Tidal harmonic analysis and prediction with least-squares estimation and inaction method

Abstract

Harmonic analysis has long been an important tool for tidal level analysis and prediction, in which Least Square Estimation (LSE) plays a major role. However, the importance of this role may now be challenged by the Inaction method (IM), which is based on normal time-frequency transform (NTFT), because this provides a new way to analyze and predict the tidal constituents by extracting the tidal constituents from the tidal level directly and obtaining the time-varying harmonic constants simultaneously. In this study IM is combined with LSE (LSEIM) to predict long-term tidal levels. Nineteen tide gauge records with long-term (more than 6 years) and continuous records were chosen for prediction tests. The results showed that, at eight tide gauges, the tidal level predictions using the LSEIM were improved in accuracy by 9–15 mm with a lead time of two months, with respect to those using purely LSE. These findings clearly suggest that the LSEIM can improve tidal level prediction accuracy with a lead time of two months.