Combining Standard Artificial Intelligence Models, Pre-Processing Techniques, and Post-Processing Methods to Improve the Accuracy of Monthly Runoff Predictions in Karst-Area Watersheds

The complex and unique topography of karst regions highlights the weaknesses traditional hydrological models which fail to fully generalize them. successive proposals standard artificial intelligence (AI) models, pre-processing techniques, post-processing methods have provided new opportunities enhance accuracy runoff prediction in areas. In this study, first, BP neural network model Elman were used for prediction. Then, performance four coupled models—formed by combining two AI Empirical Modal Decomposition (EMD) Ensemble (EEMD), with previously mentioned models—was investigated. Finally, triple-coupled formed applying method quantile mapping (QM) previous was estimated. Nash–Sutcliffe efficiency (NSE), mean absolute percentage error (MAPE), root square (RMSE), peak threshold statistics (PPTS) selected evaluate analyze forecasting results above models. demonstrated that had best effect better than QM–EMD–Elman an NSE value 0.73, MAPE 0.75, RMSE 34.60, PPTS 2.36.