Abstract
Two well-known third generation wave models, SWAN and WAVEWATCH-III, with different assumptions for high cut-off frequency were used to evaluate the interaction of low and high frequency components in wave spectral evolution. The results showed that WAM cycle 3 formulation overestimates the energy content in frequency band of 0.5–1 Hz for Gulf of Mexico, which suggests using cut-off frequency close to 0.5 Hz rather than 1 Hz would improve the simulated bulk wave parameters. The evaluation of WAM cycle 4 and a newer nonlinear formulation implemented recently for white capping in SWAN also showed the better performance of wave model in oceanic scale with cut-off frequency close to 0.5 Hz. However, WAM cycle 3 was more sensitive to cut-off frequency as well as to the exponent used in the expression for the frequency tail, than other formulations in SWAN. The use of f− 5tail shape, rather than the f− 4 form for the frequency spectrum beyond both cut-off frequencies used in this study, resulted in better agreement between simulated and observed wave parameters for most of the formulations implemented in these models. Also, it was demonstrated that WAM cycle 3 with dynamic cut-off frequency outperformed the corresponding configuration with static cut-off frequency. The suggested modifications for cut-off frequency and the expression for high frequency tail in SWAN substantially ameliorates the widely known underestimation of the average wave period associated with the WAM cycle 3 formulation, and reduces the amount of calculations needed for other formulations.