Initialized predictions with CESM3 development tags

[islasimpson]

To more efficiently explore the impact of parameterization/tuning choices for CESM3, we plan to explore things in an initialized framework, with a particular focus on ENSO.

The plan:

• Use tag beta06 + mods of 165 (b.e30_beta06.B1850C_LTso.ne30_t232_wgx3.165 #90) if possible. If this fails with the existing restarts then @gustavo-marques will provide user_nl_mom for compatibility.
• Run a "climo initialized" simulation to examine model drift and for the computation of anomalies. One member initialized from November 1st of each year from 1991-2020. With historical forcings.
• Run a 1998 El Nino case, initialized from November 1st 1997 and run for 18 months with historical forcings.

Existing initial conditions (as of May 14th 2025):

• Ocean: /glade/derecho/scratch/gmarques/archive/g.e30_a05cish.GJRAv4.TL319_t232_wgx3_hycom1_N75.2025.028/rest/
  /glade/derecho/scratch/gmarques/archive/g.e30_b06.GJRAv4.TL319_t232_wgx3_hycom1_N75.2025.081/rest/
• ERA5 atmospheric fields on CAM7 L58 ne30np4 gll grid: /glade/campaign/cesm/development/espwg/CESM3_ERA5_IC/ne30np4_L58_gll - will need to be put into a restart file format compatible with CAM7
• @wwieder - do you have information on land restarts we can use?

Additional output:

• @gustavo-marques suggest any additional daily ocean outputs that would be needed.

@swrneale @PeterHjortLauritzen @cecilehannay @dlawrenncar @gustavo-marques @wwieder @sgyeager @nanr

[adamrher]

ERA5 atmospheric fields on CAM7 L58 grid: /glade/campaign/cgd/cas/islas/ERA5_nudging_input_6h/ne30pg3/L58/ - will need to be put into a restart file format compatible with CAM7

That will be difficult because the inic file needs to be on the dycore GLL grid, not the physics grid. I can just make one for you using betacast and ERA5 model level data?

[islasimpson]

Thanks. Julio's mods below worked, so I can create the Nov 1st initial dates needed for this with that code

[JulioTBacmeister]

I can provide ICs for SE if it helps. My code is set up to regrid onto the GLL grid as well as the physgird

[islasimpson]

Oh ok. What do I change to run it on the GLL grid instead?

[islasimpson]

Or if you already have them, that works too! Need November 1st for 1991 to 2020

[JulioTBacmeister]

The YAML file should look something like this ( I put * next to what I think you'll have to pay attention to) Dst: ne30np4 * DstVgrid: L93 IC_for_pg: True * Output_base_dir: null Resubmit: 0 StepBy: day StepN: 1 TheProcYear: -1 day: 1 hour: 0 month: 11 year: 2004

…

On Wed, May 14, 2025 at 3:24 PM islasimpson ***@***.***> wrote: Oh ok. What do I change to run it on the GLL grid instead? — Reply to this email directly, view it on GitHub <#91 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACGLMTQSGHXXA4TZ7ZP7T2326OYBHAVCNFSM6AAAAAB5EMU3H6VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGMJVGA2DAMQ> . You are receiving this because you commented.Message ID: ***@***.***>

-- *My working day may not be your working day. Please do not feel obliged to reply to this message outside of your normal working hours.*

[islasimpson]

OK, thanks. I'll try that.

[islasimpson]

This worked. I put the files on the gll grid for Nov 1st at /glade/campaign/cesm/development/espwg/CESM3_ERA5_IC/ne30np4_L58_gll and edited the discussion above to relfect that new path.

[JulioTBacmeister]

Could do that too.

…

On Wed, May 14, 2025 at 3:30 PM islasimpson ***@***.***> wrote: Or if you already have them, that works too! Need November 1st for 1991 to 2020 — Reply to this email directly, view it on GitHub <#91 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACGLMTRKZKELI3PFKM4XJOL26OYWZAVCNFSM6AAAAAB5EMU3H6VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGMJVGA2DMNI> . You are receiving this because you commented.Message ID: ***@***.***>

-- *My working day may not be your working day. Please do not feel obliged to reply to this message outside of your normal working hours.*

[wwieder]

Hey Isla,

For land initial conditions, I'm doing a historical runs on the ne30 grid that are basically clones of CESM_156, but they are running with the 5.4 surface data, LMWG_dev #102.

