This test case is set up to create the domains for the MISMIP3D test case, but the
tests themselves must be manually run.

There are two domain sizes:
minimal_width - this is only 3 cells wide laterally, and it useful for long spinups
full_width - this is the full width laterally, and is needed for the perturbation experiments

Running the setup scripts would create the domains for the size selected.  Note the highest
resolutions of the full_width size must be generated manually because they take non-trivial
amounts of time/memory to complete.  Still, the setup script organizes the tools needed, making
this fairly straightforward.

After running the desired resolution Stnd experiment to steady state, setting up and running the
perturbation experiments must be done manually.  See the other README file in this directory
for instructions.


One strategy is to run Stnd to steady state for a coarse resolution and then use the final state
of the coarse run as the initial condition for a finer resolution.  I compared this piggyback strategy
to a full spinup of the 1km resolution (with GLP on) and found the steady state solution for both
strategies to have an identical grounding line.  The full spinup took about 25,000 years to equilibrate
while the piggyback method took about 5000 years.
To do this, first make sure you've run the standard IC set up script on the finer resolution file, then:
MPAS-Tools/grid_gen/landice_grid_tools/interpolate_to_mpasli_grid.py -m d -s coarse-res-output-or-restart-file-with-final-time.nc -d finer-res.nc --thickness-only
This makes use of the extrapolation that the 'd' method uses, and interpolates the thickness field only.

