TOWARDS A REAL-TIME TIDAL ANALYSIS AND PREDICTION

In the practice of tidal analysis and prediction, the number and kind of astronomical tidal components that are to be included in a tidal mo del depend on the length of available tidal record and the desired acc uracy of prediction. Since tidal frequencies, including shallow water constituents, are distributed unequally in a few narrow frequency band s, an inappropriate selection of tidal constituents to be included in the analysis and prediction may cause the normal equations to become i ll-conditioned, or even singular, and the prediction to become poor. T his investigation shows how to construct lumped tidal frequencies whic h better characterize ocean fides with diminishing length of observati onal series. Further, a sequential tidal analysis model is proposed an d an algorithm for its implementation is presented, which can rigorous ly update a tidal solution when the number of observations increases. The algorithm also brings in automatically additional tidal constituen ts without a large amount of computation work; the CPU time for this a nalysis is only about 4 percent of that for the conventional harmonic technique. The sequential algorithm for ocean tidal analysis and predi ction has a potential to be used in tide gauge stations for providing continuous up-to-date tidal prediction.