NUMERICAL AND OBSERVATIONAL STUDIES FOR HIGHLY ACCURATE ANALYSIS OF DIRECTIONAL SPECTRUM

Abstract

 An ultrasonic Doppler type directional Wave Meter (DWM) has been used for directional wave observations since 1990's in Japan. In those days, due to poor computer power, directional spectra were estimated by using four time series data of water surface elevation and oblique components of three water particle velocities measured at an upper layer in water. Recently, we carried out 31 components measurements of water surface elevation and 30 oblique components of water particle velocities at 10 layers. In this case, since hundreds of simultaneous integral equations can be used for estimating directional spectra, estimation accuracy of directional spectra can be certainly evaluated quantitatively by using various combinations of the hundreds of equations. In this study, first, we examined the estimation accuracy through numerical simulations for DWM with Bayesian Directional spectrum estimation Method (BDM), Extended Maximum Entropy Method (EMEP), and Extended Maximum Likelihood Method (EMLM). Then the results were verified with the filed data observed by DWM. Various combinations of simultaneous integral equations observed by DWM were solved by BDM, EMEP, and EMLM. According to the results of the filed observations, the estimates by BDM and EMEP using information of water surface elevation and water particle velocities in all the layers, compared with EMLM, yield directional spectra with sharp and narrower directional spreading, and thus an accuracy improvement of the directional spectrum estimation can be expected by BDM and EMEP. These results are agree well with the results of the numerical simulations, where the estimates by BDM, EMEP, and EMLM were compared with the true directional spectra.