Deep Learning Data-Intelligence Model Based on Adjusted Forecasting Window Scale : Application in Daily Streamflow Simulation

Streamow forecasting is essential for hydrological engineering. In accordance with theadvancement of computer aids in this eld, various machine learning (ML) models have been explored tosolve this highly non-stationary, stochastic, and nonlinear problem. In the current research, a newly exploredversion of an ML model called the long short-term memory (LSTM) was investigated for streamowprediction using historical data for forecasting for a particular period. For a case study located in a tropicalenvironment, the Kelantan river in the northeast region of the Malaysia Peninsula was selected. Themodelling was performed according to several perspectives: (i) The feasibility of applying the developedLSTM model to streamow prediction was veried, and the performance of the developed LSTM modelwas compared with the classic backpropagation neural network model; (ii) In the experimental process ofapplying the LSTM model to the prediction of streamow, the inuence of the training set size on theperformance of the developed LSTM model was tested; (iii) The effect of the time interval between thetraining set and the testing set on the performance of the developed LSTM model was tested; (iv) The effectof the time span of the prediction data on the performance of the developed LSTM model was tested. Theexperimental data showthat not only does the developedLSTM model have obvious advantages in processingsteady streamow data in the dry season but it also shows good ability to capture data features in the rapidlyuctuant streamow data in the rainy season.Validerad;2020;Nivå 2;2020-03-02 (johcin)