Hydrological Model Performance Metrics Explained: NSE, KGE, RMSE & More
📌 Description
Evaluating hydrological model performance is not as simple as reporting a single metric.
In this video, we take a comprehensive and practical look at performance metrics in hydrological modelling, explaining why relying on just one metric (such as NSE) can be misleading.
We cover:
• Why single-metric evaluation fails
• A detailed explanation of 10 commonly used performance metrics
• Their mathematical formulation, interpretation, and limitations
• How different metrics capture different aspects of the hydrograph (peaks, low flows, timing, volume)
• Practical examples demonstrating how metrics can lead to different conclusions
• Best practices for multi-metric model evaluation and calibration
Metrics covered in this video:
• NSE (Nash–Sutcliffe Efficiency)
• Log-NSE
• KGE (Kling–Gupta Efficiency)
• R² (Coefficient of Determination)
• RMSE (Root Mean Square Error)
• MAE (Mean Absolute Error)
• PBIAS (Percent Bias)
• VE (Volumetric Efficiency)
• PFB (Peak Flow Bias)
• TPE (Timing of Peak Error)
This video is essential for hydrologists, water engineers, researchers, and students working in rainfall–runoff modelling, flood analysis, and model calibration.
________________________________________
⏱️ Timestamps
00:00 – Introduction: The Problem with Single Metrics
00:43 – Why One Metric Is Not Enough
02:27 – NSE: Definition, Use, and Limitations
03:13 – Log-NSE and Low Flow Performance
03:45 – KGE and Its Components
04:40 – R², RMSE, and MAE Explained
06:46 – Bias Metrics: PBIAS and VE
07:33 – Peak and Timing Metrics: PFB and TPE
08:04 – Numerical Examples: Comparing Metrics
11:58 – Best Practices for Model Evaluation
12:57 – Key Takeaways
________________________________________
🔑 Keywords
#hydrologytutorial #hydrologicalmodelling #modelperformance #performancemetrics #rainfallrunoffmodel #modelcalibration #modelvalidation #peakestimation #civilengineering #nse #correlationcoefficient #kge #RMSE #PBIAS