model crashes with do_sppt=F and pert_clds=T

[pjpegion]

Description

Due to the lack of logic checking, if SPPT is not active and pert_clds is true, the cloud fraction gets modified by an uninitialized array and crashes.

Short term solution is to make sure pert_clds is false if do_sppt is false at the namelist level.

We will need to add a logic check to GFS_typedefs.F90 to ensure namelist compatibility:

if (.not. do_sppt) then pert_clds=.false. endif

[climbfuji]

Would it be better to abort the model run and tell the user to fix the namelist?

[pjpegion]

Or that.

[junwang-noaa]

@pjpegion Are you going to create a fv3atm PR with the code changes?

[arunchawla-NOAA]

@pjpegion what is the status of this ?

[DeniseWorthen]

@pjpegion Can you give a status update?

[pjpegion]

This is an open issue. Do you want me to implement Dom's suggestion?

[DeniseWorthen]

I don't know enough about these options to know what the full true/false matrix might be between the options.

Is there is only one combination of true/false that is valid (either scientifically or by code-implemenation)?

[DeniseWorthen]

I don't know enough about these options to know what the full true/false matrix might be between the options.

Is there is only one combination of true/false that is valid (either scientifically or by code-implemenation)?

[pjpegion]

@DeniseWorthen, scientifically pert_clds is only valid when SPPT is active. My suggestion was to just set pert_clds to false if SPPT is not on, Dom suggest aborting the model and telling the user to fix the namelist.

[junwang-noaa]

@pjpegion I agree with Dom, it might be better to abort the model run so that the users will be aware of the inconsistent namelist setting and get it fixed before they run the model and interpret the results.

[pjpegion]

OK, I'm attending a workshop at ECMWF this week. I will work on this when I get back. -Phil

On Mon, May 9, 2022 at 3:12 PM Jun Wang @.***> wrote:

@pjpegion https://github.com/pjpegion I agree with Dom, it might be better to abort the model run so that the users will be aware of the inconsistent namelist setting and get it fixed before they run the model and interpret the results.

— Reply to this email directly, view it on GitHub https://github.com/ufs-community/ufs-weather-model/issues/856#issuecomment-1121156023, or unsubscribe https://github.com/notifications/unsubscribe-auth/AJIRVJH27QOAAPAG3R5FFOTVJEML7ANCNFSM5FOZPDKA . You are receiving this because you were mentioned.Message ID: @.***>

-- Phil Pegion (he/him/his) Physical Scientist NOAA/Physical Sciences Laboratory (303) 497-7897 @.***

[pjpegion]

I'm working on the logic check, and will put it in GFS_typedefs.F90 after the namelist is read in. Is it appropriate to just put a stop here like other logic checks? Would this only abort the atmosphere model, and not ocean and ice?

[pjpegion]

the stop doesn't work. it only aborts the atmos model, the other components just hang.