ocean convergence

[dhofer@climate_unibe_ch]

Hi all,

I have some problems with the convergence in the ocean model. My settings:

CCSM3.0.1 beta24
CRAY XT3 (linux)
branch for b30.032b

According to the log-files, the problem occurs on the first day in the barotropic driver when the pcg_chrongear is called ("POP exiting... no convergence at time step xx"). I have increased the DT_count, but that didn't help either. Anyone an idea?

Regards,
Dominik

 

[njn01]

Most often when you encounter a pop nonconvergence error in the first few timesteps, it is an indication that the forcing fields supplied to the ocean model are somehow incorrect.

It is a good idea to decrease the timestep (increase DT_COUNT) and try again, as you have done, but if you need to decrease the timestep by anything more than about 20%, the problem is most likely not a timestepping instability.

 

[ckcastillo@purdue_edu]

Hi, I am also experiencing the non-convergence error, though not within the first few timesteps - I am ~80 years into my error. The dt_count is currently at 23. Could you suggest a good value to try changing it to? Thanks!

 

[njn01]

Usually, an increase of 10-20% in dt_count is usually sufficient. (Note that increasing dt_count decreases the timestep.) So in your case, setting dt_count to 25-28 ought to allow you to integrate past the point of nonconvergence. Oftentimes, you can later reduce dt_count after integrating for a year or so with the smaller timestep.