Assessing machine learning models for streamflow estimation: a case study in Oued Sebaou watershed (Northern Algeria)

ABSTRACT

This paper proposes runoff models based on machine learning to estimate daily streamflows in Oued Sebaou watershed, a Mediterranean coastal basin located in northern Algeria. Therefore, we applied random forest (RF), artificial neural networks (ANN – under different training algorithms), and locally weighted linear regression (LWLR) using as input combinations of current and past rainfall amounts and previous values of streamflow. We selected streamflow and rainfall records to calibrate and validate the stated approaches. We used root mean square error (RMSE) and correlation coefficient (R) to evaluate the accuracy of the models. Analyses of the results show that RF provided the best outcomes for both training (RMSE = 4.7458 and R = 0.9834) and validation (RMSE = 2.3617 and R = 0.9719). The ANN calibrated with the Levenberg-Marquardt algorithm presented the second-best result, outperforming its counterparts and LWLR.

Graphical Abstract

KEYWORDS:

• runoff
• machine learning
• training algorithms
• neural networks
• random forest
• Algeria

Editor A. Castellarin Associate Editor H. Tyralis

Editor A. Castellarin Associate Editor H. Tyralis

Acknowledgements

The authors are grateful to the Directorate General for Scientific Research and Technological Development for supporting this research, and to the engineers of the National Agency of Water Resources (ANRH), who provided us with the necessary data.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

The authors reported there is no funding associated with the work featured in this article.