Using Long short-term memory neural network to forecast water level at the Quang Phuc and the Cua Cam stations in Hai Phong, Vietnam

Abstract

    In this article, the LSTM (Long Short-Term Memory) model is applied to predict the river water level without utilization of rainfall - forecast information and terrain data. The data required for simulation are hourly water levels at hydrological stations in Hai Phong city. The model was formulated to predict water level at the Quang Phuc station and the Cua Cam station, in Hai Phong city for many cases from 1 to 5 hours of lead time. Although the model does not require many input data such as climate, geography, land-use for rainfall-runoff simulation, the prediction results are very stable and reliable: the Nash - Sutcliffe efficiency (NSE) is higher than 97.8% and the root mean square error (RMSE) is lower than 0.10m for 3 hours of lead time prediction. The result illustrated that the LSTM model is able to produce the river water level time series and useful for the practical flood forecasting