Abstract
Accurate wave modeling is crucial for coastal management, navigation, and marine safety, particularly in complex coastal environments like Monterey Bay. Here, we investigated the performance of a global wave model, specifically the third-generation WAVEWATCH III model, downscaled to Monterey Bay, California, over a twoyear period. We employed two different source term packages, namely ST4 and ST6, for wind input. Four distinct grids were generated, with three of them being regular grids and one being unstructured. A two-way nesting approach was applied for three grids, with resolutions in the latitude of 0.5 degrees, 0.2 degrees, and 0.05 degrees, respectively. The fourth grid is unstructured, with maximum and minimum resolutions of 2 km and 0.15 km, respectively. Boundary conditions for the unstructured grids were obtained from the two-way nesting grids. Additionally, the model was forced by CFSv2 wind data with resolutions of 0.2 degrees. This study focuses on the highest-resolution model, which utilizes an unstructured grid. Significant wave heights were validated against data from five NDBC buoys, six CDIP buoys, 22 CLASI buoy locations, eight spotter buoys, and altimeter data. Across all 41 buoy locations and altimeter data, the model exhibits excellent agreement with the measurements in terms of statistical properties. Furthermore, we observed that ST4 outperformed ST6 in terms of scatter index and Pearson's correlation coefficient, while ST6 exhibited lower RMSE and bias. Regarding computational time, it was found that ST4 runs 25 % slower than ST6. In addition to significant wave height, wind sea, and swell were also compared based on one-dimensional wave spectra. Eleven buoys were used to validate the swell, with both ST4 and ST6 showing similar statistical performance for wind sea while ST6 should be used in swell conditions because it runs faster and gives better results.