Author(s): Yin-Sung Hsu; Hsiang-Chun Chou; Wan-Li Liu; Hsi-Chin Hsieh; Wen-Cheng Chen; I-Min Chang
Linked Author(s):
Keywords: Rating curve; Neural-Network model; Self organizing Map; Backpropagation network
Abstract: The stage-discharge rating curves are graphs depicting the relation between stage and flow rate. Since the discharge measurements are not automatic and continuous, the rating curves that provide a convenient way to estimate discharge are fundamental to water resource engineering. However, there are many variables in the natural river. The relations of stage and discharge are irregular and complicated, and need modifying by many methods. In this study, Artificial Neural Networks (ANNs) including Self-Organizing Map (SOM) and Backpropagation Network (BPN) are applied to establish a stage-discharge relation for the Zhuo-Hsui River. SOM is used in a preliminary process that classifies the measurement data into several clusters. According to each cluster, the relation to stage, width of water surface, submerged flow area, and discharge is established by the learning method of BPN. Based on the comparison of the simulated and measured results, it is shown that the ANN approach is much closer to the observed values than conventional techniques.
Year: 2004