Statistical models and distributions of current velocities with application to the prediction of extreme events

Abstract

A multiple input system cross-spectral analysis method for ocean current velocities, involving the major meteorological and hydrographical forcing terms, is described and analyzed. The model uses joint observed data sets of forcing and current velocities to calculate initially unknown transfer functions that correlate each forcing term to the current velocity in transform space. From further forcing data sets, spectral estimates of the current velocity can be calculated with a higher level of significance than from a shorter, observed, series. The method is applied to data from the Öresund, between Sweden and Denmark, and the Gullmar Fjord at the Swedish west coast. The results show that this method is suitable to estimate the statistical parameters of a current in a specific area, using the transfer functions, if the forcing is known for a long time and the current for a briefer period. It has advantages over purely statistical models, as it accounts for the coupling between the statistical parameters of the current and the forcing. It also has advantages over complex hydrodynamic model, as it is simple to use and interpret. The paper also contains a brief analysis of statistical distributions of current velocities sampled with various instruments in different hydrographic regimes. It is concluded that currents can be regarded as normally distributed when dominating tidal components have been removed by filtration.