DeepFlowNet: Deep Learning Based Daily Water Flow Forecasting of Test River

River flow Forecasting is crucial for managing hazards and water resources, particularly in hilly areas where there is a significant reliance on downstream flows. In water resource management, incorporating methods such as time series, conceptual modeling, and artificial neural networks (ANNs) improves river flow forecasting accuracy. In this research, three most popular deep learning frameworks such as Deep Neural Network (DNN), Gated Recurrent Unit (GRU) and Long Short Term Memory Network (LSTM) have been employed to forecast the flow of river Test accurately. This dataset contains the daily gauged flow rate(m^3/sec) of the River Test, and climatic data (precipitation and temperature). A set of ten input combinations with lagged river flow values are developed. The target value is smoothed using a Gaussian filter with a sigma value of 5. The results show that LSTM performed the best followed by GRU and DNN. In measurable terms, LSTM, GRU and DNN models was demonstrated by R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>= 0.93, 0.915, and 0.876, RMSE = 0.393, 0.412, and 0.498, MAE = 0.247, 0.257, and 0.323, MAPE 0.337, 0.328, and 0.432, and MSE = 0.155, 0.169, and 0.248 respectively. The results strengthen the argument that deep learning algorithms particularly LSTM outperforms other above-mentioned algorithms for forecasting rivers flow with supreme accuracy.