[Proj] Proj4 Datum Grids

[Martin Desruisseaux]

Hello

Le 19/04/2018 à 09:57, support at mnspoint.com a écrit :

> yes, that is exactly the idea in Proj .. if the datum is not Proj
> predefined then you should use the towgs84 statement.
>
Indeed, but just to nuance, the Helmert transformation (as done by the
+towgs84 parameters) is still an approximation not equivalent to using
datum shift grids. Furthermore the question was if one "can perform any
datum shift with the appropriate +towgs84 entry", which is not always
true. In some cases (e.g. from Martinique 1938 to RGAF09), the
transformation needs to be expressed relative to another datum than WGS
84. This is the reason why TOWGS84 has been removed in WKT 2 [1] (a.k.a
ISO 19162) and replaced by a new construct - BOUNDCRS [2] - which allows
to specify the datum in addition of resolving other ambiguities (e.g.
the sign of rotation parameters).

    Martin

[1] http://docs.opengeospatial.org/is/12-063r5/12-063r5.html#144
[2] http://docs.opengeospatial.org/is/12-063r5/12-063r5.html#125