Short-term Water Level Forecast Using ANN Hybrid Gaussian-Nonlinear Autoregressive Neural Network

The aim of this study is to develop the best forecast model using hybrid Gaussian-Nonlinear Autoregressive Neural Network water level with multiple hoursahead for Melaka River. developmentof flood models crucial and has led risk control, policy recommendations, a reduction in human life loss, flood-related property destruction. In research, Artificial (ANN) approach was used by modeling forecasting time series.ANN selected due its high reputation abilities learn from time-series data pattern. A total 2782data period one month wasused ANN training, validation, testing flash flood. study,Hybrid Gaussian Nonlinear (Gaussian-NAR) as series. This study's primary focus find most appropriate steps ahead, which are 1 hour, 3 hours, 5 7 hours. accuracy measures measured Pearson R R-squared accurate time-step ahead. result indicates that hours squared 86.7%. Gaussian-NAR 3-hour 99.8 percent had performance result.