[QGIS-Developer] CRS problems with WGS84 and ETRS89

[Kristian Evers]

Hi Bo,

You are on the right path. It is true that ETRS89 is plate-fixed and that WGS84 is not and that the two slowly drift apart. There are two facts that makes this more complicated than most of us would like:


1.       There exists 6 realizations, or versions, of WGS84 that differ up to a meter between the two versions that are farthest apart. Since in general in GIS we never specific *which* of the six WGS84’s the uncertainty in the transformation is larger than the 30 years of continental plate movement.

2.       To do an accurate transform between a global reference like WGS84 and a plate-fixed frame like ETRS89 you need to know the observation time of your coordinate. What if you WGS84 coordinate was measured in 1995? Then you only need to account for a few years of plate movement compared to a coordinate from today where 30 years is needed.

Because of the two above facts the default transformation between WGS84 and ETRS89 is the identity transformation. Improving the situation required both time-stamped coordinates and knowledge of the specific WGS84 version in use.

For the Danish case, if you really need high accuracy transformations you can take a look at https://github.com/NordicGeodesy/NordicTransformations where transformations with cm-accuracy between ITRS2014 (basically WGS84 today) and the Danish realization of ETRS89 is available in the form of PROJ parameter and grid files.

/Kristian

From: QGIS-Developer <qgis-developer-bounces at lists.osgeo.org> On Behalf Of Bo Victor Thomsen
Sent: 25. september 2019 08:24
To: QGIS Developer Mailing List <qgis-developer at lists.osgeo.org>
Subject: [QGIS-Developer] CRS problems with WGS84 and ETRS89


Hi list members -

I have question regarding coordinate reference systems.

As I understand the difference between the datums WGS84 and ETRS89:

  *   They are based on the same ellipsoid. And in 1989 there were - for all practical purposes - no difference between the the 2 datums coordinate wise
  *   However, the datums are "drifting" apart because the ETRS89 is coordinate bound  to the Eurasian continental plate and WGS84 is not.
  *   The continental drift is ca 2.5 cm / year. And around 70 cm in all.

So a specific position in Denmark (or Europe) in  the same mathematical projection i.e UTM32N should have 2 different x,y values using the WGS84 datum  and ETRS89 datum. And the present "distance" should be ca. 70 cm.

However, that is not the case in QGIS (v. 3.8):  A point digitized with position (550000, 6080000) in ETRS89/UTM32N has the (almost) exact same x/y values in WGS84/UTM32N when you change the projection in the map window from ETRS89/UTM32N to WGS84/UTM32N. The x/y values differs around .1 mm. Certainly no "drift" around 70 cm.

(Doing the same experiment using ED50 instead of WGS84 show the expected difference around 200 m between ED50 and ETRS89)

How come ? Or is the error-source sitting ca. 70 cm in front of the screen ?? (I have a 34" large flatscreen :-)



--

Med venlig hilsen / Kind regards



Bo Victor Thomsen
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/qgis-developer/attachments/20190925/5ab33753/attachment-0001.html>