n1850.ne30_tn14.hybrid.20240822

[mvertens]

Purpose:
Exactly like n1850.ne30_tn14.hybrid.20240808 EXCEPT

• for an updated code with a bug fix oslo_aero (corrected dust emissions)
• updates to the BLOM for hybrid coordinates
• new initial conditions for BLOM and CICE
• NOTE: a memory leak is present in the run and hits agains the node limit when running past 11 years.
  See memory leak in n1850 case NorESM#550

---

Description:

• create_newcase commnad
  ./create_newcase --case /cluster/projects/nn9560k/mvertens/cases/n1850.ne30_tn14.hybrid.20240822 --compset N1850 --res ne30pg3_tn14 --project nn9039k --run-unsupported

• user_nl_cam

 interpolate_nlat   = 192
 interpolate_nlon   = 288
 interpolate_output = .true.
 history_aerosol = .true. 
 zmconv_c0_lnd  =  0.0075D0
 zmconv_c0_ocn  =  0.0300D0
 zmconv_ke =  5.0E-6
 zmconv_ke_lnd =  1.0E-5
 clim_modal_aero_top_press =  1.D-4
 bndtvg = '/cluster/shared/noresm/inputdata/atm/cam/ggas/noaamisc.r8.nc'

• user_nl_blom
  ICFILE = "/cluster/shared/noresm/inputdata/restart/NOIIAJRAOC20TR_TL319_tn14_ppm_20240816/rest/1775-01-01-00000/NOIIAJRAOC20TR_TL319_tn14_ppm_20240816.blom.r.1775-01-01-00000.nc"

• user_nl_cice
  ice_ic = "/cluster/shared/noresm/inputdata/restart/NOIIAJRAOC20TR_TL319_tn14_ppm_20240816/rest/1775-01-01-00000/NOIIAJRAOC20TR_TL319_tn14_ppm_20240816.cice.r.1775-01-01-00000.nc"

• xmlchange commands

./xmlchange NTASKS=3200
./xmlchange NTASKS_OCN=256
./xmlchange ROOTPE=0
./xmlchange ROOTPE_OCN=3200
./xmlchange BLOM_VCOORD=cntiso_hybrid,BLOM_TURBULENT_CLOSURE=
./xmlchange STOP_OPTION=nyears
./xmlchange STOP_N=20
./xmlchange REST_N=1
./xmlchange REST_OPTION=nyears
./xmlchange RESUBMIT=1
./xmlchange --subgroup case.run JOB_WALLCLOCK_TIME=48:00:00
./xmlchange --subgroup case.st_archive JOB_WALLCLOCK_TIME=03:00:00

---

Compset long name
1850_CAM%DEV%LT%NORESM%CAMoslo_CLM51%SP_CICE_BLOM%ECO_MOSART_DGLC%NOEVOLVE_SWAV_SESP

---

Model grid long name
a%ne30np4.pg3_l%ne30pg4.pg3_oi%tnx1v4_r%r05_w%null_z%null_g%gris4_m%tnx1v4

Case directory:

• Locally on betzy: /cluster/projects/nn9560k/cases/mvertens/n1850.ne30_tn14.hybrid.20240822
• On github: TODO

---

Code version on github (either NorESMhub or user's fork):

• On github: https://github.com/mvertens/NorESM/tree/feature/noresm2_5_alpha04_v3

---

Diagnostics:

• ocean/ice: https://ns2345k.web.sigma2.no/datalake/diagnostics/noresm/adagj/n1850.ne30_tn14.hybrid.20240822/

• atm:

• https://ns2345k.web.sigma2.no/datalake/diagnostics/ADF/plots/n1850.ne30_tn14.hybrid.20240822_1_10_vs_Obs/website/index.html

• https://ns2345k.web.sigma2.no/datalake/diagnostics/ADF/plots/n1850.ne30_tn14.hybrid.20240822_1_10_vs_n1850.ne30_tn14.hybrid.20240808_1_10/website/index.html

• Comparison to NorESM2-MM piControl: https://ns2345k.web.sigma2.no/datalake/diagnostics/ADF/plots/n1850.ne30_tn14.hybrid.20240822_1_10_vs_N1850frc2_f09_tn14_20191001_1200_1209/website/index.html

---

Output:

• Raw output: /nird/datalake/NS9560K/noresm3/cases/ n1850.ne30_tn14.hybrid.20240822
• CAM Climos: TODO

---

Contacts:
@mvertens, @adagj

---

Extra details:

[monsieuralok]

@mvertens I have tested NF1850 case with yours tag; it is also happening there; Memory start to increase just after every mid month; if I remember well; CAM is reading somedata mid months probably there could be allocation of memory I will check it.

[mvertens]

@monsieuralok - I have discovered that this memory leak also occurs in CESM simulations. @jedwards4b is tracking this down in CAM right now and believes that it in tphysac. I'll update the issue once I know more.

[jedwards4b]

Just to clarify - I'm not sure we are seeing the same leak - mine appears to be a smaller increment and not associated with the 15th of the month as yours is.

[mvertens]

@jedwards4b - I agree. It would be good to see what the source of the CESM memory leak is at this point and see if it impacts what I am seeing.

[jedwards4b]

ESCOMP/CAM#1136

[monsieuralok]

@mvertens I guess this solved memory leak issue issue; variable lam and phi are not deallocated in file :- components/cam/src/chemistry/oslo_aero/src_cam/mo_drydep.F90

probably same in @jedwards4b

components/cam/src/chemistry/mozart/mo_drydep.F90

but, it has solved my memory leak issue atleast in NorESM but, I need to double check.

Case folder on Betzy:- /cluster/projects/nn9560k/alok/NorESM-mariana/cases/NF1850.ne30_tn14.hybrid.20240822_N43

You look for SourceMods and only file mo_drydep.F90

[mvertens]

@monsieuralok - since lam and phi are allocatable variables in mo_drydep.F90
real(r8), allocatable :: lam(:), phi(:)
my understanding is that they will be automatically deallocated upon exit of the subroutine. If they were pointers that were allocated, this would be a different case. Do you agree with this?
That said - in looking at your mediator log file in /cluster/work/users/agu002/archive/NF1850.ne30_tn14.hybrid.20240822_N43/logs your memory profile does indeed look better - and the memory leak I saw happened earlier in the year. But just to make sure - can you please run longer than that to confirm your hypothesis. I would be very surprised if not deallocating allocatable variables solves this.

[mvertens]

@monsieuralok - to clarify - lam and phi are allocatable local variables - not module variables - and so according to the fortran standard they should be deallocated upon exiting the subroutine.