Abstract
Tidal currents are one of the major attributes of flows in the sea. This paper describes the maximum likelihood method for the harmonic analysis of tidal currents. Assuming that measurements of tidal currents contain errors that follow the multivariate normal distribution, it can be explored the way to selecthe dominant components of tidal currents. Tidal currents were synthesized with harmoniconstants by numerical simulation, and AIC and BIC were employed to select the dominant components. As a result, BIC was a more reliable criterion than AIC when using many sample data. In addition, this method was applied to surface currents in Tokyo bay.
