Monthly streamflow simulation and driving mechanism analysis using periodization-based autocorrelation and machine learning methods

Key Points

• To improve the accuracy of monthly streamflow simulation predictions using periodization-based autocorrelation and machine learning methods.
• Applied periodization-based autocorrelation to determine optimal lag times for different flow periods
• Optimized model combinations using TOPSIS scoring method
• Implemented SHAP analysis to identify driving factors for monthly flow dynamics.
• Optimal lag periods determined for base flow, transitional flow, and peak flow phases
• Achieved a Nash efficiency coefficient of 0.91 with a combined random forest and LSTM model
• Increased predictive accuracy by 68.2% compared to single models, enhancing extreme flow simulation accuracy.

Abstract