Skip to content

FAQ

How do I use cloud (OSS/S3) data?

Set data_cfgs.source: cloud and configure storage.s3 in ~/hydro_setting.yml (bucket, prefix, access keys, endpoint URL). Cloud reads use zarr caches under s3://<bucket>/zarr/; missing caches are generated from raw files (write access required). source is a hard selection - it does not fall back to local automatically.

Why do I get an EndpointConnectionError when using cloud?

Check that this machine can reach the OSS/S3 endpoint, and that the storage.s3 block in ~/hydro_setting.yml is correct. Remember that a project-level .hydro_setting.yml overrides the user-level file key-by-key.

Why does my streamflow contain negative values like -28.28?

Raw USGS files mark missing flow as -999 (cfs). If the -999 values are converted to m^3/s (multiplied by 0.0283168) instead of being replaced with NaN, you get negative garbage values that corrupt metrics. Regenerate the data cache/zarr with the -999 -> NaN handling so missing days are excluded from objective functions.

What does "param_range_file not provided ... Using default MODEL_PARAM_DICT ranges" mean?

It is informational: no custom parameter range file was given, so the built-in parameter ranges are used. To customize ranges, set training_cfgs.param_range_file to a YAML file (strict validation: missing files, unknown/missing parameters, or invalid [min, max] ranges fail fast).

What is the difference between calibration, evaluation, and simulation?

  • Calibration searches for the parameter set that minimizes the objective (e.g. RMSE).
  • Evaluation runs the calibrated parameters on a period (e.g. test) and reports metrics (NSE, KGE, RMSE, PBIAS, ...).
  • Simulation runs the model with any parameters and returns the simulated series.

Evaluation internally runs a simulation, so you normally do not need to run simulation separately after calibration unless you want the raw simulated series.

Why is source: cloud in my YAML ignored?

source is read from data_cfgs.source. If it is missing, storage.default_source in hydro_setting.yml is used. Also verify you are running from the project root - a project-level .hydro_setting.yml overrides the user-level one.

Which Python versions are supported?

Python >= 3.11 (see pyproject.toml).