CAM4 idealized physics

[wrboos@fas_harvard_edu]

I am trying to do an idealized, dynamical core only (i.e. Held-Suarez) run of CAM4, and have attempted to configure and build it using the CCSM system by specifying CAM_CONFIG_OPTS="-phys ideal" in the compset.xml file for the new case. But after running create_newcase, configure, and the build script, the model crashes before taking its first time step.

I noticed that this configuration method did not set the atm_ideal_phys namelist, so I set that manually in the cpl.buildnml.csh file before building (I did this manually because I could not find a method to specify a namelist option for drv_in using the configure script). The model did run, but it failed to break zonal symmetry when I started it from an isothermal resting state. I tried specifying pertlim = 0.1 but this crashed the model too.

So, is there a use case, or somewhere I can find relevant namelists or .xml files for a Held-Suarez run with CAM4?

Thanks,
Bill Boos

 

[eaton]

There is no ccsm compset set up for running ideal physics forcings. For this reason (and also because I'm not enough of an expert with the ccsm scripts to have a good idea about how to fix this) I think you'll have better luck doing this type of run using the CAM standalone scripts. When CAM is run in standalone mode its build-namelist utility will automatically set the atm_ideal_phys variable in the drv_in namelist file.

Once you've done that, that's really all there is to setting up an ideal physics run. The problem you ran into with the isothermal resting state initial conditions may just be that more time was required to spin up the model. I don't know how long it should take starting from rest. When I've tested the ideal physics mode in the past I've used an initial file that the dynamics folks here use for baroclinic wave test cases. These initial files are available from the data repository. The available files are:

atm/cam/inic/fv/cami_baroclinic_wave_0.47x0.63_L26_c071227.nc
atm/cam/inic/fv/cami_baroclinic_wave_1.9x2.5_L26_c071227.nc
atm/cam/inic/fv/cami_baroclinic_wave_4x5_L26_c070821.nc

To use these files just set the ncdata variable in your namelist input.

 

[wrboos@fas_harvard_edu]

Just wanted to post my experience using idealized physics in CAM in case it is helpful to any other users.

I initially tried to get idealized physics runing in CAM 3.1/3.5. But idealized physics (== Held-Suarez setup) was broken for the finite volume core of CAM3.1 and 3.5 (as confirmed by Brian Eaton at NCAR in another bulletin board message). We did get the idealized physics to run in the semi-Lagrangian and Eulerian cores, but the results had some peculiarities when compared to Held-Suarez output from other models (e.g. the CAM jets were too far poleward). Furthermore, the structure of these peculiarities (e.g. jet core position) was sensitive to which of these cores we used and the resolution. I even showed our output to Christiane Jablonowski and she agreed they looked odd but didn't know why. It's possible I made an error in model configuration, but we chose not to spend time debugging since CAM 4 had released by that time and it produced successful Held-Suarez runs that calibrated well with other model cores.

When running idealized physics with CAM4, you will need to use the standalone CAM (i.e. it won't work if you try to build & run CAM using the CCSM scripts). Also, you may need to create your own initial condition file if you want to start from a state of rest; if you do so be sure to include a perturbation to break zonal symmetry (or I would be happy to share our i.c. file).