[Qgis-user] GDA94 and WGS84 transformations

[Greg Troxel]

Stephane Goldstein <s.n.g at gmx.com> writes:

> I'm having trouble understanding how QGIS handles the conversion from the
> Australian datum to WGS84 in the UTM grid.
>
> GDA94 (EPSG:4283) /  MGA94 Zone 5X (EPSG:2835x)
> WGS84 (EPSG:4326) /  UTM South     (EPSG:3265x)

The basic issue here is that WGS84 is a fuzzy notion.  There are
actually multiple datums with the name WGS84, including one that is
often called "original WGS84" or some similar name.  Then there are
several of the form WGS84(GNNNN) where G1762 is the current one.

When people say WGS84 without being specific, or they use a codepoint
like 4326, they are saying, whether they realize it or not:

   I have coordinates in one of the realizations of WGS84 and I don't
   know which.

This leads to the "several meters of fuzz" notion.

Note that WGS84(G1762) is equivalent to modern ITRF (2005, 2014) at the
sub-cm level, maybe a few mm.

I have coordinates in NAD83(2011).  I suspect this is a bit like the
modern version of GDA94.  However NAD83 has multiple realizations also.

I also have coordinates in ITR2014, determined by 30h GPS observations
against CORS.

If I hand-put those in a gpx and don't label them, so they are treated
as WGS84 unspecified,, and plot them against NAD83(2011) in qgis with
the project CRS set to NAD83(2011), then they don't line up.

If I label the gpx as ITRF2014, then I get the transformation I want.

So basically the answer is that if you are being precsise to more than a
few meters, you should not be using the WGS84 datum family codepoints
*at all*, but only codepoints for specific realizations of individual
datums within datum families.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 194 bytes
Desc: not available
URL: <http://lists.osgeo.org/pipermail/qgis-user/attachments/20201001/465e2a07/attachment.sig>