Streamflow forecasting using Artlficial Neural Networks

No Thumbnail Available
Mohammed Eisa, Hatim
Journal Title
Journal ISSN
Volume Title
Streamflow forecasting is an essential prerequisite to provide basic information on a wide range of flow-related activities and problems on natural and regulated rivers, such as, irrigation development, flood control, hydro-power generation, drinking water supply, and early warning systems. Streamflow forecasts can be provided by the tool of hydrological modeling. Many Different types of models have been reported in the hydrologic literature. While conceptual or physically-based models are important in the understanding of hydrological process, there are many practical situations where the main concern is with making accurate predictions and forecasts at specific locations. In such situation it is preferred to implement a simple “black box” model to identify a direct mapping between the inputs and outputs without detailed consideration of the internal structure of the physical process. This study presents the use of the recently developed modeling approach which is known as the Artificial Neural Networks (ANNs) to the problem of streamflow forecasting, and investigate its capability and applicability in simulating both rainfall-runoff as well as channel routing processes. The performance of the ANNs was then compared with other numerical simulators such as the Total Linear Model (TLM), the Modified Linear Perturbation Model (MLPM), and the Autoregressive (AR) Model. All the models were applied to the medium sized (166,875 km2, 8°-15°N, 34°-39°E) Atbara River sub-basin which is part a of the Nile River basin. Six years of real data of lumped daily rainfall and streamflow were used during the study. The results indicate that the nonlinear ANN model approach provides superior performance over all models across the full range of flow levels. Consequently, these results suggest that the ANN approach may provide a superior alternative to the Total Linear Model (TLM) and the Modified Linear Perturbation Model (MLPM) approaches for developing input-output simulation and forecasting models in situations that do not require modeling of the internal structure of the watershed.