Parameter setting for new e3sm_to_cmip task

[chengzhuzhang]

I'm testing zppy3.0.0 rc2 and try to understand how to set up the configuration file. This case is a little special because the target grids land and atmosphere data are remapped to are difference: land: esmfaave, and atm: traave. I'm having trouble to feed e3sm_to_cmip the right input files.

The session of my ts and e3sm_to_cmip is as follows, @forsyth2 could you please help take a look?

[ts]
active = True
years = "2000:2014:15"
walltime = "00:50:00"

  [[ atm_monthly_180x360_aave ]]
  input_subdir = "archive/atm/hist"
  input_files = "eam.h0"
  frequency = "monthly"
  mapping_file = /g/g92/zhang40/zppy_test/map_ne30pg2_to_cmip6_180x360_traave.20231201.nc
  vars = "FSNTOA,FLUT,FSNT,FLNT,FSNS,FLNS,SHFLX,QFLX,TAUX,TAUY,PRECC,PRECL,PRECSC,PRECSL,TS,TREFHT,CLDTOT,CLDHGH,CLDMED,CLDLOW,U,ICEFRAC,LANDFRAC,OCNFRAC,PS,CLDICE,CLDLIQ,T,AODDUST"
# Needed for mixed-phase partition
#  vars = "LANDFRAC,CLDICE,CLDLIQ,T"
 # ts_fmt = "cmip"


  [[ land_monthly ]]
  input_subdir = "archive/lnd/hist"
  input_files = "elm.h0"
  frequency = "monthly"
  mapping_file = "map_r05_to_cmip6_180x360_esmfaave.20231110.nc"
  vars = "FSH,RH2M,LAISHA,LAISUN,QINTR,QOVER,QRUNOFF,QSOIL,QVEGE,QVEGT,SOILICE,SOILLIQ,SOILWATER_10CM,TSA,TSOI,H2OSNO,TOTLITC,CWDC,SOIL1C,SOIL2C,SOIL3C,SOIL4C,WOOD_HARVESTC,TOTVEGC,NBP,GPP,AR,HR"
  extra_vars = "landfrac"
[e3sm_to_cmip]
active = True
frequency = "monthly"
walltime = "00:50:00"
ts_num_years = 15
years = "2000:2014:15"

  [[ atm_monthly_180x360_aave ]]
  # for ilamb
  input_files = "eam.h0"

  [[ land_monthly ]]
  # for ilamb
  input_files = "elm.h0"

[forsyth2]

https://github.com/E3SM-Project/zppy/blob/main/zppy/defaults/default.ini is the key to all zppy parameters.

[e3sm_to_cmip]
# Metadata json file
cmip_metadata = string(default="inclusions/e3sm_to_cmip/default_metadata.json")
# variables to run e3sm_to_cmip on
cmip_vars = string(default="")
# Model component having generated input files (eam, eamxx, elm, mosart, ...)
input_component = string(default="")
# Name of the grid used by the relevant `[ts]` task
ts_grid = string(default="180x360_aave")
# Name of the `[ts]` subtask to depend on
ts_subsection = string(default="")

The target grids land and atmosphere data are remapped to are difference

It sounds like you just want to set ts_grid to be different between the two subtasks then.

[chengzhuzhang]

Thank you for the pointer, Ryan!
Now I have a working version of the configuration file:

[e3sm_to_cmip]
active = True
frequency = "monthly"
walltime = "00:50:00"
ts_num_years = 15
years = "2000:2014:15"

  [[ atm_monthly_180x360_aave ]]
  # for ilamb
  input_files = "eam.h0"
  ts_grid = "180x360_traave"

  [[ land_monthly ]]
  # for ilamb
  input_files = "elm.h0"
  ts_grid = "180x360_esmfaave"

I think we discussed this before, although e3sm_to_diags sub tasks only need one parameter ts_grid, it might be more intuitive to have same set of parameters as those for ilamb jobs. For instance, here is my ilamb parameters:

[ilamb]
active = True 
nodes = 8
walltime = "2:00:00"
partition = compute 
short_name = 'extendedOutput.v3.LR.historical_0101'
ts_atm_grid = '180x360_traave'
ts_land_grid = '180x360_esmfaave'
ts_num_years = 15
years = "2000:2014:15"

I don't remember if there are other advantages for use just one parameter ts_grid for e3sm_to_cmip task...

[forsyth2]

@chengzhuzhang Looking at #650 (comment), it seems that was just to reduce the number of parameters. We can separate them though if that's more intuitive.

[chengzhuzhang]

I think to separate them might be more intuitive and consistent to ilamb. It looks like ts_grid is a new parameter only used by e3sm_to_cmip. I can see that the command here can be more complicated with two component depended grids. Maybe we can try support both ts_atm_grid/ts_lnd_grd and ts_grid to be more use-error tolerant.

[forsyth2]

Ok I can work on that for the next rc

[chengzhuzhang]

thank you! Another thing about updating e3sm_to_cmip task as a separate one is that the parameter ts_fmt is removed, I believe. Is it possible to implement a parameter removed warning to help user transition?