Currently the case is in the early 1900s of an I-HIST case and I'm planning to create monthly restart files from 1990-2023 once we get there. The only negative here is that we're only getting 56 simulation years / day and the queue on derecho has been really slow to get into (> 7 hours between overnight resubmits for a 16 node request)

A few other notes. This case was created from a CTSM5.3.041 tag with:

• namelist changes to turn on the right snow thermal conductivity schemes in the right places,
• a different parameter file than what's being used in the coupled model
• new ctsm5.4 surface data (created with CMIP7 data), but also not yet being used in cesm_dev cases.

I'm not sure the latest beta_06 tag has these features (mainly for namelist control on the snow thermal conductivity schemes over glaciers), but we can use source mods for this that will be close enough, as Cecile did for CESM cases 155, 156 and beyond

Alternatively, we could make a cesm branch tag to get the right land settings out of the box, which would avoid confusion and errors, but may take more SE time than is necessary for the runs you're wanting to do.

If all of the above sounds good I'll let you know when the land restarts are ready in /glade/derecho/scratch/wwieder/archive/ctsm53041_54surfdata_ne30_102_HIST/rest/

[islasimpson]

This sounds good to me. I think at this point since these aren't production runs, what we really need is just something half reasonable to initialize the land with. Since the primary focus for now will be on ENSO, the mismatches you referred to don't sound like they would be a big problem. One thing though, I wonder if it's worth starting to output the monthly restart files earlier than 1990 if that's still possible? Maybe 1970? In case we want to run other ENSO cases further back in time. @sgyeager, do you agree? If it's already got there, then don't worry about it. I think it'll take some time to get all the other pieces of this in order, so it's OK if the queues are being slow. Thanks!

[sgyeager]

I agree. It would be great to assemble a collection of monthly restarts spanning 1970-2020, if possible.

[wwieder]

OK this should be simple enough. Is 1971-2023 OK? That will be easiest for me.

[sgyeager]

yes, that would work. Thanks!

[gustavo-marques]

Here are monthly restarts from a G case using beta06. Please use this simulation instead of 2025.028.
/glade/derecho/scratch/gmarques/archive/g.e30_b06.GJRAv4.TL319_t232_wgx3_hycom1_N75.2025.081/rest/

[wwieder]

Monthly land restarts are here with the caveat that (currently) atmospheric CO2 is held at 2014 concentrations through 2023. Rerunning this now.

/glade/derecho/scratch/wwieder/archive/ctsm53041_54surfdata_ne30_102_HIST/rest/

[wwieder]

actually, I didn't save interim (monthly) restart files to the st_archiver. Doing this all again now using the correct CO2 streams.... sorry for the rookie mistake

[wwieder]

monthly land initial conditions are now up through Dec 2000 and the run is continuing through 2023.

[gustavo-marques]

I did the same as @wwieder and did not have DOUT_S_SAVE_INTERIM_RESTART_FILES=True. I will run for another cycle with DOUT_S_SAVE_INTERIM_RESTART_FILES=True

[cecilehannay]

Here is some details about the configuration we used for 165 (#90)

create_newcase

• Tag: cesm3_0_beta06
• compset: B1850C_LTso
• resolution: ne30pg3_t232_wg37

SourceMods

/glade/campaign/cesm/cesmdata/cseg/runs/cesm2_0/b.e30_beta06.B1850C_LTso.ne30_t232_wgx3.165/SourceMods/src.cam/beljaars_drag.F90
/glade/campaign/cesm/cesmdata/cseg/runs/cesm2_0/b.e30_beta06.B1850C_LTso.ne30_t232_wgx3.165/SourceMods/src.clm/SoilTemperatureMod.F90
/glade/campaign/cesm/cesmdata/cseg/runs/cesm2_0/b.e30_beta06.B1850C_LTso.ne30_t232_wgx3.165/SourceMods/src.drv/shr_flux_mod.F90 

