Title: Nonlinear multiple regression analysis for predicting seasonal streamflow using climate indices for New South Wales

Authors: Rijwana Esha; Monzur A. Imteaz

Addresses: Department of Civil and Construction Engineering, Swinburne University of Technology, Melbourne, VIC 3122, Australia ' Department of Civil and Construction Engineering, Swinburne University of Technology, Melbourne, VIC 3122, Australia

Abstract: This paper presents development of streamflow prediction models with long-lead timescale using the Multiple Non-Linear Regression (MNLR) technique. Four major climate indices which were found to be influencing the streamflow of New South Wales (NSW) are used for this purpose. The developed models with all the possible combinations show good results in terms of Pearson correlation(r), Mean Absolute Error (MAE), Root Mean Square Error (RMSE) and Willmott index of agreement (d). The outcomes of MNLR models are compared to the best models of Multiple Linear Regression (MLR) analysis. MNLR models are evident to outperform the MLR models in terms of Pearson correlation (r) values, confirming the non-linear relationship between seasonal streamflow and large-scale climate drivers. Though the correlation values are not very high, they are statistically significant. The correlations obtained varied from 0.38 to 0.53 during calibration period, while it improved during the validation period, ranging from 0.52 to 0.63.

Keywords: multiple nonlinear regression; MNLR; multiple linear regression; MLR; climate indices; streamflow; seasonal forecast; New South Wales; NSW.

DOI: 10.1504/IJHST.2024.135186

International Journal of Hydrology Science and Technology, 2024 Vol.17 No.1, pp.101 - 116

Received: 23 Sep 2021
Accepted: 11 Aug 2022
Published online: 01 Dec 2023 *

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