[Qgis-user] Reproject from WGS84 (EPSG: 4326) to ETRS89/UTM32N (EPSG: 25832) considering the epoch

[Greg Troxel]

Jochem Kail via QGIS-User <qgis-user at lists.osgeo.org> writes:

>> It is a major problem in GIS that data are mislabeled as being in WGS84
>> and that it is treated as an interchange format.  Both practices are
>> errors, but also the standard approach.
>
> In summary, I now do suspect, that the data from 2011 were mislabeled
> as being WGS84. Based on what I do understand so far, the ITRF
> reference frame in 2011 was ITRF2000, which corresponds to the WGS 84
> (G1150) realization. However, most likely the RTK network has been
> used as reference stations. Therefore, the geographic coordinates are
> most likely based on the ETRS89 reference system (not WGS84) and given
> that the data have been mapped in 2011, based on the ETRF realization
> ETRF2000.

You said "DGPS" earlier and now you are saying "RTK".  These are not the
same thing.   To get this right, you need to be clear on what was done
before, from what reference station, and what CRS that reference station
was operating in.   That should have been collected as source metadata
at the time, and the proper CRS label applied to the dataset, which is
easier said than done, to fix in hindsight.

The idea that a DGPS (or RTK) network in Europe, in 2011, was in
ETRF2000, is highly plausible More or less, surveying (that uses
beyond-site coordinates) and national mapping activities very strongly
tend to use the national/regional datum.

Further, in the US, while we use "NAD83(2011)", networks are all set up
for the 2010.0 epoch of that (technically dynamic) datum, even today.  I
am unclear on EU practice in this regard.

It remains a fair, if probably unimportant, question to ask whether the
DGPS/RTK base station reference coordinates were in ETRF2000 epoch
1989.0 or in some different epoch.

The point of ETRS is to be plate fixed, and thus velocities of points on
the ground, when expressed in ETRF2000, should be near zero.  At 1 mm/y,
we're talking 15 mm, and at 1 cm/y, it's 15 cm.

> No I am still struggeling to find a way to do a time-dependent
> transformation in QGIS from ETRS89/ETRF2000 to the recent
> ETRS89/ETRF2020.

You are writing ETRS89/ETRF2000 as if it is a compound object, which it
isn't.  ETRS89 is a "system", which gives rules for defining frames, so
there are no cooprdinates in ETRS89.  ETRF2000 is a "realization" of the
system, also called a datum or reference frame.

I am unaware of people treating ETRFxxxx as dynamic, even though it
technically seems defined that way.  But I am underclued about ETRS/ETRF
practices.

I would suggest that you relabel your 2011 data as being in ETRF2000 and
see what happens.  Also use command-line proj:

  $ projinfo  -s ETRF2000 -t ETRF2020

which returns a lot of operations that surprise me, being time-dependent
transforms with non-zero rates (despite both ETRF2000 and ETRF2020 both
having base epoch of 1989.0).  But I say again, I am underclued about
ETRS.