CESM B2000 compset error with Floating Invalid in POP (forrtl: error (65): floating invalid)

[usha k]

I am Using CESM2.1.5, I tried making new case with B2000 compset, --compset 2000_CAM60_CLM50%BGC-CROP_CICE_POP2%ECO%ABIO-DIC_MOSART_CISM2%NOEVOLVE_WW3_BGC%BDRD, similar to the B1850 compset.

Initially i got the error of ice: no. blocks exceed max: increase max to 2
which i solved following : very slow CICE running on Derecho

./xmlchange NTASKS_ICE=96

./xmlchange CICE_MXBLCKS=16

./xmlchange CICE_AUTO_DECOMP=false,

Now I am getting the in cesm.log as :

(tidal_min_regions_set_cvmixT) num_region,k,lat,lon,tidal_diff = 2 48 64.01 355.71 46.697880146050487
(tidal_min_regions_set_cvmixT) num_region,k,lat,lon,tidal_diff = 2 49 64.01 355.71 83.436298362531815
forrtl: error (65): floating invalid
Image PC Routine Line Source
libpthread-2.28.s 000015554A81ECF0 Unknown Unknown Unknown
cesm.exe 00000000023D985E diagnostics_mp_ch 3285 diagnostics.F90
cesm.exe 00000000021F7086 ocn_comp_mct_mp_o 656 ocn_comp_mct.F90
cesm.exe 00000000004370F3 component_mod_mp_ 728 component_mod.F90
cesm.exe 000000000041DF11 cime_comp_mod_mp_ 3452 cime_comp_mod.F90
cesm.exe 0000000000436D60 MAIN__ 125 cime_driver.F90
cesm.exe 000000000041A33D Unknown Unknown Unknown
libc-2.28.so 000015554A27CD85 __libc_start_main Unknown Unknown
cesm.exe 000000000041A25E Unknown Unknown Unknown


I tired switching of the tidal mixing by making following changes in user_nl_pop but still the error persist.

&tidal_nml
ltidal_mixing = .false.
ltidal_min_regions = .false.
ltidal_max = .false.
ltidal_stabc = .false.

! ---- Disable horizontally varying background ----
lhoriz_varying_bckgrnd = .false.

! ---- Use constant background diffusivity (explicitly define) ----
bckgrnd_vdc1 = 1.0e-5
bckgrnd_vdc2 = 1.0e-5
bckgrnd_vdc_eq = 1.0e-5
bckgrnd_vdc_psim = 1.0e-5
/
&overflows_nml
overflows_on = .false.
overflows_interactive = .false.
/

I have also attached all the log files for your reference
Kindly help to rectify this error. What could be the issue ?

 

[fischer]

Hi Usha,

Can you try rebuilding and running your case with debug turned on. This will help narrow down the location of the floating invalid.
To turn on debugging and rebuilding you can do the following.

./xmlchange DEBUG=TRUE
./case.build --clean-all
./case.build

Thanks
Chris

 

[fischer]

Hi Usha,

I was having trouble getting the long compset name to build for me. But then I tried defining a B2000 alias in cime_config/config_compsets.xml. I was able to run the B2000 compset at f09_g17 for 6 days using both gnu/12.4.0 and intel/2023.2.1.

Chris

 

[usha k]

Hi Chris,

I tried running with debug=True, could not find the source of the error. the line 3285 of diagnostics.F90 points to the line :
if (diag_ke>ke_thresh.or.diag_ke<0_ check_KE = true
and the line 656 ocn_comp_mct.F90 points to these:
if (check_KE(100.0_r8)) then
!*** exit if energy is blowing
call output_driver
call exit_POP(sigAbort,'ERROR: k.e. > 100 ')
endif
call output_driver

Also, i have one more doubt, in pop_in, all the input files are in gx1v7 grid, except the files used in &tidal_nml, &forcing_shf_nml,&niw_nml and &ecosys_forcing_data_nml, they are gx1v6
there are no tidal files for the gx1v7 grid in the repository. Could that be the problem?..

Also I am using the refcase file (for year 2000) from Globus as mentioned in one of the threads.

 

[fischer]

Hi Usha,

I'm not sure if the gx1v6 files would cause issues. My test that I ran has the gx1v6 files. I'm wonder if using a year 2000 ref case isn't the issue. I'm going to move this to the POP forum for them to look at it.

Chris