Abstract
A short-term flood prediction algorithm with a 2D dynamic wave model and a particle filter is proposed to consider the uncertainties of hydrologic input data and channel roughness. The particle filter makes it possible to utilize a non-linear and non-Gaussian model for estimating time variant channel roughness and inlet flow uncertainty by considering sequentially updated water stages. The proposed method was applied to the Katsura River located in Kyoto, Japan, and it was verified first through a synthetic experiment. The experiment result shows that the algorithm successfully traces the hidden true values, which are the correct inlet discharges and Manning’s roughness coefficient, on a real-time basis. The prediction results were also compared with observed water stages, and they showed good agreements with the observed water stage.
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call