This study proposes a computationally efficient solution to stream flow forecasting for river basins where historical time series data are available. Two data-driven modeling techniques are investigated, namely support vector regression and artificial neural network. Each model is trained on historical stream flow and precipitation data to forecast stream flow with a lead time of up to seven days. The Shoalhaven, Herbert and Adelaide rivers in Australia are considered for experimentation. The predictive performance of each model is determined by the Pearson correlation coefficient, the root mean squared error and the Nash-Sutcliffe efficiency. The performance of our data-driven models are compared to that of a physical stream flow forecasting model currently supplied by Australia's Bureau of Meteorology. It is concluded that the data-driven models have the potential to be useful stream flow forecasting tools in river basin modeling.
Reference:
Steyn, M. et al. 2017. Short-term stream flow forecasting at Australian river sites using data-driven regression techniques. International Work-Conference on Time Series Analysis (ITISE 2017), University of Granada, Granada, Spain , 18-20 September 2017
Steyn, M., Wilms, J. M., Brink, W., & Smit, F. (2017). Short-term stream flow forecasting at Australian river sites using data-driven regression techniques. http://hdl.handle.net/10204/9895
Steyn, Melise, Josefine M Wilms, W Brink, and F Smit. "Short-term stream flow forecasting at Australian river sites using data-driven regression techniques." (2017): http://hdl.handle.net/10204/9895
Steyn M, Wilms JM, Brink W, Smit F, Short-term stream flow forecasting at Australian river sites using data-driven regression techniques; 2017. http://hdl.handle.net/10204/9895 .
Paper presented at the International Work-Conference on Time Series Analysis (ITISE 2017), University of Granada, Granada, Spain , 18-20 September 2017