Namelists

• add to user_nl_cam:

solar_irrad_data_file= '/glade/campaign/cesm/development/cross-wg/inputdata/SolarForcingCMIP7piControl_c20250103.nc'
bnd_topo = '/glade/campaign/cgd/amp/pel/topo/cesm3/ne30pg3_gmted2010_modis_bedmachine_nc3000_Laplace0050_noleak_20250325.nc'

micro_mg_dcs= 600.D-6
cldfrc_dp1 =  0.05
clubb_c8 = 4.6 

• add to user_nl_clm:

nfix_method = 'Bytnerowicz'
use_c13 = .true.
use_c14 = .true.
use_c13_timeseries = .true.
use_c14_bombspike = .true.
use_init_interp = .true.
do_grossunrep=.false.
reseed_dead_plants = .true.
paramfile = '/glade/campaign/cesm/development/amwg/inputdata_extra/ctsm5.3.012.Nfix_params.v13.c250221_upplim250.nc'

• add to user_nl_mom:

INPUTDIR = "INPUT"
MAXTRUNC = 2000
REMAPPING_USE_OM4_SUBCELLS = False
BODNER_DETECT_MLD = True
MLE_USE_PBL_MLD = False
MLE_DENSITY_DIFF = 0.02
MLE%
USE_BODNER23 = True
USE_CR_GRID = True
CR_FILE = "mle-cr-tx2_3v2-v8_20250215.nc"
MIN_WSTAR2 = 1.0E-09
BLD_DECAYING_TFILTER = 8.64E+04
MLD_DECAYING_TFILTER = 2.592E+06
BLD_GROWING_TFILTER = 300.
MLD_GROWING_TFILTER = 3600.
%MLE
KPP%
KPP_CVt2 = 1.0
%KPP
RESTART_CHECKSUMS_REQUIRED = FALSE

• add to user_nl_drv:

histaux_l2x1yrg = .true.

@nanr

[nanr]

@Cecile Hannay ***@***.***> : - compset: B1850C_LTso This looks like a B1850 compset. Is there an equivalent transient compset that I should be using. *~Nan Rosenbloom* ***@***.*** ORCID ID =0000-0001-7389-3346

…

On Wed, May 21, 2025 at 3:42 PM Gustavo Marques ***@***.***> wrote: I did the same as @wwieder <https://github.com/wwieder> and did not have DOUT_S_SAVE_INTERIM_RESTART_FILES=True. I will run for another cycle with DOUT_S_SAVE_INTERIM_RESTART_FILES=True — Reply to this email directly, view it on GitHub <#91 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AEBD3OMLGTZRAJLB7P4QDKD27TXMLAVCNFSM6AAAAAB5EMU3H6VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGMRSGU4TSOI> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[cecilehannay]

The transient equivalent of B1850C_LTsois BHISTC_LTso.
I have not used it yet but I believe it should work out of the box.

[nanr]

Last question @cecilehannay (for the moment): What does the 'so' mean in the compset name?
B1850C_LTso

Nan

[cecilehannay]

Specified oxidants

[nanr]

Note: HIST vs HISTC: Which should I be using?

I have been using this longname in my create_newcase call
_HIST__CAM70%LT_CLM60%BGC-CROP_CICE_MOM6_MOSART_DGLC%NOEVOLVE_WW3_SESP

Compset: BHISTC_LTso
HISTC_CAM70%LT_CLM60%BGC-CROP_CICE_MOM6_MOSART_DGLC%NOEVOLVE_WW3_SESP

[cecilehannay]

I had a look into directory and it seems to me that CAM_NML_USE_CASE is not set properly in the compset BHISTC_LTso. As a result it is using some 2000 forcing instead of historical forcing.

