Understanding and simulating coastal extremes along the East Coast of the United States with global numerical models of different complexity

Sea level rise and non-climatic anthropogenic factors are increasingly exposing coastal populations worldwide to extreme events such as storm surges and flooding. Coastal extremes arise from mechanisms operating across multiple spatial and temporal scales, necessitating integrated modeling approaches. This thesis explores methodologies for modeling coastal extremes in a global framework, with primary focus on developing MUSE (Multiscale Unstructured model for Simulating the Earth ocean), a novel baroclinic global ocean model based on unstructured meshes. The unstructured-grid capability enables high coastal resolution while seamlessly integrating the large-scale circulation. The first implementation and assessment of MUSE exhibit skillful tidal sea level, demonstrating promising capabilities for short-term coastal extreme applications. Storm surge simulations along the United States East Coast (USEC) for hurricanes Dorian (2019) and Ian (2022) reveal that bottom friction formulation, atmospheric forcing resolution, and coastal mesh refinement critically influence surge prediction. Sensitivity to mesh resolution and bottom friction depends primarily on regional topography and dynamics rather than specific storm characteristics. Results identify land-sea-river coupling as the most critical future development for improving surge accuracy. Nevertheless, this work establishes foundational knowledge for future development of MUSE and demonstrates its potential as a complex multi-scale model for coastal extremes. Analysis extends to compound flooding events from precipitation and storm surge along the USEC, using coupled climate models at different atmospheric resolutions. Higher atmospheric resolution better captures extreme event statistics and seasonality, especially regarding precipitation. A persistent dipole pattern of sea level pressure anomaly drives onshore winds that elevate sea surface height and facilitates moisture transport toward the event location. A detection index based on this pattern provides encouraging results for identifying compound events. Comparison of MUSE and global models with different complexities, numerical methods and resolutions clarifies the respective strengths and limitations for coastal extreme modeling.