Spectral element simulation with new version CESM2

[BoHuang]

Dear all,

I am a new user working with CESM2 model. Recently, I want to build a spectral element simulation with CESM2. I choose FHIST_BGC component and ne30_g16 grids for test. However, when I build the simulation, I get the error message “out=CAM build-namelist - ERROR: No default value found for ncdata”. I add “ncdata = "/cluster/work/users/boh/cesm/input-data/atm/cam/inic/se/cami-mam4_0000-01-01_ne30np4_L32_c200331.nc"” to user_nl_cam, I get another error message “out=CAM build-namelist - ERROR: No default value found for bnd_topo”. If I add bnd_topo (bnd_topo = "/cluster/work/users/boh/cesm/input-data/atm/cam/topo/USGS-gtopo30_ne30np4_16xdel2-PFC-consistentSGH.nc") in user_nl_cam, the error as “err=CAM build-namelist - ERROR: Use of topography not currently supported with SE dycore”. I have no idea how to fix this problem. Is there anyone can help me or give me any hint? Thank you very much.

Regards,
Bo

 

[nusbaume]

Hi Bo,

Sadly the Spectral Element dycore is only officially supported with simple physics configurations in CESM2.1.3 (and earlier). It will be expected to work for all physics configurations in CESM2.2, but that version will likely not be released for at least another few months. So, unless there is an explicit reason why you want to use this dycore, I would recommend using the Finite Volume (FV) configuration instead, which uses the "f09_g16" grid.

Hope that helps, and have a great day!

Jesse

 

[BoHuang]

Hi Jesse,

thank you so much for the confirmation. I want to use the SE dycore due to I want investigate the interaction between recent land cover change and climate change. If the grid size is too large (see 1deg), the land cover change is not significant. It is difficult to quantify the land cover change and climate change. Therefore, I want to use the refine grid in my focus region (as Europe, East Asia). For my case, how do you think run the simulation with f02_g16 grid?

Greetings,
Bo

 

[nusbaume]

Hi Bo,

If 1-degree resolution is an issue, then you can certainly use either the "f05_g16" grid, which has a horizontal resolution of 0.47 degrees N x 0.63 degrees E, or the "f02_g16" grid you mentioned, which has a horizontal resolution of 0.23 degrees N x 0.31 degrees E. Sadly these will be expensive simulations (as they will have that resolution globally), but they should be relatively straight-forward to setup and run (just do the same as you would for an out-of-the-box CESM simulation, but with one of those grids specified in create_newcase).

If those options aren't doable due to their computational cost, then you can either just wait until CESM2.2 is released, or you can try contacting the scientists in the manuscript you found directly to see if they would be willing to help you use the developmental version of the model, which might then allow you to gain access to variable resolution grids.

Hope that helps, and have a good weekend!

Jesse

 

[BoHuang]

Hi Jesse,

great. Thank you so much. Your suggestion is really helpful. I will do it.

Have a nice weekend!

Regards,
Bo

 

[Qiyu8]

Hi Jesse,

I want to build BW1850_f09_g17 component and ne16_g4 grids for test. I heard that Spectral element simulation is faster than FV dycore, How to get the CESM2.2 version? Is there anyone can help me?

Regards,
Qiyu8