Add diagnostic presets to pre-industrial+concentrations configuration

[blimlim]

This pull request adds Standard and Detailed diagnostic request presets for each submodel to the pre-industrial+concentrations configurations. Switching between the presets can be controlled by changing the relevant symlink in each submodel's control directory.

The changes to these presets compared to the previous default diagnostic requests are:

Atmosphere
Detailed: same as previous requests, though bugged wetland fraction field fld_s08i248, CO2 flux variables fld_s03i326 and fld_s03i327 which don't work in concentration mode, and fld_s03i173 which is named incorrectly are removed. Soil temp averaged over tiles added in.
Standard: Same as Detailed with most variables on model theta levels removed and most daily tiled data removed.

Ocean
Detailed: Same as previous requests, with all 3D monthly variables duplicated to yearly output. Variable renaming is removed, so that vanilla names are used. Variables that were requested previously requested but were not output by the model are not included.
Standard: Same as detailed but with several 3D monthly fields removed (kept at yearly frequency), as described here.

Both ocean diag_tables are generated via the make_diag_table script, and hence produce output with one variable per file.

Ice
Detailed: Same as existing historical configuration (which included daily output for several fields), though with the bugged fields identified here removed.
Standard: Same as Detailed, with daily output removed.

Ice history namelist settings have been moved from the cice_in.nml to the separate ice_history.nml to enable swapping via symlinks.

[blimlim]

One thing to note: The addition of the separate ice_history.nml namelist to control CICE's output requires the latest changes to payu from this pull request in order to work. CICE will not find the required settings and will crash without these changes.

I don't think these changes have made their way into the dev version of payu yet, and so the configuration might not work currently if you try to run with it.

[anton-seaice]

The cice stuff looks good. is diagnostics_requests the best name for it ? I think diagnostics_profiles or history_profiles is better?

[minghangli-uni]

For the ocean diag_table, I am wondering why both yearly and monthly frequency fields are included at the same time? Was this decided at the release meeting? At the end of the day, both frequencies contribute to the same diagnostics, with yearly diagnostics essentially being an average of the monthly ones. Storing both can create redundancy, causing larger datasets and inefficiencies in data management. And maintaining both frequencies might cause confusion and require more computational resources.

[blimlim]

For the ocean diag_table, I am wondering why both yearly and monthly frequency fields are included at the same time?

That's a good point, and not something I have a great reason for. When setting up the diag_table_standard, the idea was to keep the regularly used 3D fields at monthly frequency, and move the less used ones to yearly. Given that some fields were being kept at yearly frequencies, I wasn't sure whether it was important to then do the same for the more popular ones too, in case it would be convenient for people doing yearly analysis. You're right that it is quite redundant though.

For the diag_table_detailed where all 3D variables are saved at monthly frequency, I agree it's probably not useful to keep the yearly 3D variables at all.

@minghangli-uni do you think the following would make more sense:

diag_table_detailed:

• As is but remove all yearly 3D fields

diag_table_standard:

• Keep the regularly used 3D fields at monthly frequency
• At yearly frequency, keep only the less regularly used fields (i.e. those which are not in the monthly frequency list)

Let me know if this would sound a bit better, or any other ideas/suggestions would be great too!

[minghangli-uni]

Sounds good to me! Thanks @blimlim

[blimlim]

I think diagnostics_profiles or history_profiles is better?

Yep I think those are easier to understand – I've renamed the directories to diagnostic_profiles but also happy with either of those suggestions.

[blimlim]

@minghangli-uni, I've made those changes for the yearly ocean data, let me know if it's looking ok.

@anton-seaice, for some reason I don't understand, the dumpfreq reverted to m. I've set it back to y but not sure why github is reporting the original value as monthly.

[MartinDix]

The standard monthly means include 4 |THETA AFTER TIMESTEP on model levels. I don't think this is useful.

We should save
1| 1 | 0 | 407 |PRESSURE AT RHO LEVELS AFTER TS |
which is necessary for calculating a vertical integral of the tracers

Apologies for missing this one before, but we don't need orography in either of the daily sets.

[minghangli-uni]

The ocean diagnostics look good to me. Thanks, @blimlim

[blimlim]

Thanks!
I've removed THETA AFTER TIMESTEP from the standard profile, and also the daily LEAF AREA INDEX (CASA-CNP GLAI).

I've added in PRESSURE AT RHO LEVELS AFTER TS to both the standard and detailed preset, using IDOM=8, corresponding to &DOMAIN NAME="DALLRH"

The detailed preset already had a copy of this variable, though on the domain &DOMAIN NAME="DALLTH",:

 &STREQ IMOD= 1, ISEC=0, ITEM=407, IDOM=2, ITIM=3, IUSE=2 / 
 &STREQ IMOD= 1, ISEC=0, ITEM=407, IDOM=8, ITIM=3, IUSE=2 /

@MartinDix, is it important to keep both of these, or is one of the domain selections incorrect?

[MartinDix]

Only the version on the rho levels gets through to the output file so you can remove the theta one. The model seems to just silently ignore the one with theta levels.

I couldn't find any messages in the log file.

[aidanheerdegen]

Looks great, thanks for all the work and constructive input from others.

[blimlim]

Great! Thanks everyone for your help!