Abstract Reliable seasonal streamflow forecasting is critical for effective water resources management in Victoria, Australia, where hydroclimatic variability is strongly influenced by large-scale ocean–atmosphere interactions. This study develops Artificial Neural Network (ANN) models driven by the lagged climate predictors El Niño Southern Oscillation (ENSO) and Indian Ocean Dipole (IOD) to forecast Victorian spring streamflow across multiple catchments exhibiting pronounced climatic and hydrological heterogeneity. Model performance was assessed with Multiple Linear Regression (MLR) used as a benchmark. Across all stations, the ANN models consistently outperformed MLR, demonstrating substantial improvements in predictive accuracy. In eastern Victorian catchments, where Pacific climate drivers exert stronger hydrological control, validation and testing Pearson Correlation Coefficients (R) attained values of 0.68–0.86, alongside substantial reductions in Mean Squared Error (MSE) from 0.04–0.06 under MLR to 0.015–0.03 using the ANN models. In central regions characterized by interacting ENSO and IOD influences, predictive accuracy improved markedly, increasing from R values of 0.18–0.38 to 0.60–0.88 and corresponding decreases in Root Mean Squared Error (RMSE) from approximately 0.17–0.22 to 0.10–0.15. Even in western catchments, where climate-streamflow coupling is comparatively weaker, the ANN reduced the Mean Absolute Error (MAE) from 0.11–0.18 to 0.08–0.12 while maintaining physically plausible bias levels. These findings demonstrate that integrating lagged climate predictors within an ANN approach enhances the representation of delayed teleconnection effects on seasonal streamflow, contributing to improved understanding of climate-streamflow interactions, advancing nonlinear hydroclimatic modelling approaches, and informing the development of climate-driven streamflow prediction frameworks in southeastern Australia, which are transferable to other teleconnection-sensitive regions.
Bani et al. (Sat,) studied this question.