PCR_GLOBWB_2

Run preflight_check.py first, then use make_clone_map.py with the gauge coordinates and drainage area to build clone and landmask maps, fetch inputs with fetch_pcrglobwb_inputs.py, convert forcing, write the .ini, and execute run_pcrglobwb.py. Extract discharge at the snapped gauge cell with parse_pcrglobwb_output.py.

It needs daily NetCDF files named exactly precipitation, temperature, and referencePotET or evapotranspiration. All depth quantities must be in m/day and temperature in degrees Celsius; mm/day precipitation or Kelvin temperature cause silent 1000x errors.

Precipitation was almost certainly supplied in mm/day instead of the required m/day, the most common silent failure (dt_001). Divide precipitation by 1000; the same trap applies to reference ET and soil storage depths.

The NetCDF variable names do not match the expected names exactly, so the model silently reads zeros (dt_009). Rename forcing variables to precipitation, temperature, and evapotranspiration or referencePotET.

No. The model runs on a daily timestep only, and groundwater is a single vertical linear reservoir per cell with no lateral inter-cell flow. It also excludes glacier dynamics and backwater effects in routing.

Extract daily discharge at the snapped gauge cell with --aggregation dailyTot and score against observed series using NSE and PBIAS conventions (satisfactory NSE 0.5, good 0.7 per Moriasi). The default global run is uncalibrated, with median gauge correlation around 0.60.