Advice on setting up 21st century SSP585 F case

[nick]

Hi, I want to set up an F case for the end of the 21st century (e.g. 2080-2100 SSP585) to run some sensitivity tests, but of course I need ocean and other boundary conditions to do that. Could someone recommend a way for me to approach this. Is it feasible to source all the data I need (SSTs/sea ice/GHG's/etc) and create my own boundary condition files. Or should I simply run my own B case for the 21st century and branch F cases off of that? Granted this latter option takes a lot of compute. I don't have access to NCAR drives.

 

[nick]

I've found this experiment in the list of public experiments, which by the name sounds something like what I want to do (i.e. a "scenario forced" FHIST case).
f.e21.FHIST_SSP585.f09_f09_mg17.CMIP6-lfmip-pdLC.001

The case files aren't available online anymore and I don't have access to NCAR drives. Would it be possible for someone to post these case files again? That way I can try see the changes that need to be made to run a scenario forced case.

 

[erik]

Our model input data is publicly available, and for F compsets that includes ocean SST's for example. You could modify those files if you want to, but you don't have to create them from scratch. You do need a subversion (svn) client to download the data to your machine.

For help on this with for example CESM2 release versions see the instructions on downloading the model and model data here...

Downloading Instructions | Community Earth System Model

www.cesm.ucar.edu

Note, when you setup a case with the model will download just the data that's needed to run the particular case in question. There is too much data to download it all at once.

 

[erik]

@nusbaume I don't see the casedirectory in the svn experiment database server for f.e21.FHIST_SSP585.f09_f09_mg17.CMIP6-lfmip-pdLC.001. Could you look into that?

 

[nick]

Thanks @erik, so the compset for f.e21.FHIST_SSP585.f09_f09_mg17.CMIP6-lfmip-pdLC.001 case I want to mimic is FHIST_BGC. That compset by default looks like it runs off prescribed observed SST data (sst_HadOIBl_bc_0.9x1.25_1850_2017_c180507.nc). If I want an SSP585 F case I'll have to change the input streams for the docn component to point to 21st century B case SST output, as well as have greenhouse gases evolve according to SSP585. I imagine the case files for the above experiment would show me what I need to do to get that kind of experiment going. @nusbaume, would greatly appreciate if you could point me to the case files and data used for this F case.

 

[nusbaume]

Hi Nick and Erik,

Sadly I don't know anything about this experiment. However, since it seems to have been part of CMIP6 maybe @strandwg knows where the case files might be? Otherwise I will try to ask around to see if anyone else has an idea.

Thanks!

Jesse

 

[strandwg]

Attached is the tarfile for the caseroot of f.e21.FHIST_SSP585.f09_f09_mg17.CMIP6-lfmip-pdLC.001

 

[nick]

Thanks very much @strandwg, @nusbaume and @erik. I'll take a look at those case files and try reproduce this experiment.

 

[nick]

I see that the above case uses the compset SSP585_CAM60_CLM50%BGC-CROP_CICE%PRES_DOCN%DOM_MOSART_CISM2%NOEVOLVE_SWAV and points to the SST file sstice_b.e21.BHIST.f09_g17.CMIP6-hist.011-SSP5-8.5.002_0.9x1.25_yrs1850-2100_timeseries_c20190703.nc.

I built a vanilla case using the same compset and while I see some SSP585-specific datasets getting referenced (e.g. various emissions data), the SST boundary conditions still point to observations (not the above 1850-2100 SSP585 file). I assume this is because in the compset specification the ocean component is just set to PRES_DOCN. I also assume the above SST file is something that one has to create using the SST tools and some B case output, but I'd like to know if the above file is available to the public (thus letting people skip the lengthy step of fetching B case SST/seaice)? I had a look in the data trunk and couldn't find any file similar to this.

Lastly, does the above sound like the right setup for a SSP585 F case? i.e. using the above compset and SST file? It seems like it would have everything covered.

 

[nusbaume]

Hi Nick,

You are correct that copying a compset won't always set every relevant input (it depends on how that particular scenario or use case is configured in each component). For SSTs in F-compset simulations the SST file is usually set by CAM itself, but it appears that for the SSP585 scenario that is not the case (my guess is because it was originally designed for fully-coupled B-compset simulations).

In terms of your particular run, it does appear that the sstice_b.e21.BHIST.f09_g17.CMIP6-hist.011-SSP5-8.5.002_0.9x1.25_yrs1850 2100_timeseries_c20190703.nc file is publicly available (at least it is publicly available on NCAR's machines). so you should be able to just use it directly instead of having to create your own version.

I believe you can use it by simply modifying the variable SSTICE_DATA_FILENAME in env_run.xml to match what is in the provided SSP F-compset case, running preview_namelists in your case directory, and then running check_input_data --download to make sure the SST file has been acquired.

Of course if for any reason you are unable to download and/or use that SST file please let us know.

Thanks, and have a great day!

Jesse

 

[nick]

Thanks @nusbaume, I don't have an account at NCAR, so I would appreciate it if someone could put this SST file on an FTP somewhere. Besides that, I think I should have everything I need to do my own SSP585 F case.

 

[cacraig]

I have added the file sstice_b.e21.BHIST.f09_g17.CMIP6-hist.011-SSP5-8.5.002_0.9x1.25_yrs1850 2100_timeseries_c20190703.nc to the inpudata repository. It should now automatically download when you run preview_namelists either separately or as part of case_build.

 

[nick]

Thanks @cacraig, I've started this run. It literally spends 3.5 days interpolating data first (and still ironing out bugs in namelists etc), but I certainly have the SST data now. Thanks.

 

[xiaoxiaokuishu]

Hi Nick,

You are correct that copying a compset won't always set every relevant input (it depends on how that particular scenario or use case is configured in each component). For SSTs in F-compset simulations the SST file is usually set by CAM itself, but it appears that for the SSP585 scenario that is not the case (my guess is because it was originally designed for fully-coupled B-compset simulations).

In terms of your particular run, it does appear that the sstice_b.e21.BHIST.f09_g17.CMIP6-hist.011-SSP5-8.5.002_0.9x1.25_yrs1850 2100_timeseries_c20190703.nc file is publicly available (at least it is publicly available on NCAR's machines). so you should be able to just use it directly instead of having to create your own version.

I believe you can use it by simply modifying the variable SSTICE_DATA_FILENAME in env_run.xml to match what is in the provided SSP F-compset case, running preview_namelists in your case directory, and then running check_input_data --download to make sure the SST file has been acquired.

Of course if for any reason you are unable to download and/or use that SST file please let us know.

Thanks, and have a great day!

Jesse

Hi, do you know where is the error come from



when i use historical SST, the model works well. But when I switch to use the sst data in the above picture, the error come.

Best