Interpret calibration results
What the diagnostics on your dashboard mean.
After a calibration run, your dashboard shows how well the model reproduces observed streamflow, on both the calibration and validation periods.
The diagnostics
- NSE (Nash–Sutcliffe Efficiency): 1.0 is perfect; above ~0.5 is generally acceptable for daily streamflow, higher is better.
- KGE (Kling–Gupta Efficiency): balances correlation, bias, and variability; closer to 1.0 is better.
- PBIAS: percent bias — positive means the model under-predicts volume, negative means it over-predicts; near zero is best.
- The hydrograph overlay shows simulated vs observed flow so you can see where the fit is strong or weak.
Compare the validation-period scores to the calibration period. Validation scores close to (or better than) calibration means the model generalizes rather than overfitting.
SWATGenX calibrates streamflow (hydrology). It does not claim calibrated nutrient, sediment, or PFAS loads — those are separate.