ESTIMATION OF WAVE DIRECTIONAL SPECTRA FROM PITCH-ROLL BUOY DATA

Estimates of the directional wave spectrum are generally extracted fro m data obtained by a heave-pitch roll buoy. Many different special est imators are now available, and there is a question as to which one is the more reliable in different environmental conditions. Routine data analysis and wave-spectra calculation require a computationally effici ent and robust estimator. Five different directional spectrum estimate s are compared for both synthetic and field data. Synthetic data consi sts of reconstructed cross spectral density matrices with and without noise, and the field data come from the Atlantic Ocean via the Surface Wave Dynamics Experiment (SWADE). Directional spreading and higher-or der statistics of the wave field arc compared for each technique. Test results indicate that the maximum likelihood method (MLM) estimator, although not considered an ''optimal'' estimator, gives the most consi stent and predictable results, and should therefore be favored over mo re traditionally used direct Fourier transform methods. Other spectral estimators, such as the maximum entropy method, iterative MLM, and ei gen vector MLM, outperform the MLM in many cases, but have unwanted ch aracteristics. such its wavenumber dependence, high sensitivity to noi se, and numerical instability. Results outline the fact that none of t he directional spectra estimators performs very well in every case and that care should be taken in data interpretation for engineering appl ications.