| Summary: | 碩士 === 國立成功大學 === 水利及海洋工程學系專班 === 94 === The harmonic analysis method needs a long-term observation data for calculation. If the observed data are missing quite a lot due to some reasons, then this method cannot estimate astronomical tide height so well. Thus the purpose of this paper is to obtain accurate astronomical tide height by using astronomical tide to retrieve the missing data. It is shown that the supplement is reasonable agreement under the condition at the data loss being below 20% in a year or 60% for more than two years. If the data loss ratio is above 20% of the data record length lies within one to two year, the supplement results cannot converge to the real of the amplitude among the tide component of S2、K1 and Sa. The reason is that the amplitude of tide components called S2 with R2 and T2 ； K1 with Ψ1 and S1 ； Sa with Ssa are mutually affected each other. Consider two component tides with close by angle frequency to constitute the resultant tide, it is necessary to make the supplement with 2 years data length can acquire better result. The same results are done by the correlation method to use 2 years data length for the supplement.
The result of harmonic analysis can get the relation between energy and frequency, but unable to obtain distribution of energy in the time domain. The wavelet transform of the tidal data can get the distribution of energy in the time and frequency domains. From the wavelet energy distribution in the time domain, it can judge the time to happen for the maximum tide range difference. And from the mean of wavelet energy distribution in the frequency domain, it can decide the type of tide based on the energy proportion of diurnal and semidiurnal tides. Owing to the energy for frequency below 0.5(1/day) is very small compared to total energy; the retrieved data from the harmonic analysis have little influence to the change of water level caused by other meteorological factors. During the process of the tidal data supplement, it can judge the convergent condition and iteration frequency from the variation of the wavelet energy spectrum. The same results are similar to the correlation method.
|
|---|
