FAQ: Data ocean slab mode (DOCN-SOM)

[YCCircle]

Hi!

I'm trying to run CESM2 with the following compset: 2000_CAM60_SLND_CICE_DOCN%SOMAQP_SROF_SGLC_SWAV, which is unsupported. It is hoped to be an aquaplanet slab ocean mode with sea ice. However, I found that there was only cam output without cice output (history files, log files, restart files). It seems that I did not turn on the sea ice component successfully. I made no changes in user_nl_cice file. I'm wondering how can I solve this problem.

Thank you so much!

Best,
Rose

 

[dbailey]

I've moved this to the simpler model forum.

 

[brianpm]

I believe that "DOCN%SOMAQP" will not do what you want. The aquaplanet SOM doesn't work with ice at this time. Instead, I think you will need to change to something like:

2000_CAM60_SLND_CICE_DOCN%SOM_SROF_SGLC_SWAV

I took a look in ./components/cpl7/components/data_comps_mct/docn/src/docn_comp_mod.F90 to compare what happens in each case. I think you can look at docn_comp_run in that module to see what I mean. In there, the model checks what kind of data ocean is being run, and you can compare case('SOM') to case('SOM_AQUAP') to see that the aquaplanet version explicitly excludes the temperature adjustment associated with freezing (this allows the slab temperature to drop below freezing, which is necessary when ice isn't present). I'm guessing that this would break the sea-ice, or at least not produce what you'd want, since the slab temperature wouldn't be interacting correctly with any ice that did somehow form.

 

[YCCircle]

I've moved this to the simpler model forum.

Thank you very much! I changed the compset to 2000_CAM60_SLND_CICE_DOCN%SOM_SROF_SGLC_SWAV following other's suggestion. I still have a question about the SOM forcing file. It's not the focus of my study, so I hope to solve this problem in the most convenient way (maybe just using the existing forcing files?). The default of my compset is UNSET. I think the most similar compset is ETEST (2000_CAM60_CLM50%SP_CICE_DOCN%SOM_MOSART_SGLC_SWAV_TEST) despite having the land component. ETEST provides a default forcing file like pop_frc.1x1d.090130.nc. I'm just wondering whether I could use the same forcing file as ETEST.

 

[YCCircle]

I believe that "DOCN%SOMAQP" will not do what you want. The aquaplanet SOM doesn't work with ice at this time. Instead, I think you will need to change to something like:

2000_CAM60_SLND_CICE_DOCN%SOM_SROF_SGLC_SWAV

I took a look in ./components/cpl7/components/data_comps_mct/docn/src/docn_comp_mod.F90 to compare what happens in each case. I think you can look at docn_comp_run in that module to see what I mean. In there, the model checks what kind of data ocean is being run, and you can compare case('SOM') to case('SOM_AQUAP') to see that the aquaplanet version explicitly excludes the temperature adjustment associated with freezing (this allows the slab temperature to drop below freezing, which is necessary when ice isn't present). I'm guessing that this would break the sea-ice, or at least not produce what you'd want, since the slab temperature wouldn't be interacting correctly with any ice that did somehow form.

Thank you for your suggestion! Sooooo useful! I'm trying to create a new case based on your suggestion. :D

 

[YCCircle]

Thank you very much! I changed the compset to 2000_CAM60_SLND_CICE_DOCN%SOM_SROF_SGLC_SWAV following other's suggestion. I still have a question about the SOM forcing file. It's not the focus of my study, so I hope to solve this problem in the most convenient way (maybe just using the existing forcing files?). The default of my compset is UNSET. I think the most similar compset is ETEST (2000_CAM60_CLM50%SP_CICE_DOCN%SOM_MOSART_SGLC_SWAV_TEST) despite having the land component. ETEST provides a default forcing file like pop_frc.1x1d.090130.nc. I'm just wondering whether I could use the same forcing file as ETEST.

I read your replies to others again in another forum. My questions may have been partially answered, but I am still confused about them. The grid of pop_frc.1x1d.090130.nc seems not to match my case (f09_f09_mg17). Do I need to interpolate? Can I use the aquaplanet default forcing file (default.som.forcing.aquaplanet.Qflux0_h30_sstQOBS.1degFV_c20170421.nc) after changing the qflux and slab ocean thickness to proper patterns? The grid of aquaplanet forcing file is the same as my current compset.