Using HIST_CAM70%LT as in /glade/work/nanr/CESM3-SMYLE/b.e30_beta06.BLTHISTC_SMYLE.ne30pg3_t232_wg37.1991-11.001
sets:
CAM_NML_USE_CASE=hist_cam_lt
(Note this casename is labeled BLTHISTC although it uses HIST_CAM70%LT and not HISTC_CAM70%LT)

Using HISTC_CAM70%LT as in
/glade/work/nanr/CESM3-SMYLE/b.e30_beta06.BHISTC_LTso_SMYLE.ne30pg3_t232_wg37.1993-11.001
sets:
CAM_NML_USE_CASE=UNSET
This doesn't seem correct to me.

I adding CAM SEs to the thread (@nusbaume, @cacraigucar)

[cecilehannay]

@brian-eaton:

To clarify what we are looking at:
Nan has been running a B case with beta06 using the compset:
HISTC_CAM70%LT_CLM60%BGC-CROP_CICE_MOM6_MOSART_DGLC%NOEVOLVE_WW3_SESP

Based on her case directory:
/glade/work/nanr/CESM3-SMYLE/b.e30_beta06.BHISTC_LTso_SMYLE.ne30pg3_t232_wg37.1993-11.001
it seems to me that HISTC_CAM70%LT in beta06 doesn't set any use_cases by default (CAM_NML_USE_CASE=UNSET).
As a result, she is getting cyclical forcing for the year 2000.

[nanr]

And for comparison, here is a parallel case using BHIST (B case with beta06) using the compset:

• HIST_CAM70%LT_CLM60%BGC-CROP_CICE_MOM6_MOSART_DGLC%NOEVOLVE_WW3_SESP
• /glade/work/nanr/CESM3-SMYLE/b.e30_beta06.BLTHISTC_SMYLE.ne30pg3_t232_wg37.1991-11.001

[nanr]

NOTE: this case was not labeled correctly. It is really a HIST case: b.e30_beta06.BLTHISTC_SMYLE.ne30pg3_t232_wg37.1991-11.001

[cecilehannay]

We are going to implement a bugfix in CAM:
ESCOMP/CAM#1315

I believe that in the mean time, you can just use:
CAM_NML_USE_CASE=hist_cam_lt

[dlawrenncar]

I think probably HISTC, but it's also possible that these are just equivalent. Someone else will need to weigh in here to confirm.

…

On Tue, May 27, 2025 at 5:20 PM nanr ***@***.***> wrote: *Note: HIST vs HISTC: Which should I be using?* I have been using this *longname* in my create_newcase call _*HIST*__CAM70%LT_CLM60%BGC-CROP_CICE_MOM6_MOSART_DGLC%NOEVOLVE_WW3_SESP Compset: BHISTC_LTso *HISTC*_CAM70%LT_CLM60%BGC-CROP_CICE_MOM6_MOSART_DGLC%NOEVOLVE_WW3_SESP — Reply to this email directly, view it on GitHub <#91 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFABYVBEP5FGAW4LYBXHQCD3ATXL5AVCNFSM6AAAAAB5EMU3H6VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGMRZGAYDQNA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[brian-eaton]

Looks like the problem is in CAM's 'config_component.xml' file, which is responsible for setting 'CAM_NML_USE_CASE'. It contains

      <value compset="HIST_CAM70%LT"     >hist_cam_lt</value>

but should instead contain

      <value compset="HIST[CE]?_CAM70%LT"     >hist_cam_lt</value>

Not sure what the best fix is for the runs being discussed above. I can create a CAM tag quickly...

[nanr]

Thanks Brian ~ What is your recommendation? Is this something I should wait for (I'm hoping to get this set of 40 runs completed in the next few days before the machine goes down). Alternatively, I could make the correction in my own sandbox ..?

[nanr]

I made the change in my sandbox and I'll check that if fixes the problem!
Thanks for the help finding the issue!

Nan

[cecilehannay]

I think that if you do:

./xmlchange CAM_NML_USE_CASE=hist_cam_lt

in your case directory, you will get the correct forcing.

[cacraigucar]

FYI - @brian-eaton will be making cam6_4_095 containing this change

[cacraigucar]

Just an FYI - The tag cam6_4_095 containing the use case change has been committed by @brian-eaton