LAUSR

Abstract Multi-peaked directional wave spectra have been estimated by the well-known maximum likelihood method and modeled by applying a numerical optimization approach that uses a gradient OQNLP algorithm. The research is based on field measurement data from a 210-km-long area at the northern coasts of the Gulf of Oman. This modeling approach allows the fitting of various spectral models. Accordingly, three directional spectrum models with combinations of frequency and directional spreading functions have been evaluated and proper models for directional waves have been presented. On the other hand, the introduced optimization algorithm is modified by adding a constraint to zeroth-order spectral moment (m0) value to improve the fitting accuracy in the frequency spectrum. It has been shown that selecting the appropriate model for each of the frequency and directional functions can significantly improve the performance of both. Among the evaluated directional models, the accuracy of the results based on both JONSWAP Sech2 and Gamma Sech2 models was about 25% better in the frequency domain and about 10% better in the full directional domain than JONSWAP Cos2s model. Besides, in most of the observed swell components, because of their wide spectrums with low peak values, the use of the Gamma frequency model performs better than the JONSWAP model.