Understanding GPS Coordinates: Formats, Datums and Precision

Every point on Earth can be described with two angles: latitude (north–south of the Equator) and longitude (east–west of the Prime Meridian). That sounds simple, but coordinates show up in several notations, and mixing them up is a common source of 'my marker is in the ocean' errors.

Decimal degrees vs. DMS

Decimal degrees (DD) write the angle as a single number: 40.7128, −74.0060. Degrees-minutes-seconds (DMS) splits each degree into 60 minutes and each minute into 60 seconds: 40°42'46"N, 74°00'22"W. Both can describe the same point. Software commonly uses DD; charts, legal descriptions and older GPS workflows may use DMS.

• Negative latitude = south of the Equator; negative longitude = west of Greenwich in the usual signed convention.
• DMS hemisphere letters (N/S/E/W) carry the direction, so do not combine them with a conflicting sign.
• Degrees and decimal minutes (for example 40°42.768') are another format often seen in aviation and marine GPS workflows.

How precise is precise?

One degree of latitude is about 111 km, which gives a useful rule of thumb for the numeric spacing represented by decimal places:

• 3 decimals ≈ 111 m of latitude
• 4 decimals ≈ 11 m
• 5 decimals ≈ 1.1 m of numeric precision
• 6 decimals ≈ 0.11 m of numeric precision — usually finer than consumer positioning accuracy

Datums: WGS84 is not the only one

A geodetic datum defines the reference frame used to interpret coordinates. Modern GNSS and web-mapping workflows are commonly expressed in WGS84-compatible geographic coordinates, while older or local datasets may use systems such as NAD27 or NAD83. The same physical point can therefore have different numeric coordinates in different datums.

Practical checklist

• Store coordinates with enough decimals for your workflow, but keep source accuracy metadata when it matters.
• Validate ranges: latitude −90…90, longitude −180…180.
• Watch for swapped lat/lng — GeoJSON uses [longitude, latitude] order.
• When copying coordinates from another service, verify the format and datum instead of assuming all systems use the same convention.

A real-world example: one place, several coordinate formats

Take a point in central New York City: 40.7128, -74.0060 in decimal degrees. The same location can be written in degrees-minutes-seconds, converted to UTM for a projected grid workflow, or expressed in MGRS for grid-based navigation. None of those formats makes the underlying observation more accurate; they are different ways to encode the same location. The safest workflow is to preserve the original source coordinates, record the datum when known, and convert only for the system that needs a different notation.

Common coordinate mistakes that look plausible

• Latitude and longitude swapped: a valid-looking pair can still land on another continent.
• Missing minus sign: western longitudes and southern latitudes can jump across the globe.
• DMS pasted into a decimal field: symbols and hemisphere letters may be parsed incorrectly or ignored.
• Wrong datum: older survey coordinates can shift when interpreted as modern WGS84 without transformation.
• Too much trust in decimal places: seven decimals on screen do not prove centimetre-level accuracy.

How many decimals should you actually store?

For general web mapping, five or six decimal places is usually a sensible storage choice because it preserves metre-scale numeric detail without pretending the source measurement is perfect. If the source is a phone fix uncertain by 8 metres, storing six decimals is harmless as long as you also understand the uncertainty. For surveying, engineering or legal boundary work, coordinate precision must be tied to the required datum, instrument, control network and professional workflow rather than a generic web rule.

A coordinate is only as trustworthy as its source

When somebody sends you a pair of coordinates, the first useful question is not how many decimals they contain. It is where they came from. A point copied from a phone GPS fix, a survey drawing, a geocoded street address and a hand-clicked map marker can all be written to six decimal places, yet they represent very different levels of confidence. The digits describe the stored number; the source describes the evidence behind it.

A geocoder usually returns the location associated with an address record or mapped feature. That might be a building centroid, a parcel approximation, an entrance, a road interpolation or another provider-specific point. A phone fix is a sensor estimate with an uncertainty radius. A survey coordinate may be tied to a formal control network and datum. Treating those inputs as interchangeable is how a clean-looking spreadsheet becomes misleading.

If you only have an address, start with the Address to Coordinates tool, then inspect the returned pin instead of copying the first number blindly. If you already have a coordinate and want human-readable context, use Coordinates to Address as a separate reverse-geocoding step. Keeping those two operations conceptually separate makes data cleanup much easier.

Why longitude precision changes with latitude

The familiar 'five decimals is about a metre' rule is only a rough latitude-based shortcut. Lines of latitude stay roughly parallel, but lines of longitude converge toward the poles. One degree of longitude spans about 111 km at the Equator, around 79 km at 45° latitude, and much less close to the poles. That means the east-west ground distance represented by a decimal place depends on where you are.

For ordinary map work this does not mean you need a custom decimal rule for every location. It means you should avoid presenting decimal digits as a universal accuracy guarantee. Store enough numeric detail to avoid introducing extra rounding error, then carry the real-world accuracy or source quality separately when it matters.

A practical QA routine before importing thousands of points

• Check the minimum and maximum latitude/longitude values. One outlier can reveal swapped columns or a missing minus sign.
• Plot a small sample on a map before importing the full file into another system.
• Look for suspicious duplicates: many identical coordinates can mean a geocoder fell back to a city centroid.
• Keep the original text fields alongside normalized numeric coordinates until the dataset is verified.
• Record the coordinate reference system or datum when the source is anything other than a normal consumer web map.

For batch work, the CSV Coordinate Validator is useful for catching range and formatting problems before mapping, while the CSV Coordinate Deduplicator can help identify repeated points that deserve a closer look.

When coordinate conversion becomes a professional problem

Most casual web workflows live comfortably in latitude/longitude on WGS84-compatible maps. Problems become more serious when coordinates come from cadastral surveys, engineering drawings, local grid systems, historic maps or legacy government datasets. A datum label that looks like minor metadata can correspond to a meaningful positional shift. In those cases, a proper coordinate transformation is not the same thing as changing notation from DMS to decimal degrees.

If money, safety, legal ownership or construction depends on the result, confirm the source CRS and use a transformation appropriate to that dataset. MapBench is useful for inspection and everyday conversions, but it should not replace the formal survey or GIS workflow required by the decision.