open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello all,

I am testing the addition of a TIME SERIES COORDINATES FILE (new in V8R5) to a mesh using lat/lon coordinates (with GEOGRAPHIC SYSTEM=1, SPHERICAL COORDINATES=YES). However, I am encountering issues where the data written out is incorrect.

To troubleshoot, I ran the model with RESULT FILE IN LONGITUDE-LATITUDE=NO. When I manually selected approximate locations from the project mesh frame, the results seemed to match expectations much better.

This makes me wonder if there is an issue with how the time series extraction is handled when the mesh is in geographic coordinates or with how coordinates are interpreted from the TIME SERIES COORDINATES FILE.

Has anyone encountered similar issues or can provide insights into how this feature handles WGS84 geographic coordinates?

Thank you,

Hello,

There is an issue opened on gitlab:
gitlab.pam-retd.fr/otm/telemac-mascaret/-/issues/989

When time series files are handled, the mesh is already in Mercator projection.
You can work around this by giving coordinates in Mercator projection (see e.g. formula in $HOMETEL/sources/utils/bief/read_mesh_coord.f subroutine).

Another way is to convert the longitude/latitude coordinates with the same formulas in OUTHIST_PREPARE subroutine of the OUT_HISTORY module:
Conversion of TCDAT(IFILE)%XHISTOUT(I) and TCDAT(IFILE)%YHISTOUT(I) for I from 1 to NRPOINT, just before calling INTERP_POINT_PREPARE.

FYI GEOGRAPHIC SYSTEM keyword is only used for tide modelling and has no impact for time series files.

Hope this helps,

Chi-Tuan

Hello,

I have suggested a fix + enhancement in
gitlab.pam-retd.fr/otm/telemac-mascaret/-/issues/989

It is rebased with main (not so different from recently released 9.0).

Can you try to use it or change your release with the modifications and tell me if it solves your issue please?

Here is the archive (same as the 2nd archive in the issue).

Hope this helps,

Chi-Tuan

Hi Chin-Tuan,

I can confirm that when I used the files you attached in the *.tar.gz file and recompiled my desktop application of Telemac, the output timeseries data using Lat/lon coordinates works correctly. However, after pulling release 9.0 from the git repository and compiling the code on our HPC, the timeseries data is incorrect. Unlike the earlier version of the code, the model will run without producing an error related to a point not being in the domain, but all the timeseries data looks exactly the same.

Thanks
Glen

Hi Chin-Tuan,

I have re-downloaded version 9 of the code from the Git repository, performed a clean install, and recompiled. I can confirm that the time-series extraction is now working as expected. I appreciate your support.

Thanks,

Hello Glen,

Good news that everything works fine for you now, thanks for your feedback, in particular for the issue with longitude/latitude.
I was surprised to read what you wrote as from 9.0, there is a stop when coordinates of points are outside the extent of the domain.

Chi-Tuan