[PROJ] System34 height difference

[Kristian Evers]

Søren,


On 22 Apr 2020, at 16:04, Søren Holm <sgh at sgh.dk<mailto:sgh at sgh.dk>> wrote:

Hi

We are using proj for transformation of some Danish coordinatesystems.

Among them are S34 and DKTM.

Usually the height is not affected much during transformations. This is
however different for S34.


Yes, System34 is only a 2D coordinate reference system, so the height is generally not meant to be used.


$ cct -t0  +proj=pipeline +step +init=DK:DKTM1
10 55 0 0
 263991.7405   1097565.8295        0.0000        0.0000 0


$ cct -t0  +proj=pipeline +step +init=DK:S34J
10 55 0 0
 234033.0592     63129.6022      -36.8103        0.0000 0


Why is there ~36m subtracted for S34?

The reason is that there is an implicit datum change happening in that transformation, which internally works in three dimensions and then consequently transforms the zero height to another datum. This, however, has no physical meaning and can be regarded as a mathematical curiosity. With PROJ > 6.0.0 this can actually be avoided by using the push<https://proj.org/operations/conversions/push.html> and pop<https://proj.org/operations/conversions/pop.html> operators:

echo 10 55 | ./bin/cct -z0 -t0 +proj=pipeline +step +proj=push +v_3 +step +init=DK:S34J +step +proj=pop +v_3
  234033.0592     63129.6022        0.0000        0.0000

The DK init file was created before that was possible. This probably should be updated, although it comes with the risk that someone now relies on the current behaviour. System34 has been submitted to the EPSG but the application is still pending. Once that find it’s way to the EPSG registry this problem should disappear entirely in PROJ. The time frame is unfortunately unknown at this time.

Why is there around 0.1mm height difference for most UTM and NTM zones?

I haven’t been able to replicate that, so hard to comment on. A guess is that there is some numerical inaccuracy involved. I don’t think there’s anything to worry about, 0.1 mm is very likely less than both the accuracy of the data and the transformation.

Is it the intention that the heights should be affected during these
transformations?

For System34 it’s yes and no. Yes, because that is the way the transformation is currently made but no since it the height isn’t really part of the system. Can’t comment on the other case.

/Kristian

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/proj/attachments/20200426/5fb36bb8/attachment.html>