Daily streamflow forecasting by machine learning in Tra Khuc river in Vietnam

Precise streamflow prediction is crucial in the optimization of distribution water resources. This study develops machine learning models by integrating recurrent gate unit (GRU) with bacterial foraging (BFO), gray wolf optimizer (GWO), and human group (HGO) to forecast Tra Khuc River, Vietnam. For this purpose, time series daily rainfall river flow at Son Giang station from 2000 2020 were employed streamflow. The statistical indices, namely root mean square error, absolute coefficient determination (R²), was utilized evaluate performance proposed models. results showed that three algorithms (HGO, GWO, BFO) effectively enhanced GRU model.
 Moreover, among four (GRU, GRU-HGO, GRU-GWO, GRU-BFO), GRU-GWO model outperformed other R² = 0.883. GRU-HGO achieved 0.879, GRU-BFO R²=0.878. combined a reliable modeling approach short-term forecasting.