# Data paths. Originals are read IN PLACE and never modified/moved.

observations:
  gridded_tx:              # PRIMARY (targets, training, CV, spatial verification)
    file: data/VnGAD_daily-0.25_1961-2023_tasmax.nc
    var: tasmax
    units: degC
    time: {start: "1961-01-01", end: "2023-12-31", n: 23010, includes_feb29: true}
    fill_value: -999.0
    provenance: "VnGAD, REMOSAT lab (Thanh N-X)"
  region_mask:
    file: /work/projects/G09/QG.25.196/OBS/mask7reg_VnGAD_0.25.nc
    vars: [vn_7regions, vn_1regions]
    provenance: "Nguyen Duy Tung, 2025; cdo setgrid to VnGAD 0.25 grid"
  station_tx:              # SECONDARY validation
    status: NOT_FOUND
    note: >
      No station Tx file located in data/ or the OBS source dir. If the gridded
      VnGAD product was built partly from stations, station verification is not
      independent. GATE ITEM: provide station Tx path + station->grid provenance.

other_products_present_not_used:
  - file: /work/projects/G09/QG.25.196/OBS/Tx_EC_0.25_2002-2023.nc   # EC-grid Tx, del29feb
  - file: /work/projects/G09/QG.25.196/OBS/Pr_1961-2023.nc           # precipitation
  - file: data/Mask_VN_S2S_EC_0.25.nc                                # EC-grid mask (wrong grid)

derived_outputs:           # created by this project only, in NEW folders
  p90:            data/processed/P90_2003_2022.nc
  ehd_threshold:  data/processed/EHD_THRESHOLD_2003_2022.nc
  ecmwf_raw_dir:  data/raw/ecmwf_s2s_1p5deg/     # to be created on approved download

ecmwf:                              # DECIDED (D1/D2): ECDS -> cdsapi -> s2s-reforecasts
  access_path: "ECDS -> cdsapi -> s2s-reforecasts (cf only, coarse 1.5deg)"
  dataset: s2s-reforecasts
  originating_centre: ecmwf
  forecast_member: control_forecast   # cf only
  predictor_grid: "1.5/1.5"           # never retrieve predictors at 0.25
  credentials:
    use: cdsapirc_ECDS
    cdsapirc: {path: ~/.cdsapirc, required_url: "https://ecds.ecmwf.int/api",
               current_url: "https://cds.climate.copernicus.eu/api",
               status: PENDING_ECDS_TOKEN}   # user configures token locally
    ecmwfapirc_legacy: {path: ~/.ecmwfapirc, url: "https://api.ecmwf.int/v1",
                        status: TOKEN_EXPIRED, use: false}   # NOT used
  request_source: "reproduce from ECDS 'Show API request code' -> config/ecds_minimal_request.json"
  grib_workflow: "ecCodes CLI (grib_ls/grib_get) + CDO -> NetCDF/Zarr -> xarray; keep original GRIB; cfgrib optional"
  note: >
    Never paste a key into chat or the repo. Run scripts/01_minimal_ecmwf_retrieval.py
    for the minimal ECDS test once ~/.cdsapirc points at ecds.ecmwf.int. Do NOT start
    the 2003-2022 bulk download yet.