I have heard people referring to as a horizontal datum to models such as WGS84 and as a vertical datum to geoids. The former is primarily used for determining latitude and longitude but also the height (and thus X, Y Z can be calculated) above the ellipsoid while the latter is used for determining height above an equal-potential surface only.

I wonder why one would call "horizontal" a full 3D datum. The fact that you have a pure vertical datum (i.e. geoid) does not mean you have to call "horizontal" the remaining one.

Here are two articles where you can find the term:

http://geography.about.com/od/geographyintern/a/datums.htm http://www.esri.com/news/arcuser/0703/geoid1of3.html


The term horizontal datum is used because it is more easily flattened into 2 dimensions and more useful for finding locations on a flat plane (compared to a vertical datum).

As in the ESRI post you referenced (image below), the Earth's surface is very uneven, so modeling this very difficult. The "ellipsoid" used in horizontal datums is close approximation of Earth's surface. While you can use a horizontal datum to determine height simply because it is a 3D model, it is not as useful as a geoid (used in vertical datums) which has some "physical" meaning because it approximates mean sea level fairly accurately. Therefore, heights are better represented or estimated with a geoid because they represent a physical estimate of mean sea level. Maybe someone else can explain it better..

enter image description here

  • In a simplistic sense I guess that a geoid is useful because it defines what is up and down hill. That is something that could have real implications. Two elevations measured from an ellipsoid could have the same value, but require travel up or down hill (I think). Perhaps an ellipsoid could also be changed to take account of seismic activity without needing to change elevation values. – Matthew Snape Oct 9 '13 at 15:09
  • Sounds good, except for the "estimated relatively easily". I recall in early '90s hearing a surveyor explain how they needed to collect sea level measurements over a very long time before they could establish an elevation benchmark in a developing country. – Kirk Kuykendall Oct 9 '13 at 15:38
  • All valid points, I've updated my answer. – SoilSciGuy Oct 9 '13 at 16:48
  • 1
    Geoid isn't based on MSL; it's based on gravity potential and it closely fits MSL (in a least squares sense). – Paul Oct 9 '13 at 17:04

A "horizontal datum" references the 2D horizontal portion of a set of coordinates. Take these coordinates:

Latitude: 29.27452 North

Longitude: 102.32512 West

Elevation: 110 meters

If I said "The horizontal datum is WGS84" and that's all, then I've introduced ambiguity in my coordinates. And if there's one thing GIS professionals hate, it's ambiguity. Horizontal datums by definition give no reference to height, so is my elevation from the ellipsoid? Above the Earth's surface? Mean sea level?

Now, I could have said that these coordinates are in EPSG:4979 - WGS84 / Geographic 3D CRS, and that explicitly states that height is relative to the ellipsoid, kind of like the coordinates you would see on your GPS display. But here's the catch: the height on your GPS receiver is probably wrong.

The raw coordinates you get from satellite GPS calculation is actually WGS84 geocentric 3D Cartesian XYZ coordinates (EPSG:4978) that your receiver turns into latitude, longitude, and ellipsoidal height (via conversion EPSG:15592) if you're lucky. Too bad nobody cares about ellipsoidal height, because what people think of when they hear the word "elevation" is distance above sea level, or orthometric height. If you're really lucky, the GPS receiver will have some sort of geoidal matrix built in, and will try to guess your orthometric height based on your ellipsoid height and the geoid tables. But there's no guarantee it got your ellipsoidal height correct to begin with, as height is the weakest component of GPS. The standard deviation of height is worse than the standard deviation of the horizontal by a factor of 10!

Long story short, (too late!) GPS sucks at figuring out elevation, and nobody wants ellipsoidal height. If you really care about the precision of your elevations, you will give a horizontal datum for the latitude and longitude, and a vertical datum for the height.


After having read the answer of SoilSciGuy, and the comments of it, I thought more about a possible reason why a 3D datum should be called horizontal. Height can be accurately defined using an ellipsoid, but their reference (height = 0) cannot be physical. Now, is this a problem? Well, during the time when there was no GPS, it was because surveyors had to use the sea level and physical benchmarks inland (derived from the sea level) as a reference for their height measurements. There were also physical benchmarks for determining x and y, but these benchmarks are also fading out with the take over of GPS technology.
So, "horizontal" seems merely a historical term. But, you can still argue by saying that as soon as there are still surveyors who use traditional methods to determine x,y or h, you can continue calling horizontal a 3D datum.


Best way to remember the difference is to use this RILT(Remember It Like This)

The horizontal-wide-lines(latitude) are similar to gym-freaks and bodybuilding building their WIDE/Horizontal-wide latts :)


  • 2
    First of all, the post is not asking what latitude is and how to remember it. Secondly, latitude is not a line. It is an angle. – multigoodverse Mar 30 '15 at 8:58
  • @ArditSulce YES SIRRR !!! – user914584 Mar 31 '15 at 9:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.