Run type

I can’t find the run type documentation for CESM2, but the analagous documentation from CESM1 can be found on this Running CESM page.

Note

There was some confusion on my part about whether the KiloCAM experiment should be a branch or a hybrid run. To extend the Reanalysis from 2019 to 2020, Kevin set up a branch run because it allowed him to change the SST forcing for the reanalysis while, at the same time, using restart files. For the KiloCAM experiment, we require using perturb_single_instance to generate the thousand members, which only works on initial files rather than restart files, so it should be set up as a hybrid run.

Initial files

The initial files from the CAM6 reanalysis are here:

/lustre/project/k1421/Reanalyses/f.e21.FHIST_BGC.f09_025.CAM6assim.011/rest/2018-01-01-00000

The individual filenames are of the form:

f.e21.FHIST_BGC.f09_025.CAM6assim.011.cam_00??.i.2018-01-01-00000.nc

Using perturb_single_instance

The &perturb_single_instance_nml namelist in input.nml has this structure:

&perturb_single_instance_nml
   ens_size               = 3
   input_files            = 'caminput.nc'
   output_files           = 'cam_pert1.nc','cam_pert2.nc','cam_pert3.nc'
   output_file_list       = ''
   perturbation_amplitude = 0.2
/

The simplest way to generate the initial ensemble is to use a symbolic link for caminput.nc and write the names of the desired perturbed initial files into the file declared in output_file_list.