ACCESS-OM2 model setup problem

COSIMA Ocean
,
1

Hi all,

I’d like to run a new 01deg_jra55v13_ryf_9091 simulation. This is a re-run from a simulation I completed successfully last year and I am using the same configuration since last time.

Unfortunately I now get a

FATAL from PE 2626: Error: salinity out of range with value -6.324080775145E+00 at (i,j,k) = (2034,1446, 63), (lon,lat,dpt) = ( -76.6500, 19.6529, 3377.5509 m)

abort notification when restarting the simulation again.

Andrew mentioned this could be caused by a model setup problem. Near the Antarctic coast (where my salinity drops) is where a few cells in the bathymetry were changed recently and maybe there is a mismatch between my configuration and this input field?

These are the input files for my MOM model:

- name: ocean
  model: mom
  exe: /g/data/ik11/inputs/access-om2/bin/fms_ACCESS-OM_97e3429_libaccessom2_1bb8904.x
  input:
      - /g/data/ik11/inputs/access-om2/input_08022019/mom_01deg

I am aware I am using a rather old executable and the reason for it is that this one allows me to save WMT diagnostics in temperature space. Could it be that the /g/data/ik11/inputs are not compatible anymore with my .exe files?

2

I think this is probably the thread with the changes made:

(lon,lat,dpt) = ( -76.6500, 19.6529, 3377.5509 m) is in the Caribbean Sea somewhere. Although I am surprised there is 3000m depth there ?

The files in /g/data/ik11/inputs/access-om2/input_08022019/mom_01deg are all from February 2019, and don’t look like they changed over time (based on their date stamps)

3

Hi @mauricehuguenin:

Thanks for starting this topic. @anton was quick to the mark here (thanks Anton!) and has pointed you to a useful topic that should help you out. I’m going to mark this topic with the community-help tag so that folks from COSIMA can provide any additional support, since I think this points to some legacy COSIMA experiments.

Please continue to post any additional comments/queries here!

Thanks,

Lawrence

4

I managed to solve this issue by changing the ice_ocean_timestep in accessom2.nml from 540 to 720. During the first few months of the simulation, I had to adjust the time step a few more times but eventually the model stabilised itself enough that it did not abort anymore after year 2 of the simulation.

While troubleshooting, I chose any of the following time steps:

The timestep is normally set to a factor of the JRA55-do forcing period of 3hr = 10800s, for example one of 100, 108, 120, 135, 144, 150, 180, 200, 216, 225, 240, 270, 300, 360, 400, 432, 450, 540, 600, 675, 720, 900, 1080, 1200, 1350, 1800, 2160, 2700, 3600 or 5400s. The default timestep for this configuration is 1350 seconds, and the model is stable with this timestep right from the start. After the first year or two of model equilibration you may be able to run with a 1800s timestep for faster throughput.

5

PS: What a relief! :grinning_face_with_smiling_eyes: