# Spatial domains and grids for S2S-FM-VN.

target_grid:                # LOCKED: the exact 0.25-deg observed Tx grid (VnGAD)
  name: VnGAD_0.25
  source_file: data/VnGAD_daily-0.25_1961-2023_tasmax.nc
  lon: {min: 102.0, max: 109.5, n: 31, step: 0.25}
  lat: {min: 8.5,  max: 23.5, n: 61, step: 0.25}
  units: degC
  valid_land_cells: 586       # from actual valid data (time-stable), of 61x31=1891
  fill_value: -999.0
  note: >
    Product domain is every cell with valid observed Tx; coastal/border cells
    that extend partly over sea or across the border are NOT dropped.

region_mask:                 # R1-R7, matches the VnGAD target grid exactly
  file: /work/projects/G09/QG.25.196/OBS/mask7reg_VnGAD_0.25.nc
  var: vn_7regions
  codes: [1, 2, 3, 4, 5, 6, 7]
  cell_counts: {R1: 51, R2: 105, R3: 37, R4: 88, R5: 63, R6: 96, R7: 101}
  on_target_grid: true
  note: >
    541 of the 586 valid Tx cells carry a region code; 45 valid (mostly
    coastal/border) cells are unclassified -> handle explicitly in regional
    tables. NOT symlinked into data/; read in place. Never redraw boundaries.

# DECISION ITEM (gate): data/Mask_VN_S2S_EC_0.25.nc is on a DIFFERENT grid
# (65x41, 101-111E/8-24N) and does NOT align with the VnGAD target grid.
# It appears to belong to a sibling S2S-EC project; not used here.
ec_mask_not_used:
  file: data/Mask_VN_S2S_EC_0.25.nc
  grid: {lon: [101.0, 111.0, 41], lat: [8.0, 24.0, 65]}
  reason: "wrong grid for the VnGAD target; confirm at gate"

predictor_domains:           # coarse; NEITHER upsampled to 0.25 before encoder
  local_regional:            # LOCAL branch
    lon: [95.0, 120.0]
    lat: [5.0, 30.0]
    area_NWSE: [30, 95, 5, 120]
    rationale: "Vietnam, Indochina, South China Sea, S China; regional circulation, moisture, land-atmosphere"
    retrieval_groups: [sfc_core]     # target-side mx2t6 + regional surface
  large_scale:               # LARGE-SCALE branch
    lon: [40.0, 180.0]
    lat: [-20.0, 50.0]
    area_NWSE: [50, 40, -20, 180]
    rationale: "tropical Indian Ocean, Maritime Continent, W/C Pacific, East Asia; ENSO, MJO, monsoon, WNPSH"
    retrieval_groups: [pl_hi]        # circulation Z500/U200/T850/q; (later: SST etc.)
  note: >
    Pressure/circulation predictors MUST be retrieved on the large-scale domain to
    capture ENSO/MJO/monsoon/teleconnections; the local box is too small for them.
    Filenames carry a domain tag: _loc (local) / _ls (large-scale).