HYCOM v2.3 Model Crash – Request for Assistance "neg. dp (m) < -10.00" Error

[Achref OTHMANI]

Dear HYCOM users,

I’m currently running a HYCOM v2.3 spin-up from 1993 to 2024 , and I'm encountering a persistent model crash with the following error:

error: neg. dp (m) < -10.00

The crash consistently occurs on March 6, 2023, at the same location—near the western model boundary, which is closed.

Here are the steps I’ve taken so far to troubleshoot the issue:

• Reduced both baroclinic and barotropic time steps (tested with very small values).

• Verified the relaxation, nesting, and blkdat.input files.

• Attempted a restart from January 5, 2023 and from January 2022, with variable viscosity and diffusion coefficients.

• Ran several short tests around the crash period, modifying only one parameter at a time.

Unfortunately, none of these approaches have resolved the issue.

I’ve attached the log file and the blkdat.input file for reference. I would greatly appreciate any insights or suggestions from the community to help resolve this.

Thank you in advance for your support.

Best regards,
Achref OTHMANI

[Alan Wallcraft]

In general, keep batrop as large as possible.  A smaller batrop is a less stable batrop.

The usual reason for a persistent blow up at the same model time is an issue with the input files.  If your nesting rmu file is wrong (relaxing near the western model boundary by mistake) it could be the 3-D nesting, but it is more likely to be the atmospheric forcing.

One test would be to replace the initial restart file with one from a different year.  If it blows up in the same place: it is the forcing.  A check on this would be to use the "correct" restart but change the year of the forcing, which should not blow up.

Alan.

[Achref OTHMANI]

Dear Alan,

Thank you for your response and the helpful suggestions.

I’ve carried out the tests you recommended:

• Replaced the initial restart file with one from a different year.

• Used the original restart, but changed the year of the atmospheric forcing.

Unfortunately, in both cases, the model still crashes on March 6, 2023, at the same location near the western boundary, with the same error (neg. dp (m) < -10.00).

I’ve also double-checked the nesting RMU files to ensure that relaxation isn't occurring near the western boundary, and everything appears to be in order.

Given that the issue persists across these configurations, I suspect it may be related to a more subtle nesting or boundary setup issue.

Any further insights would be greatly appreciated.

Best regards,
Achref OTHMANI

[Alan Wallcraft]

a) Start from a case that works, e.g.:

restart, forcing and nesting from 1 year before

confirm that this works.

b) Replace the (a) case restart with the one that started the blow up segment (edit the model day in the .b file).  If this fails, then the restart is bad and you should rerun the segment before the original blow up with a smaller baroclinic time step.

c) If (b) works, turn off the nesting and rerun (still with the blow up restart).

d) if (c) works, rerun the original blow up case turning off nesting.

e) if (d) fails and (c) works, then it is the forcing.

It is possible that turning off nesting will blow up, but if it works then the above steps should clarify the issue.

Alan.