JEDI-ACE

Workflow to use VIND (Versatile Implementation for Native Data) interfaced with the JEDI components and METplus for Atmospheric Composition Evaluation (ACE).
VIND repo

Supported platforms

• Platforms with JCSDA spack-stack package.
• Module loading scripts:
  derecho_intel (v1.8.0), derecho_gnu (v1.9.2), derecho_oneapi (v1.9.2), orion_gnu (v1.8.0)

Tested models

• Lambert CC projection: WRF-Chem
• Reducing Gaussian Lon-Lat: GEFS-Aerosols
• Regular Lon-Lat: MERRA-2

Tested measurements

• TROPOMI NO₂ and CO
• AOD from MODIS, VIIRS, OCI (PACE), AERONET
• TEMPO NO₂, CO
• PANDORA NO₂
• AirNow O₃ and PM_2.5

Clone the code

Clone this repo into <folder> with the command below:
git clone https://github.com/weiwilliam/JEDI-ACE.git <folder>

Build VIND (VIND-bundle)

1. Create the <repo>/vind-bundle/build folder
2. Create virtual python env <repo>/venv if you do not have one.
   source ush/setup.sh <repo path> <platform> <compiler>
3. cd <repo>/vind-bundle/build
4. ecbuild <path/to/vind-bundle>
5. make -j <n>
6. ctest to check executables work properly

Existing builds

Derecho (oneapi): /glade/work/swei/projects/JEDI-ACE/vind-bundle/build
Orion (gnu): /work2/noaa/jcsda/shihwei/git/caliop_opr/genint-bundle/build (older version)

1. Update VIND_BUILD in ush/setup.sh to /glade/work/swei/projects/JEDI-ACE/vind-bundle/build
2. source ush/setup.sh <your/repo/path> <platform> <compiler>
   It will create venv for you and point your executables to my build.

• It may encounter permission issue

Use of this interface

• Prerequisites: observation files in IODA format and model outputs supported by VIND VIND README.

Options currently available:

Platform	Compilers
Derecho	gnu, intel, oneapi
Orion	gnu

1. Create Python venv under the cloned repo by running: source ush/setup.sh </repo/path> <platform> <compiler>

2. yamls/main and yamls/hofx3d include examples of input YAML files for different models (e.g. MERRA-2, WRF-Chem) and different observations (e.g. AOD, tracegas).
   For example to evaluate WRF-Chem trace gas you can use main/main_wrfchem_tracegas.yaml and it uses hofx3d/TraceGas_WRFChem.yaml.
   README.md under yamls/<main, hofx3d> provides more details.
   Modify or add your own YAML based on your application

3. Run your experiment by executing pyscripts/vind_vrfy.py <main yaml>

Preprocesses for use case of WRF system

1. Create air_potential_temperature for JEDI application:
   ncap2 -O -s "air_potential_temperature=T+300" <wrfout> <new wrfout>
2. Create o3_molpermol from o3 (ppmv) for JEDI application:
   ncap2 -O -s "o3_molpermol=o3*1e-6" <wrfout> <new_wrfout>
3. Cropping IODA file:
   Use pyscripts/get_wrfout_polygon.py to create a polygon .csv file for your domain boundary.
   Run pyscripts/crop_iodafile.py -i <global/IODA/file> -o <WRF/domain/IODA/file> -p <WRF/domain/polygon/csv>
4. Use P_HYD to represent air_pressure.
   The PSFC is a diagnostic variable derived through hydrostatic function, so the air_pressure_levels based on akbk, ptop, and PSFC are more close to hydrostatic. It may cause half level pressure from PB+P is not between two adjacent full level.
5. To get ak and bk values from wrfout,
   ak = C4F + P_top - C3F * P_top
   bk = C3F

Reference links

UFO operators: JEDI document/UFO