Tropical cyclone-driven storm surge and wave database for the US North Atlantic and Gulf coastlines

Historical datasets of tropical cyclone–driven storm surge and waves at moderate coastal resolution are scarce, limiting coastal hazard analysis and AI/ML-based surrogate model development where field observations remain sparse. We present a publicly available hindcast database of surge and wave conditions for 232 U.S. landfalling and impactful storms (1981–2021). We applied the coupled ADCIRC+SWAN system across the entire U.S. North Atlantic and Gulf coastline on a coastal-refined unstructured mesh achieving practical nearshore resolution ( ~100-500 m) to have a computational feasibility. We forced simulations with parametric wind fields from the Generalized Asymmetric Holland Model fitted to NOAA best-track data. For each event, we provide hourly NetCDF files containing water surface elevation, significant wave height, and peak wave period. Users can apply these fields as boundary conditions for higher-resolution local models, train ML model predictors, and conduct coastwide extreme-value analyses. We validated simulations against numerous NOAA tide gauges and NDBC buoys, demonstrating robust water level skill with documented wave biases. This comprehensive basin-scale database enables coastal flood hazard assessment across multiple decades of historical storm surge events.