 

[brianpm]

The "f09_f09_mg17" grid specifies that the grid for both atmosphere and ocean will be "f09" meaning a 0.9°x1.25° latitude-longitude grid using the finite volume dynamical core. This is how we have usually run aquaplanets with the SOM. Having the data ocean model and the atmosphere on the same grid is slightly more efficient because there is no remapping of the fluxes that are sent between the components. When sea-ice is turned on, I think the CICE model has not been able to be run on the same grid without using some simplifications; that is when the POP grid comes into play.

Running SOM+sea-ice in aquaplanet has been done, but it is not supported because there is no agreement at this time about the best way to get the configuration running. When I did some experiments in the old version of the model, I was able to get the model to run by switching to "thermodynamic sea ice" (the dynamics turned off). Some description of that is in this document: https://www2.cesm.ucar.edu/models/simpler-models/files/CAM_aqua_som_description.pdf

I'm not sure what is required in the current model to do the same thing. You may need to set kdyn=0 in user_nl_cice to turn off the dynamics, and you might need to set ktherm=0 for very simple thermodynamics. This configuration *should* be able to run on the lat-lon grid defined in the domain file, which you will need to provide to the sea ice model via the namelist parameters grid_file and kmt_file.

Other users have reported that it's difficult to get a stable, non-glaciated climate when sea ice is turned on. So you might need to tinker with the settings to get a climate that is suitable for your application. Good luck!

 

[YCCircle]

The "f09_f09_mg17" grid specifies that the grid for both atmosphere and ocean will be "f09" meaning a 0.9°x1.25° latitude-longitude grid using the finite volume dynamical core. This is how we have usually run aquaplanets with the SOM. Having the data ocean model and the atmosphere on the same grid is slightly more efficient because there is no remapping of the fluxes that are sent between the components. When sea-ice is turned on, I think the CICE model has not been able to be run on the same grid without using some simplifications; that is when the POP grid comes into play.

Running SOM+sea-ice in aquaplanet has been done, but it is not supported because there is no agreement at this time about the best way to get the configuration running. When I did some experiments in the old version of the model, I was able to get the model to run by switching to "thermodynamic sea ice" (the dynamics turned off). Some description of that is in this document: https://www2.cesm.ucar.edu/models/simpler-models/files/CAM_aqua_som_description.pdf

I'm not sure what is required in the current model to do the same thing. You may need to set kdyn=0 in user_nl_cice to turn off the dynamics, and you might need to set ktherm=0 for very simple thermodynamics. This configuration *should* be able to run on the lat-lon grid defined in the domain file, which you will need to provide to the sea ice model via the namelist parameters grid_file and kmt_file.

Other users have reported that it's difficult to get a stable, non-glaciated climate when sea ice is turned on. So you might need to tinker with the settings to get a climate that is suitable for your application. Good luck!

Thank you for your useful suggestions! As for the grid_file and kmt_file, there are several files to choose (https://svn-ccsm-inputdata.cgd.ucar.edu/trunk/inputdata/ocn/pop/gx1v6/grid/). I'm not very sure what the differences are and which one to choose. Do you have any suggestions?

 

[brianpm]

I would recommend trying to start on the FV grid rather than the POP grid. You could set:

grid_file = '/glade/p/cesmdata/inputdata/share/domains/domain.aqua.fv1.9x2.5_c20170421.nc'
kmt_file = '/glade/p/cesmdata/inputdata/share/domains/domain.aqua.fv1.9x2.5_c20170421.nc'

Those are for the 2° grid. If you are running with "fv09" (1°), you could try to point at my (old, unsupported) file:

/glade/work/brianpm/model_data/domain.aqua.fv0.9x1.25.nc

The main reason to try to use the atmosphere grid rather than the POP grid is that the POP grid has converging grid lines at the "poles." Usually these are not part of the ice domain because they are hidden by land areas. When these parts of the grid are exposed to the sea-ice model, there are a number of potential numerical problems. *I don't know whether this is still an issue with later versions of CICE, but when I tried to get sea ice in aquaplanets working many years ago, this was a major problem.*