bad mpas L93 default ncdata files

[adamrher]

What happened?

The FMTHIST compsets blow-up of the box using mpas, which @gdicker1 tracked down to missing values in the real world inic files for the 120 km and 480 km grids:

<ncdata hgrid="mpasa480" nlev="93" >atm/cam/inic/mpas/cami_01-01-2000_00Z_mpasa480_L93_CFSR_c240814.nc</ncdata>
<ncdata hgrid="mpasa120" nlev="93" >atm/cam/inic/mpas/cami_01-01-2000_00Z_mpasa120_L93_CFSR_c240814.nc</ncdata>

The issue seems to be that the ~80 km model top in L93 is well above the ~60 km CFSR top used to create the inic. One solutions is to use the ERA5 model level dataset, which has an ~80 km top and only requires extrapolation into the top most layer of L93 mpas. New inic's could be generated using betacast or the mpas inic generator.

I have not checked whether the analytic ic inic's are similarly problematic:

<ncdata hgrid="mpasa480" nlev="93" analytic_ic="1" >atm/cam/inic/mpas/mpasa480_L93_notopo_coords_c240814.nc</ncdata>
<ncdata hgrid="mpasa120" nlev="93" analytic_ic="1" >atm/cam/inic/mpas/mpasa120_L93_notopo_coords_c240814.nc</ncdata>
<ncdata hgrid="mpasa60"  nlev="93" analytic_ic="1" >atm/cam/inic/mpas/mpasa60_L93_notopo_coords_c240814.nc</ncdata>

What are the steps to reproduce the bug?

Run FMTHIST with mpasa120_mpasa120 or mpasa480_mpasa480.

What CAM tag were you using?

cam6_4_038

What machine were you running CAM on?

CISL machine (e.g. cheyenne)

What compiler were you using?

Intel

Path to a case directory, if applicable

No response

Will you be addressing this bug yourself?

Yes, but I will need some help

Extra info

No response