Vertical interpolation to GFS hybrid levels.

[marencibia]

Hi,

I am trying to initialize global MPAS from 0.25d real time GFS data and use the same vertical levels in the model. Using the default config_extrap_airtemp = lapse-rate in my namelist, I get this error which seems to indicate it can't interpolate temperature at the model top:

ERROR: extrap_type == 2 not implemented for target_z >= zf(1,nz)
ERROR: *****************************************************************
ERROR: Error in interpolation of t(k,iCell) for k= 127, iCell= 1
CRITICAL ERROR: *****************************************************************

For the levels I am using a table for config_specified_zeta_levels pulled from here (column Z(h)): https://www.emc.ncep.noaa.gov/gmb/wx24fy/misc/GFS127_profile/hyblev_gfsC128.txt

I can however successfully generate static fields with config_extrap_airtemp set to the other available options linear and constant, but would like to be able to run the default.

Any advice is greatly appreciated, thanks.

 

[mgduda]

It looks like the 'lapse-rate' option will fail if the height of an MPAS model level is greater than or equal to the height of the highest level in the column of input data. It seems like we could relax this check so that if the height of the MPAS level is equal to the height of the highest level in the input data, we can proceed (no extrapolation required).

Could you modify this line of code, changing '>=' to '>' ? MPAS-Model/src/core_init_atmosphere/mpas_init_atm_cases.F at v8.3.1 · MPAS-Dev/MPAS-Model

Alternatively, it might be worth verifying that the 'lapse-rate' option works if you reduce the height of the MPAS model top from 80110 m to something like 80109.9 m in your namelist when generating a vertical grid (config_vertical_grid = true in the &preproc_stages namelist group).

 

[marencibia]

Thanks for the reply Michael! Unfortunately the init cases hotfix did not get around this error.

I was hopeful for the namelist fix and also tried the second highest level (77323m) in my list, but both choices give the same error.

Since I am using a file for config_specified_zeta_levels, doesn't this override the model top setting in the namelist? I also tried removing the top level here, but that didn't work either.