Converting cubed sphere to lat/lon grid with JEDI - JEDI training and hands-on examples

Prerequisite¶

Please review here to learn about setting up your work environment and configuring the jedi-bundle.

Before starting, verify:

$JEDI_BUILD environment variable is set
fv3jedi_converttostructuredgrid.x exists in $JEDI_BUILD
Input files exist in the jedi_utils/run_convert_to_latlon/inputs directory (background, observation, and auxiliary files such as geometry configurations)
You have access to compute resources (interactive node or batch system)

Usage and limitations¶

fv3jedi_converttostructuredgrid.x can be used to convert a cubed sphere file to a structured (lat/lon) grid. However, the user needs to specify which model level they want to perform the conversion on. This means that the executable doesn’t automatically regrid all the model levels.

An example of an input YAML file is below. You will need to specify:

Input geometry (under state geometry)
Details about the input file (under state)
Details about the structured grid interpolation. More details available [here].(src/oops/base/StructuredGridWriter.h)

YAML structure:¶

states to structured grid:
- state geometry:
    fms initialization:
      namelist filename: geometry_input/fmsmpp.nml
      field table filename: geometry_input/field_table_gmao
    akbk: geometry_input/akbk72.nc4
    npx: 13
    npy: 13
    npz: 72
    field metadata override: geometry_input/geos_cf.yaml

  state:
    datetime: 2020-09-03T15:00:00Z
    filetype: cube sphere history
    state variables: [DELP,NO2,NO,O3,PS,SPHU]
    datapath: geos_c12/
    filename: geos_cf.bkg.20200903_150000z.nc4
    state variables: &statevars
    - air_pressure_thickness
    - volume_mixing_ratio_of_no2
    - volume_mixing_ratio_of_no
    - volume_mixing_ratio_of_o3
    - air_pressure_at_surface
    - water_vapor_mixing_ratio_wrt_moist_air
    field io names: &field_io_names
      air_pressure_thickness: DELP
      volume_mixing_ratio_of_no2: NO2
      volume_mixing_ratio_of_no: "NO"
      volume_mixing_ratio_of_o3: O3
      air_pressure_at_surface: PS
      water_vapor_mixing_ratio_wrt_moist_air: SPHU


  structured grid interpolation:
    grid type: regular gaussian
    local interpolator type: atlas interpolator
    interpolation method:
      type: finite-element
    number of latitude gridpoints: 12
    variables to output: *statevars
    pressure levels in hPa:
    - 925
    - 850
    - 700
    - 500
    - 250
    all model levels: true
    bottom model level: false
    datapath: outputs
    prefix: converttogauss

Running convertstate¶

$MPIEXEC "-n" "6" $JEDI_BUILD/fv3jedi_converttostructuredgrid.x converttolatlon_geos.yaml 2>&1 | tee log_convert_to_latlon.txt

In this example, two files will be generated. The “PressureLevels” file will include the variables to output values in latlon grid for levels specified under pressure levels in hPa. Since, all model levels is true the “modelLevels” files will include the variables to output values in latlon grid for all model levels. To learn more about other options refer to StructuredGridWriter.h in OOPS.

Ctest example:¶

An example of this application is ctest fv3jedi_test_tier1_converttolatlon_gfs in fv3-jedi. In this example, filetype of fms restart is converted from cubed sphere to latlon grid.