Machine learning prediction of sediment yield index

• Published in: Soft computing (Berlin, Germany) Germany), 2023-11, Vol.27 (21), p.16111-16124
• Main Authors: Meshram, Sarita Gajbhiye, Hasan, Mohd Abul, Nouraki, Atefeh, Alavi, Mohammad, Albaji, Mohammad, Meshram, Chandrashekhar
• Format: Article
• Language: English
• Subjects: Algorithms; Application of Soft Computing; Artificial Intelligence; Artificial neural networks; Computational Intelligence; Control; Engineering; Machine learning; Mathematical Logic and Foundations; Mechatronics; Modelling; Multilayer perceptrons; Natural resources; Neural networks; Performance evaluation; Radial basis function; Robotics; Root-mean-square errors; Soil contamination; Soil erosion; Watersheds
• ISSN: 1432-7643; 1433-7479
• DOI: 10.1007/s00500-023-07985-5

Sediment output affects soil health maintenance, reservoir sustainability, environmental contamination, and natural resource preservation. Three different algorithms, the artificial neural networks-radial basis function (ANN-RBF), Artificial Neural Networks-Multilayer Perceptron, M5P tree, were used for this purpose in the Narmada river watersheds, India. For this purpose, fifteen different scenarios are considered as inputs to the models. For selecting the best-fit model, the performance of selected models was evaluated using performance criteria such as root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination ( R 2 ). The results indicated that the ANN-RBF models outperformed the other models in terms of accuracy, with RMSE, MAE and R 2 of 26.72, 19.84 and 0.98, respectively. The current study’s findings support the applicability of the proposed methodology for modeling the sediment yield index and encourage the use of these methods in alternative hydrological modeling.