Upgrading suite to v13.5

Unified Model
1

Hello,

Because of the upcoming requirement for the new system to support only vn13.5 and up I’m trying to upgrade my suite from vn13.0 to vn13.5. I followed the guidance in Upgrading a suite to the latest UM version and Upgrading suite to vn12.0 - #3 by James_Weber and upgrading through each version one by one. I then have updated the ‘Sources’ in fcm_make_um for ‘um_sources’ and ‘casim_sources’ to a branch with vn13.5.

Running the suite it gets through fcm_make_um but is giving an error message during glm reconfiguration:
Error from routine: CHECK_JULES_HYDROLOGY
Error message: No value for ti_max

Are there some more steps that should be done before running the suite to bring things up to vn13.5? The suite I am working in is u-dp232.

Thanks for your help,
Russell

2

Hi,

There appears to be mismatch in settings being applied to the glm_um_recon1 task. ‘ti_max’ is required when l_top is true in jules_hydrology namelist. It seems that l_top is set to .false. by default but .true. in ga7/8/coMorph optional configurations. However, I am not sure which of these is being selected for glm_um/ glm_um_recon.

Are you able to compare the settings with another vn13.4/5 GA8/9 Nested configuration - maybe the task runtime settings (suite.rc, suite-runtime.rc ) also need to be updated manually?

–
Mohit

3

Hi Mohit,

I see in the glm namelist - rose-app.conf that l_top was true but ti_max wasn’t there so I have added it but now I am getting,
‘Error reading namelist jules_deposition
IoMsg: A READ operation tried to read past the end-of-file.
Please check input list against code.’

It seems like something needs to be updated or reconfigured but I’m not sure what.

Thank you,
Russell

4

Hi Russell,

It looks like the upgrade macros have not been run for the app/glm_um folder. The settings in that rose-app.conf are still at vn13.0.
The rose app-upgrade command will have to be run for the intervening versions in all relevant folders including glm_um, fcm_make_createbc and um_createbc.

–
Mohit

5

Hi Mohit,

Great, thanks I’ve upgraded those folders and the glm_recon finishes but immediately runs into an error on the first glm_fcst time with,

Error from routine: CHK_OPTS_MOD:CHK_REAL_STR called by CHECK_RUN_BL
Problem performing defensive checks (CHECK_RUN_BL)
c_gust, (-0.107E+10), has failed to pa c_gust, (-0.107E+10), has failed to pass any

Do you have a sense of the issue here?

Thank you,
Russell

6

Hi,

This is due a new namelist item (c_gust) being added by a branch used in the suite: branches/dev/adrianlock/vn13.5_c_gust_nml.
Since this is at vn13.5 itself the corresponding upgrade macro will not be applied automatically. So, you can :

  • either temporarily add c_gust=4.0 by hand to [namelist:run_bl] (to app/um and app/glm_um)
  • or, checkout Adrian’s branch above and modify app/glm_um (and um/) /rose-app.conf: meta=path-to-branch/rose-meta/um-atmos/HEAD
    then run rose-edit in app/glm_um and app/um for Rose to detect the missing item and add it via the fixer macros.
    The second option means that item will be added to any copies of the suite.

Similarly, if any other branches used contain a rose metadata change (items added /removed/ renamed), all the rose-meta.conf files will have to be merged in a single working copy and pointed to via ‘meta=’

Mohit

7

Hi,

I did the first option and continued on to the fcst recon steps where I have run into another error related to unrecognised namelist options:

Error reading namelist RUN_Precip
IoMsg: The variable name ‘CASIM_AEROSOL_COUPLE_CHOICE’ is unrecognized in namelist input.
Please check input list against code.

Is this related to your last point about metadata changes? Could you explain this a bit further? I’m not sure I follow.

Thanks very much

8

Hi,

This seems to be an odd one. The casim_XX variables are currently not supposed to be in the namelist. However, I can trace them in the ral3p1 opt configuration right back to the first suite (u-cv482) that you have copied over, although in a deactivated state (with !!). One of the changes in your suite -the rose --macro-fix one- seems to have reactivated them, hence the error as these are not present in the code.
You could try after commenting these out by hand, but I am not sure how many others have been inadvertantly added.
I would also suggest asking someone familiar with Nesting suites to review the changes as the automatic upgrades over 5 versions could mean some new/ recommended Science settings are not being applied.

–
Mohit

9

Hi Mohit,

Sorry for the late response, I have been working through the namelist and making changes based on the errors I was getting for adding/removing options. I’ve been able to complete the reconfigure steps and the model has successfully run so it seems that things have been upgraded successfully to 13.5 however I am nervous from your last comment about some science settings potentially not being applied. I will seek some help to review my suite.

Thanks very much,
Russell