Tidal Current Predictions Using Rotary Empirical Orthogonal Functions

Abstract

In the conventional point tidal analysis approach, a set of tidal harmonic constituents is derived from each time series of currents. These sets of tidal constituents are then used to predict the tidal currents. For a large database of current time series, either generated theoretically (e.g., from numerical modeling) or collected experimentally (e.g., by remote sensing) the resultant database of harmonic constituents can be prohibitively large. The procedure of tidal prediction becomes time consuming and tedious. In this paper, an efficient and fast way to predict the tidal currents simultaneously at many locations is developed. The surface current data collected using ocean surface current radar in Poole Bay has been analyzed using the technique of rotary empirical orthogonal functions (EOF) in the time domain. It is found that the first EOF mode accounts for 97% of the total variance, while the first two EOF modes together account for 98%. The time evolutions of the first two EOF modes are mainly semidiurnal. Using the tidal harmonic constituents for the time evolutions of these two EOF modes, the dominant tidal signals, namely, M2 and S2, in Poole Bay are well reproduced. The spatial variation of the current ellipse characteristics (namely, the length of the semimajor axis, eccentricity. and angle of inclination) and Greenwich phase angles of M2 and S2 can be expressed in terms of the spatial variation of the flows in the first two EOF modes. The tidal currents can he predicted with an accuracy of better than 90%. EOF method reduces the tidal current prediction database from the number of current time series to two sets of tidal harmonic constituents.