Extreme storm surges along China’s southeastern coast exhibit nonstationary behavior under climate change. These variations, including long-term trends, accelerate exceedance of critical flood thresholds and may cause larger impacts than projected. Understanding their timing and magnitude is essential for coastal adaptation. However, insight into this region’s extreme surge variability has been limited by scarce long-term tide gauge records. This study overcame this limitation by reconstructing hourly storm surges from 1979 to 2019 using a numerical model. Temporal patterns and magnitudes of extreme surge changes were identified. A novel nonstationary statistical model, integrating a nonstationary Generalized Extreme Value (GEV) distribution with a state-space approach, was developed to analyze surge evolution. Comparison of 100-year return levels under stationary and nonstationary assumptions reveals strong regional differences in how extreme surges are changing. The study refines nonstationary statistical methods for extreme surges and provides critical information for developing climate-resilient coastal adaptation strategies in southeastern China.