[Qgis-user] Problems with transformation including height values

[Greg Troxel]

Bo Victor Thomsen via QGIS-User <qgis-user at lists.osgeo.org> writes:

> Hi list -
>
> I have a problem with transforming a csv file containing x,y /and /z
> values from EPSG:4326 to EPSG:7416, i.e I want to transform the height
> value from LongLat/ WGS84 ellipsoid height to UTM32/Etrs89 with
> vertical reference DVR90.

Even if qgis supported vertical transforms, and with GDAL now, you'll
have more problems:

  Accuracy is harder and more important in the vertical.  Coordinates
  with a 1-2m error in the horizontal are considered pretty good.  But
  for analyzing flooding etc. that is enormous.  I suggest understanding
  the accuracy of your input data.

  EPSG:4326 is a 2-D coordinate system, with latitude and longitude
  only.  So talking about ellipsoidal heights in EPSG:4326 does not make
  sense.
    https://epsg.org/crs_4326/WGS-84.html
  You probably mean
    https://epsg.org/crs_4979/WGS-84.html

  WGS84 is an ensemble, which consists of a number of datums.  By
  labeling your data as WGS84, you are saying that it is in one of the
  member datums, but you **don't know which one**.  This is unlikely to
  actually be the case or make sense.  The earliest member of WGS84 is
  about 2m off from the rest and thus the ensemble is treated as low
  accuracy.  I suggest going back to the data and figuring out which
  ensemble member it really is in.  Note also that the qgis/gdal/proj
  world uses null transforms at times when the instrinsic error of the
  CRS is high; this results in wrong results in the US when transforming
  between WGS84 and NAD83.  (I don't know if this affects WGS84/ETRS89.)

  It is also possible, and I'd say highly likely, that your data is not
  actually in WGS84.  To obtain WGS84, you need to use autonomous GPS --
  meaning no differential*.  It is thus necessarily low accuracy.  If
  you use SBAS, such as EGNOS, then your coordinates are in the
  reference system of the SBAS system.  If you are using a
  multi-constellation receiver without SBAS, good luck figuring out what
  it's actually doing and if so please post on the proj list :-) If you
  are using RTK, then your coordinates are in the RTK reference
  station/network CRS.  If you are using static carrier-phase
  positioning, then your coordinates are in the coordinates of the
  reference stations you used.

  If someone had data in another frame and transformed it to WGS84, look
  into how that was done.  It is possible it was in say ITRF2014 and
  just relabeled.  That's sort of ok because ITRF2014 and WGS84(G1762)
  are essentially equal, but using the ensemble label is still harmful.


* Unless you have special access to internal US DoD data, perhaps.