Modeling the damming effect on hydrological alteration and prediction of discharge in Padma River by proposing PSO based novel hybrid machine learning algorithm
Abu Reza Md. Towfiqul Islam1, Swapan Talukdar2, Shumona Akhter1, Kutub Uddin Eibek1, Md. Mostafizur Rahman3, Ashraf Dewan4, Quoc Bao Pham5,6, Nguyen Thi Thuy Linh 7,8*, Swades Pal9, Thi Ngoc Canh DOAN10, Duong Tran Anh11, Sobhy M. Ibrahim12, 13
This study quantified the hydrological alteration of the Padma River basin caused by the construction of Ferakka Barrage using different statistical models. The negative trend of the average discharge in the dry season was detected, while the average discharge was lower than environmental flows. PSO based ensemble machine learning models were developed. PSO-Reduced Error Pruning Tree (REPTree), PSO-random forest (RF), and PSO-M5P were the optimal fit for average, maximum, and minimum discharge prediction (RMSE = 0.14, 0.3, 0.18) respectively.