-
Notifications
You must be signed in to change notification settings - Fork 789
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
depth to depth or two CRS with orientation down issue? #4118
Comments
I've investigated that, and I believe the root cause if that operations such as EPSG:11488 are abusing the "Geog3D to Geog2D+GravityRelatedHeight (gtx)" method, in a way it doesn't anticipate. This method should only be used for a Geographic3D CRS to a compound CRS. Here you use 2 compound CRS. The code in PROJ that deals with "Geog3D to Geog2D+GravityRelatedHeight" and "Geog3D to Geog2D+Depth are actually the same, with the only difference that it does a vertical axis inversion prior to applying the geoid model if the target vertical CRS has a down orientation. I'm not sure how that should be resolved. We are really a in very advanced use case. I believe it would be cleaner for those depth to depth transformation to use EPSG:1084 "Vertical Offset by Grid Interpolation (gtx)" to handle the vertical transformation with vertical axis=UP, and add an explicit height/depth reversal operation before and after (which will likely require extra tricks in fixStepsDirection()...). Or you might want to write custom PROJ based operations with an explicit pipeline |
Thanks @rouault, I actually started with EPSG:1084 doing vertical only transforms but ran into issues with the horizontal so we thought we would enforce that the correct horizontal was used with the tidal datum models we produce. It may be that I didn't have interpolation_crs_code set when I did my initial testing -- I'll retry using EPSG:1084. You had suggested the pipelines in #3819 when I was asking about doing extra transforms to 3D and I actually had those working. I recently went back to concatenated (seems more object oriented/modular) when we started leaning away from trying to get all our tidal datums to have transforms to a singular 3D pivot. I meant to ask on the mailing list about what the right compound to compound would be since I only saw the EPSG:1084 (vertical only), EPSG:1088 (3d to compound) or EPSG:1115 (hydroid) as the closest options. I'll go back to the drawing board and test a bit more. Thanks again. |
Of particular concern to us is to retain "3D" CRS attribution (whether Geog3D or compound) in source data, through to coordinate expression in certain target CRS connected by reversible transformations. A straightforward example is a (reversible) compound to compound via concatenated operations pair, through a pivot in the Geog3D; e.g., EPSG:10496 "ETRS89 + DVR90(2013) height to ETRS89 + DVR90(2023) height (1)": Correction of prior NOAA examples (last two use concatenated operations): NOAA:8506 = NAD83(FBN) + LMSL (WAjuandefuca01/vdatum_3.4_20141224/1983-2001) depth NOAA:8507 = NAD83(FBN) + MHHW (WAjuandefuca01/vdatum_3.4_20141224/1983-2001) depth NOAA:8511 = NAD83(FBN) + LMSL (WAjuandefuca01/vdatum_3.4_20141224/1983-2001) height NOAA:8512 = NAD83(FBN) + MHHW (WAjuandefuca01/vdatum_3.4_20141224/1983-2001) height NOAA:8550 = NAD83(FBN) + NAVD88 height NOAA:8546 = NAD83(FBN) + NAVD88 depth |
Problem description
In my custom proj.db, attached, I am trying to connect three compound CRS depth objects with a concatenated operation of two grid_transformations. My concatenated_operation of the heights seems to work. I then tried just calling the grid_transformation and got unexpected results. I may have made the depth operations incorrectly but swapping the source and target has no effect which seems wrong even if I had something set wrong.
Example of problem
proj.zip
Expected Output
Depth and height operations would hopefully be equal but opposite sign.
Environment Information
Installation method
The text was updated successfully, but these errors were encountered: