[PROJ] OGC blog post summarising Web-mapping misalignment problem

[Martin Desruisseaux]

Le 22/08/2019 à 10:58, Joaquim Luis a écrit :

> Hello, with the risk of going a bit off topic I have another question
> about the different WGS84, which is: why are they different? They all
> use the same ellipsoid, right? So the difference is in the origin of
> the referencing system? And if yes, why does it change in the several
> WGS84 realizations?
>
The origin move slightly between realizations. But in addition of that
translation, the axes may also have a slight rotation and a slight scale
factor (I did not verified if it was the case for WGS84 realizations).
Those changes exist both because of improvement in the accuracy of the
measurements used for defining the WGS 84 reference frame, but also
because datum defined by satellites have their origin at the center of
mass of Earth and that center of mass moves continuously (because of
plates tectonic, convection movements in Earth mantle, etc.).

I think that users looking for stability should use the reference frame
defined by their country instead than any satellite-based datum. For
example for locating points in Australia, use the CRS defined by
Australian mapping agency. National CRS are defined respective to the
continental plate where the country is located. Even if they define a
relationship between national CRS and satellites-based CRS, they take in
account that their continental plate does not move in the same way than
the Earth center of mass and they provide more accurate transformations
between old and new CRS than if we were using the satellite-based CRS. I
would suggest to reserve the use of satellite-based CRS like WGS84 to
the cases where data are already in that CRS anyway (e.g. GPS data), or
span a geographic area too wide for being expressed with a national CRS.

    Martin