Author(s): Mohamed Saber; Ryoya Furuie; Ahmed Emara; Sameh Kantoush; Tetsuya Sumi; Emad Mabrouk
Linked Author(s): SAMEH KANTOUSH, Tetsuya Sumi, Mohamed Saber
Keywords: Suspended Sediment Concentration (SSC) Machine Learning Memetic Programming Miwa Dam Sediment Management
Abstract: This study investigates the application of memetic programming (MP) to predict suspended sediment concentration (SSC) in river systems, focusing on Japan's Miwa Dam. The MP model, using hourly data from 2020 and 2023, was tested under three scenarios: real-time predictions, short-term future predictions, and hybrid approaches combining historical and real-time inputs. Results showed high prediction accuracy, with correlation coefficients (R) ranging from 0.88 to 0.99 and normalized Nash-Sutcliffe coefficients (NNSC) between 0.77 and 0.98. Real-time predictions using historical data yielded the best outcomes. The MP model captured SSC dynamics during flood events and demonstrated reliable short-term predictions (R2:0. 97–0.99), though accuracy declined for extended forecasts. These findings highlight the MP model's potential to enhance reservoir operations, reduce sedimentation issues. Future work should include additional parameters and applications to other reservoirs.
Year: 2025