Title: Analysing of rainfall-runoff modelling using a hybrid DNN-SGD optimisation in Sub Basin of Brahmaputra River, India

Authors: Subha Sinha

Addresses: Department of Civil Engineering, Bakhtiyapur College of Engineering, Bakhtiyapur, India; Department of Science and Technology, Government of Bihar, India

Abstract: The main objective of this research is to improve the accuracy of runoff prediction and assess the effectiveness of the proposed DNN-SGD model. The performance of the DNN-SGD model is evaluated using standard metrics, including the coefficient of determination (R2), mean squared error (MSE), and root mean squared error (RMSE). The significance of findings demonstrating the superior performance of the proposed DNN-SGD model compared to other widely used methods, including ANN, DNN, ANN-PSO, and ANN-SGD. The results indicate that the DNN-SGD model achieved a remarkably high R2 of 0.99998, indicating its ability to capture a large proportion of the variability in the observed data. Moreover, it obtained the lowest RMSE value of 0.002252 and MSE value of 0.000507, further confirming its superior accuracy and predictive capabilities. Overall, this study providing an advanced rainfall-runoff modelling approach for water resource management in the Brahmaputra River Sub Basin and other similar regions.

Keywords: rainfall; runoff; flood; deep neural network; DNN; stochastic gradient decent; SGD; Brahmaputra River; ANN PSO; water management; India.

DOI: 10.1504/IJHST.2025.143132

International Journal of Hydrology Science and Technology, 2025 Vol.19 No.1, pp.52 - 72

Received: 20 Feb 2023
Accepted: 12 Nov 2023

Published online: 03 Dec 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article