OSVAS

Step 2. Download validation data from ICOS specialized stations

Notebook execution

The Python workflow launcher executes the notebook Flux_downloader.ipynb to retrieve and process validation data. Like Step 1, this notebook can be executed via:

1. Automatic conversion (default): The launcher uses jupyter nbconvert to convert the notebook to Python and execute it without requiring a Jupyter server.
2. Direct Python execution: A pre-converted Flux_downloader.py script is available for standalone execution.
3. Interactive Jupyter: The notebook can be run manually with jupyter notebook or jupyter lab.

Validation data processing

The notebook downloads ICOS flux data and creates OBSTABLE SQLite files for use with HARP or custom verification tools. The process is configured via the Validation_data YAML section, which defines:

• ICOS datasets to download using their DOI
• Variable mapping from ICOS to observation variable names
• Unit conversions
• Validation period

Configuration example:

Validation_data:
  validation_start: '2021-05-01 00:00:00'  # Validation period start
  validation_end: '2021-07-01 23:30:00'    # Validation period end
  common_obstable: false                   # Write to station-specific directory
  common_fctable: false
  dataset1:
    doi: https://meta.icos-cp.eu/objects/fPAqntOb1uiTQ2KI1NS1CHlB  # ICOS dataset DOI
    timedelta: 30                          # Time step (minutes)
    variables:
      SW_OUT: SW_OUT                       # Map ICOS to observation variable names
      LW_OUT: LW_OUT
      SW_IN: SW_IN
      LW_IN: LW_IN
      TS_*: TS_*                           # Wildcard: include all TS_* variables
      SWC_*: SWC_*                         # Wildcard: include all SWC_* variables
    units:
      SW_OUT: W/m2                         # Specify observation units (for conversion)
      LW_OUT: W/m2
      LW_IN: W/m2
      SW_IN: W/m2
      TS_1: degC
      TS_2: degC
      SWC_1: percent
      SWC_2: percent
  dataset2:
    doi: https://meta.icos-cp.eu/objects/V8Wjs15Fj0aDX1xmiNx2zPA-
    timedelta: 30
    variables:
      H: H                                 # Sensible heat
      LE: LE                               # Latent heat flux
    units:
      H: W/m2
      LE: W/m2

OBSTABLE output

The notebook creates SQLite files with validation observations in HARP-compatible format:

Directory structure:

• Station-specific: sqlites/OBSTABLES/validation/{STATION_NAME}/
• Cross-site (if common_obstable: true): sqlites/OBSTABLES/validation/common_obstables/

File structure: Monthly OBSTABLE files are created, containing observations from all configured datasets for that month.

Soil initialization profiles

If Initialization_data is configured in the station YAML, the notebook can compute soil temperature and humidity profiles from validation data for use in SURFEX initialization:

Initialization_data:
  metadata_filename: 'ICOSETC_ES-LMa_VARINFO_METEO_L2.csv'  # ICOS metadata file
  reuse_soil_profile: false        # Reuse previously computed profiles
  Init_to_namelist: true           # Apply profiles to OPTIONS.nam
  Init_to_prep: false              # Apply profiles to PREP files
  soil_depths_ts: [0.05, 0.1, 0.2, 0.3, 0.4]  # Soil temperature depths (m)
  soil_depths_swc: [0.05, 0.1, 0.2, 0.3, 0.4] # Soil moisture depths (m)

When this block is present, the notebook:

1. Reads the ICOS metadata file to locate soil observations at the specified depths
2. Extracts soil temperature (TS_) and moisture (SWC_) profiles from validation data
3. Saves profiles as NumPy arrays: profiles/{STATION_NAME}/tg_profile.npy and hug_profile.npy

These profiles are then applied by Step 3 when Init_to_namelist or Init_to_prep is enabled.

Soil depths are derived with the following priority: 1. Explicit lists in YAML: Initialization_data.soil_depths_ts (for TS) Initialization_data.soil_depths_swc (for SWC) If only one generic soil_depths key is found it is used for both. 2. ICOS Metadata file defined by Initialization_data.metadata_filename, looked up in config_files/{station_name}/. Check here to learn how to reach these files for every ICOS station.

Variable sampling rate

If multiple validation datasets with different sampling rates are provided, all data are upsampled to the finest (smallest) timedelta before creating the OBSTABLE.