I am not interested in converting ellipsoidal height into AHD heights, because I have a good DEM. However, this whole business about geocentric vs geodesic coordinate systems is thoroughly confusing me.

My points are recorded in WGS84, and included xyz information (latitude, longitude and ellipsoidal height). I am wondering whether there is some pre-processing I need to do to convert this xyz data into just xy, or whether just extracting xy information (latitude and longitude) will give me the accurate position for the GPS point recorded.

I am sorry if this question is v basic, but after having tried to research this myself, I have only become more confused.

Perhaps I can narrow it down further - I think the key is to reproject from ESPG:4978 to ESPG:4326 (the 3D and 2D forms of WGS84). Basically what I want is to use the xyz coordinate information to return accurate xy coordinates.

Is this necessary (I have a feeling it is) and if so, any tips for how one goes about this?

  • What coordinate reference system is DEM in? What's its resolution? How accurate are the WGS84 coordinates? – mkennedy Sep 26 at 16:38
  • @mkennedy The DEM is in SWEREF 99 TM, at a 2X2m resolution. The GPS point coordinates are very accurate, to the milimeter scale supposedly (used one of those land surveyor GPS receivers that are mounted on a tripod, not sure what they are called) – Visithuru Sep 26 at 19:04
  • I was misled by the reference to AHD which I took to be Australian Height Datum! As @neogeomat says, it should be fine. – mkennedy Sep 26 at 20:26

The ellipsoidal height does not affect the latitude longitude. so only extracting xy from xyz should be fine.


If you have longitude, latitude and ellipsoidal height, you have coordinates over an ellipsoid (the WGS84 ellipsoid in your case) and related to a coordinates frame (the WSG84 geodesic frame in your case). Imagine that, a single ellipsoid among all possible, with its center, axes (ellipsoid axes) and origin of coordinates (center meridian) positioned and oriented in a particular position and orientation, among all possible.
They are coordinates in the WGS84 geodesic reference frame.

Longitude and latitude of a point in space are the coordinates of its normal projection towards an ellipsoid in some reference frame. The ellipsoidal height is the distance between the point and the ellipsoid, measured on that normal direction.
All points of space located on the same normal line, have equal latitude and longitude.

The ellipsoid is a mathematical reference surface that we use to be able to perform calculations between points in space through normal projections to that curved surface. It makes sense from, and only until, its shape is similar to that of Earth.

On the other hand, if you have X, Y and Z, coordinates of a Cartesian system of orthogonal axes to each other, what you have are the coordinates of a point in space, typical three-dimensional coordinates that we use to locate any point in a Cartesian system. Where the center of that system is located, how the axes are oriented and what scale of measurements are used, define the frame of reference of that system.
There is also a geocentric Cartesian reference frame called WGS84.

The position of the GPS network satellites, for example, is referenced to that frame. Calculations of satellite positioning inside of the satellite receivers are generally referred to that Cartesian framework.
However, it is very common that ,within or outside the receiver, there is a processing of the location that immediately converts it from geocentric X, Y, Z to geodetic Latitude, Longitude, Ellipsoidal Height, both in frames called in the same way: WGS84. This is so, because it is much easier for humans to know where we are if they tell us our longitude, latitude and ellipsoidal height than if they tell us our distance to a list of three geocentric Cartesian axes. But it is easier for the recivers to calculate our position the other way.

We can realize what kind of coordinates we drive: some are millions of meters away from the center of the earth, others are a small amount of degrees away from a central meridian and from the equator.

To finish clarifying the situation, or confusing it, there is also a kind of projection to the plane of a geodetic coordinate system, also called WGS84: the famous EPSG:4326.
It converts Longitudes into X coordinates and Latitudes into Y coordinates of a flat and orthogonal coordinate system, centered where the point of the space corresponding to Longitude 0 degrees, Latitude 0 degrees and Ellipsoidal Height 0 meters (or all points projected on it) on the WGS84 geodetic frame (which includes in its definition the WGS84 ellipsoid).

Therefore, if you are sure of having longitude and latitude coordinates, in that order, referenced to the WGS84 geodetic reference frame, you can treat them as if they were X and Y, respectively, from a flat orthogonal coordinate system and you would have your points automatically projected in EPSG:4326. And you could get rid of ellipsoidal height without any inconvenience.

If instead, you have millions of meters in your coordinates, it is very likely that you are handling three-dimensional coordinates of a Cartesian system referred to the geocentric WGS84 frame, and you will have to reproject them without getting rid of any of them.

  • "All points of space located on the same normal line, have equal latitude and longitude." but consider negative altitudes. on a polar flattened ellipsoid, surface normals intersect the equatorial plane away from (x=0,y=0,z=0) . Points on the internal equatorial plane could be from equatorial points (lat=0) or points on the intersecting normal (lat != 0) - unlike geocentric coordinate altitudes which all intersect at (x=0,y=0,z=0). Its a degeneracy that rarely affects practice however.... – Spacedman Sep 27 at 8:01
  • 1
    You are right professor. In fact, the normal ones to the ellipsoid are full of intersections between them. Probably the geodetic coordinate system is equally plagued with ambiguities, of which we do not realize because in practice we use a relatively small range of ellipsoidal heights. – Gabriel De Luca Sep 27 at 9:45

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.