GPS, geocoding, digitising, manual entry, place-name matching and calculated representative points have different accuracy characteristics.

3. What error is acceptable for the task?

A few metres may be irrelevant for a continental map and unacceptable for a property boundary.

4. Will rounding change the analysis?

If the values feed transformations, topology, joins or detailed geometry, preserve sufficient precision internally.

5. What should the audience be allowed to infer?

Avoid a display that visually promises centimetre-level certainty for a location that is only known approximately. Also consider privacy and sensitivity.

These questions often lead to different answers for storage, analysis, display, and publication. That is normal.

More digits are useful only when they carry useful information

Modern databases can store long coordinate values cheaply, so there is little reason to round source coordinates repeatedly during computation. The problem is not that long numbers exist, but that every digit may be treated as evidence.

A coordinate's meaningful precision comes from the process that created it and the use that follows. Keep enough detail to avoid introducing material error, retain the original value when provenance matters, and choose the displayed precision deliberately.

The best coordinate is not the one with the longest decimal tail. It is the one whose numeric representation, documented accuracy and intended use agree with one another.

References

Darwin Core terms. Defines coordinate precision separately from coordinate uncertainty, illustrating why decimal representation should not be treated as positional accuracy.