Add processing of climatological Thompson MP tracers to chgres_cube #120
Conversation
Baseline initial module to read thompson mp climo file.
Add read of thompson data pressure levels. Compute horizontal interpolation weights to transform between thompson and model grid.
Merge branch 'develop' into feature/chgres_thom_mp
Add horizontal interpolation of qnifa and thomp press. Begin new vertical interpolation routine for thomp tracers.
Merge branch 'develop' into feature/chgres_thom_mp
Add vertical interpolation of qnifa.
Add write of qnifa.
Add processing and write of qnwfa. Rename some variables.
Time interpolate thompson fields (linearly) using logic borrowed from sfcsub.F
Merge branch 'develop' into feature/chgres_thom_mp
Switch to Dom's climo database with much more logical records.
Read in path/name of thomp climo file via namelist. Add logic to process or not process thomp tracers based on this namelist setting. Add clean up of thomp fields when no longer needed.
Add Thomp tracers to lateral boundary file.
Change thompson tracer record names in coldstart/boundary files.
Account for thomp tracers in output file num-tracer records. General cleanup.
Isolate thomp tracer processing into subroutines.
Remove some unused variables.
There was a problem hiding this comment.
This is great work. We tested it successfully for global C96 and regional C768 runs with the full FV3_GSD_v0 suite. For the global setup, the results are very close to the old aeroconv utility.
For the regional setup, the new chgres_cube version works better than the old aeroconv utility. With the aeroconv input files, the regional C768 runs crashed after 24h because of excessive aerosols in the mid to upper levels (aeroconv was never fully tested for regional setups). With the chgres_cube input files, the 48h forecast completes successfully. Further, chgres_cube also adds the aerosol climatology fields to the lateral boundary conditions, not only to the initial conditions. aeroconv can only do initial conditions. This difference and an apparently higher resolution for interpolating the aerosol climatology before creating the initial conditions in chgres_cube are more realistic in any case.
Thanks for this great work.
Successfully tested with two 48-hour model runs - a C96 global uniform and a C768 regional.
For details, see #106.