Scheduled Downtime
On Tuesday 24 October 2023 @ 5pm MT the forums will be in read only mode in preparation for the downtime. On Wednesday 25 October 2023 @ 5am MT, this website will be down for maintenance and expected to return online later in the morning.
Normal Operations
The forums are back online with normal operations. If you notice any issues or errors related to the forums, please reach out to help@ucar.edu

how to change dtime and the corresponding namelists: NSPLIT, NSPLTRAC, NSPLTVRM

zhuchenxia

New Member
Hello, everyone

The used CESM version is CESM1.2.2, which is not the currently supported vesions, but the problem may be related to other CESM versions.

We run the CESM F compset (F_2000_CAM5) with the resolution (f19_g16) in year (2002). The details for the case are in Attach files (Script).
At first, we met a problem with CESM F compset long-term running. It only run for 5 months and stopped with the error "Opened file ./F_Restart_InitialFiles6.clm2.h0.2002-06.nc to write 1441792 NetCDF: Numeric conversion not representable". We have solved the long-term running problem by made changes to: 1.ATM_NCPL (value="72") in env_run.xml (which means dtime has changed to 24h/72=1200s); 2.add namelist changes (nsplit=6, nspltrac=2,nspltvrm=2) in user_nl_cam, and now the model can successfully run for 10 months. It can be concluded that changing dtime and the corresponding namelists can make successful CESM F compset long-term run.

Now we want to run CESM F compset in a quicker speed, and we would like to increase the timestep (dtime). However, I have the question about the numerical relationship between dtime and nsplit, nspltrac and nspltvrm. The used values above (dtime=1200, nsplit=6, nspltrac=2, nspltvrm=2) are from a thread online. For example, if I change dtime to 6h, what the values for nsplit, nspltrac and nspltvrm should be?

The used machine is our own machine nuistwgj and the compiler pgi/17.10. The config details can be seen in Attach files (config_compileres.xml, config_machines.xml and mkbatch.nuistwgj).

Any suggestion may help. Thank you very much.
 

Attachments

  • Attach files.zip
    14 KB · Views: 1

jedwards

CSEG and Liaisons
Staff member
By changing the timestep and the relationship between overall timestep and dynamic substepping of the model you are changing the
models science. This is part of the models scientific tuning and we cannot make any recommendations about changing it. A safer way to
change the model performance is by using more or faster processors or a faster communications network.
 

zhuchenxia

New Member
Thanks, jedwards, for reminding me not to change dtime arbitrarily.
Later, I will only change processors to improve the model performance.
 
Top