Harmonic analysis of short-term tidal level prediction model for tidal reaches

The prediction for tidal level of the tidal river is affected by the upstream runoff and the tidal wave in the outer sea. The aim of this study is to overcome the harmonic constant of tidal level prediction cannot remain changed for tidal reaches. A new idea named harmonic analysis combined with the least square method, Fourier-based corrected interpolation and ridge regression analysis method was proposed. The key step of the idea is as follows: firstly, the new harmonic constant can be determined via harmonic analysis of water level station data over the corresponding period. Secondly, the Fourier-based correction method is used to interpolate the short-term tidal level data and make the data interval equal. Thirdly, the application of the ridge regression analysis method is used to deal with the multicollinearity problem which may occur in the short-term harmonic analysis. Besides, the selection rules of tidal data from data length, tide type and starting point were considered. Finally, based on an exponential function which can describe relationship between the new harmonic constant and the upstream runoff, the tidal level prediction model of Baimao station in the Yangtze River tidal reaches is constructed. The result shows that the average error of model in the water level was smaller than 0.20 m, and all the time points with a total water level error smaller than 0.25 m exceeded 69% of all points. The mean error in the tide time was smaller than 34 min, and the maximum error was smaller than 60 min. The experimental results conformed to the trend of the observed water level changes, which verified the feasibility of the established model.