Outputs

Output Reference

Directory structure

Single domain (per_domain_output: false)

output_dir/
├── CONUS_ensemble.nc
├── metrics.csv
├── hydrographs/
│   ├── hydrograph_01085500.png
│   └── ...
├── skill_maps/
│   ├── map_kge_ensemble_mean.png
│   ├── map_winner_kge.png
│   ├── boxplot_kge.png
│   └── vpu_breakdown_kge.png
├── interactive_metrics_map.html
└── animation_streamflow_mean.gif

Multi-domain (per_domain_output: true)

output_dir/
├── {domain_id}/
│   ├── {domain_id}_ensemble.nc
│   └── hydrographs/
├── metrics.csv          ← combined across all domains
├── skill_maps/
└── interactive_metrics_map.html

Ensemble NetCDF (*_ensemble.nc)

The primary operational output. Contains ensemble statistics for every flowpath feature_id at every timestep.

Variable	Description
streamflow_mean	Equal-weight arithmetic mean across formulations
streamflow_median	Median across formulations
streamflow_min	Minimum across formulations (small ensembles)
streamflow_max	Maximum across formulations (small ensembles)
streamflow_p05	5th percentile (large ensembles)
streamflow_p95	95th percentile (large ensembles)

Written as float32. Dimensions: (time, feature_id).

Metrics CSV (metrics.csv)

One row per (gage, source) pair.

Column	Description
feature_id	NHD+ feature ID of the gage flowpath
gage_id	Zero-padded 8-digit USGS gage ID
lat, lon	Gage location (WGS84)
source	ensemble_mean or formulation name
nse	Nash-Sutcliffe Efficiency
kge	Kling-Gupta Efficiency
pbias	Percent bias

Hydrographs

One PNG per gage with valid observations. Shows observed streamflow (black) and simulated ensemble mean (blue) with optional uncertainty band and individual formulation lines.

Skill Maps

File	Description
map_{metric}_{source}.png	Spatial scatter map, points coloured by metric value
map_winner_{metric}.png	Map showing which formulation has the best score at each gage
boxplot_{metric}.png	Distribution of metric scores per formulation
vpu_breakdown_{metric}.png	Stacked-bar chart of win rates by VPU (CONUS)