Author(s): Xiaoqi Zhang; Yufang Ni
Linked Author(s):
Keywords: No Keywords
Abstract: The prediction of downstream river water level is crucial for the decision-making of upstream reservoir systems and the control of navigation and ecology within the river. However, the modelling process of traditional hydraulic models requires a variety of data types, including the parameters of river section and riverbed morphology, the water level and flow information of hydrological stations so on. The purpose of this study is to build a river water level prediction model with the help of the Gate Recurrent Unit (GRU) to capture the correlation on water levels and flows between different stations. Taking the hydrological stations in the middle and lower reaches of the Yangtze River in China as the research area, the historical days water level process of the target hydrological station and the historical days flow process of Yichang station are selected as the input conditions for the model. Results show that (1) the water level prediction model for the Shashi station, Hankou station, Jiujiang station and Datong station have respectively established; (2) The robustness of the constructed river hydrological forecasting model has been verified; and (3) The root mean square error is within the range of [0.07,0. 08], the Nash efficiency coefficient is as high as 0.999, the average absolute error is within the range of [0.05,0. 06], the maximum absolute error does not exceed 0.35m, and the qualification rate is over 80%, therefore, the proposed model can accurately predict the water levels of downstream stations.
Year: 2024