Variables passing through CMEPS to SCHISM

[yunfangsun]

Description

To Reproduce:

Additional context

Output

[danishyo]

Using Ali’s Sandy case to test schism_3d (51fc37a) branch, found variables exchanged from WAV to OCN have some zero values along the boundary, even with LON coupling, which are not shown in coastal_app (c9d2936) branch. See below figures, outputs are from schism nc files:

In order to trace which part (schism/cmeps/ww3) might causing this issue, we further checks CMEPS outputs for both branch, coastal_app is normal, but schism_3d has zero values along the boundary, see below figures:

Since Sxx/Syy are not output in previous tests, we choose another variable Hs which both CMEPS & WW3 output its value to see any difference, Hs in WW3 shows normal values along the boundary, but some zero values in CMEPS outputs. This suggests something wrong when WW3 output variables to CMEPS, see below figure:

[yunfangsun]

wait to consolidate all branches

[danishyo]

Sandy case with latest coastal_app update (85e35c), checking LON coupling from both model output.
Previous zero values along boundary still exist, see the following figure:

I need to remove "MED med_phases_history_write" in run sequence to avoid signal 11 errors, thus no CMEPS output netcdf (like ufs.cpld.cpl.hi.2012-10-27-00300.nc) to check further.

[danishyo]

Sandy case with latest coastal_app update (85e35c), checking LON coupling from both model output. Previous zero values along boundary still exist, see the following figure:

I need to remove "MED med_phases_history_write" in run sequence to avoid signal 11 errors, thus no CMEPS output netcdf (like ufs.cpld.cpl.hi.2012-10-27-00300.nc) to check further.

@uturuncoglu FYI

[uturuncoglu]

@danishyo I have no idea why this is not working. As I remember there was a version of code that works. Right? We could compare the files and source to find the underlying issue. Can you do that?

[danishyo]

@danishyo I have no idea why this is not working. As I remember there was a version of code that works. Right? We could compare the files and source to find the underlying issue. Can you do that?

Yes, version c9d2936 is working and have no zero boundary issues, I can have a look, but may need guidance.

[uturuncoglu]

@danishyo okay it is good to know. In your case, we need to look at difference in ww3 and schism side for any suspicious code change. At this point I don't think it is related with CMEPS etc. So, did some experiment with components one by one. Start with the version of schism and ww3 that are working (from the hash) and for example update ww3 (maybe with some small steps) run and check. If everything fine then do the same thing for other component. Btw, you could narrow down the place and find the commit that creates issue.

[danishyo]

@uturuncoglu , since c9d2936 with lon-coupling working and no zero boundary issues, I tried to replace its SCHISM/SCHISM-ESMF with version from 85e35c , the SXX also shows NO zero boundary issue (like original one in c9d2936). This means code is OK for the current version from SCHISM side.
From ww3 output, SXX also shows NO zero boundary issues in both 85e35c and c9d2936, I have no clues which component might lead this error.

[uturuncoglu]

@danishyo I think we need to understand what is going on here. If you don't could you list briefly your tests in here. it could be like a very simple table that has all the components (CMEPS, CDEPS, SCHISM and WW3) and their hashes and the issue is shown nor not. Something like following,

Run #	CMEPS	CDEPS	SCHISM	WW3	Boundaries are
1	hash1	hash2	hash3	hash4	Ok/Not Ok

So, then I could say something about it. BTW, I am assuming that input files are same for all.

[danishyo]

@uturuncoglu , see below

<style> </style>

Run #	CDEPS	CMEPS	SCHISM	SCHISM-ESMF	WW3	Zero Boundary Issue
Test1 (coastal-app: c9d2936)	0dc0044	437d5e6	cfb58a5	12e2ce4	80a5ad5	No
Test2 (coastal-app: 85e35c3)	dcec20b	9c37623	5ef34a8	d3ab56d	0a34d19	Yes
Test3	0dc0044	437d5e6	5ef34a8	12e2ce4	80a5ad5	No
Test4	0dc0044	437d5e6	cfb58a5	d3ab56d	80a5ad5	No
Test5	0dc0044	437d5e6	5ef34a8	d3ab56d	80a5ad5	No
Test6	dcec20b	437d5e6	5ef34a8	d3ab56d	0a34d19	Yes
Test7 (coastal-app: 0ce7489)	dcec20b	9c37623	31984c6	2c9d178	e49cee9	Yes