Abstract
Parallel implementation of an unstructured Simulating Waves Nearshore (SWAN) model with the Wave Model (WAM) cycle 4 formulation was used to evaluate the performance of a third-generation wave model over large spatial scales. Data from a network of National Data Buoy Center (NDBC) buoys and the Wave Current Information System (WAVCIS) stations were used to assess the skill of the input and output of the wave model. The simulation results reveal that the underestimation of energy in the low-frequency band (0.12-0.17 Hz) can be ameliorated if the model is calibrated using site specific in situ measurements instead of the Pierson-Moskowitz spectra. This process led to more than a 25% decrease in the root mean square error between simulated significant wave height and in situ observations. Use of the verified model for the Gulf of Mexico, with bed friction computed from grain-size distribution, as opposed to a default constant bed-friction formulation, showed that the wave height difference can exceed 1.5 m or 40% of local wave height for a large spatial extent during extreme events, such as Hurricane Dennis. In addition, with the use of eddy viscosity bed-friction formulation with usSEABED (U.S. Geological Survey), the sediment data results were in better agreement with the in situ observations during Hurricane Dennis, with less than a 4% increase in computational cost. The mean wave-height distribution over several cold fronts also demonstrates the influence of bed grain-size parameterization in wave transformation, especially in water depths shallower than 15 m, thereby demonstrating the significance of this study in advancing our understanding of sediment-transport modeling.