Can MOM6 save restart file with fields at one moment?

[Xuan]

I see users can revise 'RESTINT' to control time interval of restart files. Restart files seem to be saved with daily mean fields? Can we save fields at a moment? Thanks!

 

[kshedstrom@alaska_edu]

Restart files are not daily mean fields, they are instantaneous. They (ideally) contain enough information for a perfect restart. If you ever find that the restarts are not perfect, there is a bug in the restart. I mean two one-hour runs with a restart between should give *exactly* the same answer as a single two-hour run.

 

[Xuan]

Thanks! I compare a single 30-day run with two 15-day runs with a restart, they are not the same. What problems could have in the restart?

 

[kshedstrom@alaska_edu]

Gosh, it depends! MOM6 has so many options and we don't test all of the combinations. What are the details of your configuration? Did you start with a standard test case? I have fixed some bugs in the OBC restarts. Are you using OBCs?

 

[Xuan]

I run a buoyancy forced experiment. It's not a standard test case. I have no idea how it restarts, the variables in MOM.res.nc are 3-D velocity, layer thickness, viscous acceleration.

 

[kshedstrom@alaska_edu]

You will either have to tell us a *lot* more about what you are doing or you will have to debug this yourself. The way to debug it yourself is to start with very short runs with DEBUG = True. You will get used to looking at the debug output to see where things diverge. Then I sometimes wind up running two debugger jobs, one starting at time zero, the other from a restart at one hour. One hour is my coupling timestep, so nothing shorter makes sense.

 

[dorukhan]

Is there an example usage of RESTINT anywhere? It says the time units are in days but when I tried setting it to 1 or 0.5 days it creates MOM.res.nc on the last day only.