Advancing High-Resolution Downscaling of Wind-Generated Ocean Waves Using WW3 on Unstructured Grids

Accurate modeling of wind-generated ocean waves is critical for understanding coastal processes, maritime operations, and coastal management. Recent advances in global wave forecasting have substantially improved large-scale predictions; however, bridging the gap between coarse-scale solutions and the finer-resolution requirements of nearshore environments remains an ongoing challenge. In this study, we present our latest developments in numerically downscaling wind-wave fields using the WAVEWATCH III (WW3) framework on unstructured grids, enabling more flexible resolution in complex coastal and shallow-water settings.

We detail a series of enhancements in WW3 aimed at improving both precision and computational efficiency. These include a new limiter implementation within the implicit scheme, GSE correction, and refined numerical integration of shallow water source terms and wave setup computations. In addition, we have optimized memory management and parallelization across the WW3 code base. By applying these techniques to a range of configurations, from simplified wind-wave growth scenarios to high-resolution global unstructured-grid models, we illustrate the upgraded performance and broad applicability of WW3, including its potential for more accurate wave climate assessments.

Lastly, we showcase a novel wave modeling framework based on a recent C++ language implementation of the unstructured solver. This approach leverages SIMD-based vectorization at the CPU level (0-level parallelism) in conjunction with domain decomposition and hybrid MPI+OpenMP parallelism, resulting in significant computational speed-ups. Such gains are especially valuable for long term runs of high resolution simulations, highlighting the framework’s suitability for future climate modeling efforts that demand high-resolution wave climatology over extended temporal scales.