Submission of Interpolation on the Fly code for inclusion in the trunk

[sga]

This is a first attempt at submitting new code to the NEMO trunk from NOCS.
The code is for "Interpolation on the Fly" (IOF) for reading 2D input fields,
currently implemented in the new SBC fld_read routine.

Attached is a tar file containing the proposed code changes and NVTK test results.
The code probably needs a good going over by Gurvan for adherence to the current coding conventions,
which it may fail! I anticipate that there may also have to be compromises on its internal workings, so I guess
this is the first step in this code submission.

trunk1220-IOF.patch is the patch file to convert trunk revision 1220 into an IOF enabled
version of NEMO. (Revision 407 referred to in the patch is that of the NOCS NEMO trunk which
is a mirror of revision 1220.)

This also patches the documentation in the DOC directory (mainly the SBC chapter)

Included are three subdirectories with the output from three runs of NVTK.

NVTK-trunk: has the four test cases of unadulterated trunk revision 1220 on our machine
NVTK-2TEST: has the four tests when IOF code that differs from the trunk is

introduced via the 2TEST directory and so represents a test of
the code when not switched on. All namelists have to be manually
altered before the ins_nvtk.ksh step for this to work.

NVTK-IOF: has the four tests when the new code is merged into the trunk code and

run with new test data.

Note that the ZAGRIF test is failed in all cases.

The test data used in the NVTK-IOF tests can be obtained by anonymous ftp from ftp.noc.soton.ac.uk,
subdirectory omfftp/IOF, name nemo_iof_ORCA2_LIM.tar.gz.

NOCS NEMO Team

[sga]

Thank you for adding this new code to the trunk (revision 1275).

We would strongly recommend however that the namelists do not include reference to the ORCA2 weights file that was included with a CORE forced test case. This weights file is only valid for this test case and will break all other configurations. the entries in the columns headed "weights filename" in the namelists should be set to a blank string to force the code not to attempt to use IOF. The CORE test case includes appropriate namelists and corresponding non-ORCA2 gridded input files for IOF.

The weights files need to be generated for each non-ORCA source grid as discussed in the documentation.
There is example code available for weights generation available from ​ftp://ftp.noc.soton.ac.uk/omfftp/NEMO/nocs_weights.tar.gz

NOCS NEMO Team

[rblod]

Replying to sga:

Thank you for adding this new code to the trunk (revision 1275).

We would strongly recommend however that the namelists do not include reference to the ORCA2 weights file that was included with a CORE forced test case. This weights file is only valid for this test case and will break all other configurations. the entries in the columns headed "weights filename" in the namelists should be set to a blank string to force the code not to attempt to use IOF. The CORE test case includes appropriate namelists and corresponding non-ORCA2 gridded input files for IOF.

The weights files need to be generated for each non-ORCA source grid as discussed in the documentation.
There is example code available for weights generation available from ​ftp://ftp.noc.soton.ac.uk/omfftp/NEMO/nocs_weights.tar.gz

NOCS NEMO Team

You're welcome.

The integration has been quiet straight forward with your patch. I didlet the reference to the weights files for CORE forging in the namelist on purpose, as an example, since CORE is not used by default, and forcing files not provided. So when reading the namelist users can have an example of classical forcing with CLIO, and interpolation with CORE, it may not be such a good idea, and can be changed

Do you wonder if we make your weight computation code available on the web site? Just tell me if it is not possible.
In addition, the interpolation is not working on our NEC computer. I'm a bit suspicious about the behaviour of the NEC compiler when dealing with pointers, since I encountered also a lot of issues with AGRIF in the past.

Rachid

[nicolasmartin]

Milestone NEMO improvements deleted