Detiding three-dimensional velocity survey data in coastal waters

Abstract

A three-dimensional data interpolation technique is proposed that efficiently removes tidal current from spatial velocity surveys. The least squares method extends prior two-dimensional detiding methods to three spatial dimensions using biharmonic splines. Biharmonic splines are fitted to velocity data from acoustic Doppler current profiler (ADCP) surveys, moorings, and ocean surface current radar (OSCR). The data are used to predict diurnal and semidiurnal tidal currents on the inner shelf off New Jersey that vary between 1 and 15 cm s-1 at spatial scales of about 20 km. The (tidal) signal to (subtidal) noise is thus O(1) in the study area. Although the main task of this study is to remove tidal variance from the ADCP survey data, an attempt is made to accurately 'predict' tidal currents from the data. The latter task is more difficult. Both artificial data with known signal-to-noise properties and actual measurements indicate that the method estimates both diurnal and semidiurnal tidal currents to within about 3.5 cm s-1 rms, or 30% of the true tidal signals. While the biharmonic splines remove tidal currents successfully, the prediction of the vertical structure of tidal currents is only fair. Some experimentation guided by physical intuition and prior knowledge of the tidal fields is necessary in order to obtain an accurate and stable solution. While this ambiguity constitutes the main disadvantage of the method, its simple algebraic expression to predict tidal currents in space and time is its main advantage. Properly weighting velocity data from different sources, such as moorings, surface current radar, and ADCP surveys of different quality, improves the quality of the fit.