Storm-permitting climate modeling highlights storm frequency’s role in future extreme sea level changes along US East and Gulf coasts

Storm-induced coastal extreme sea levels (ESLs) pose severe threats to infrastructure, economies, and ecosystems. However, projecting future ESL changes is hindered by the coarse resolution of climate models used in assessment reports, which fail to accurately capture tropical cyclones (TCs) and nor’easters. Here, we demonstrate that high-resolution Community Earth System Model (CESM) simulations realistically reproduce observed wind- and pressure-induced daily-mean extreme dynamic sea levels (DSLs), including the most extreme events. Under a high-emission scenario, we show that 50-year return DSLs (DSL50) decrease along the U.S. Texas and Northeast coasts but increase along the U.S., Louisiana, and Southeast coasts from 2006 to 2100, creating substantial spatial discrepancies between total DSL50 changes and mean DSL rise. Along the Gulf and Southeast coasts, DSL50 trends are primarily driven by TC frequency changes, while nor’easter frequency changes tend to govern trends along the Northeast coast. These findings challenge the traditional assumption of stationarity in storm-induced ESLs, underscore the importance of high-resolution climate models for regional coastal risk assessment, and represent a critical step toward more accurate projections of future ESLs in a warming climate.