do_mynnedmf = .false. causes SCM to stop?

Submitted by gthompsn on Sun, 09/19/2021 - 11:09

A true false option that was set to true in the GSD namelist and I want to turn it off - because two different cloud fraction schemes are called with or without this option.

So when I set it to false, SCM doesn't run and I just get the error message:

 

An error occurred in ccpp_physics_init: An error occured in mynnedmf_wrapper_init: Logic error: do_mynnedmf = .false.. Exiting...

 

If this means that I have conflicting namelist options, and the wrapper logic can detect it, why not print something to tell me what the conflicting other namelist item is?  I have also tried setting the bl_mynn_edmf option to zero.  No idea what to try next I suppose except reading source code until I find the root cause.

I suspect this has been answered elsewhere, but I'll put a response here in case others have the same question:

For some context, the CCPP physics was implemented "on top of" the existing GFS physics. While the CCPP expects which schemes get called to be configured by the suite definition file (SDF), the original GFS physics relied on a series of namelist flags. As of today, there is redundancy between the scheme "activation" flags and the SDFs. If you have a SDF configured to run the MYNNEDMF scheme, you must also set the do_mynnedmf namelist variable to true. Otherwise, during the CCPP physics initialization stage, the scheme will check for whether the activation flag is true and stop the model if it is not. This redundancy, although certainly not user-friendly right now, is temporary and will be fixed as the system matures into operations. The same behavior exists for both the UFS and SCM. If the SCM is expected to run the same physics as the UFS, this user-unfriendliness will have to exist until it is also fixed in the UFS, unfortunately.

If you want to use the MYNN EDMF scheme, but just turn off the mass flux part, set the bl_mynn_edmf namelist variable to false. If you want to exercise a different option with the MYNN scheme, see the documentation here: https://dtcenter.ucar.edu/GMTB/v5.0.0/sci_doc/namespacemynnedmf__wrapper_accff67ce1292402e1aabd6b73fe61d74.html#arg_table_mynnedmf_wrapper_run

Look for the variables that start with bl_mynn*.