海平面上升下沿海极值海水位非平稳性特征演变分析

杨思茹; 杨洋央; 方佳毅; 莫芷慧; 朱思颖; 张峰; 胡潭高

doi:10.13928/j.cnki.wrahe.2026.02.007

水利水电技术（中英文） ›› 2026, Vol. 57 ›› Issue (2) : 95 -106. DOI: 10.13928/j.cnki.wrahe.2026.02.007

复杂灾害链与水旱巨灾风险评估专栏

海平面上升下沿海极值海水位非平稳性特征演变分析

杨思茹 ¹ ,
杨洋央 ¹ ,
方佳毅 ¹^,² ,
莫芷慧 ¹ ,
朱思颖 ¹ ,
张峰 ³^,⁴ ,
胡潭高 ¹^,²

作者信息 +

Non-stationary evolutionary characteristic analysis of coastal extreme water levels under sea level rise

Siru YANG ¹ ,
Yangyang YANG ¹ ,
Jiayi FANG ¹^,² ,
Zhihui MO ¹ ,
Siying ZHU ¹ ,
Feng ZHANG ³^,⁴ ,
Tangao HU ¹^,²

Author information +

文章历史 +

PDF (1530K)

摘要

【目的】全球海平面上升破坏了沿海观测水位的平稳性假设，导致沿海极值水位重现期和重现水平发生巨变。因此，需要在非平稳性假设下分析海平面上升对极值水位的影响，为沿海地区海岸防护工程设计提供科学参考。【方法】以浙江温台沿海的验潮站为例，通过趋势分析，检验验证温州、瑞安、坎门三个验潮站观测数据的非平稳性；构建广义极值分布(GEV)和广义帕累托分布(GPD)非平稳模型，以时间协变量表征参数时变特征，利用贝叶斯框架下的马尔可夫链蒙特卡罗(MCMC)方法估计参数后验分布；通过赤池信息量准则(AIC)和贝叶斯信息量准则(BIC)评估模型性能，并结合有效重现水平(ERL)和预期等待时间(EWT)分析非平稳条件下的致灾风险。【结果】结果表明：浙江沿海的三个验潮站数据具有明显的趋势性，因此选用非平稳极值理论模型更为合理；GPD模型在拟合优度上优于GEV模型；非平稳条件下，百年一遇极值水位重现期显著缩短。【结论】GPD模型因对尾部高频极值的敏感性，更适合非平稳条件下的风险预测。模型阈值的选取会影响极值水位的预估结果。在海平面上升条件下，沿海地区面临的洪水灾害风险的频率和强度将显著增大，建议动态调整沿海工程防洪标准。研究成果为非平稳极值理论在气候变化的应用以及海防工程设计提供参考。

Abstract

[Objective] The rising global sea level has undermined the stationary assumption in coastal water level observations, [Results] ing in significant alterations in the return period and return level of extreme water levels in coastal regions. Hence, it is imperative to assess the impact of sea level rise on extreme water levels under a non-stationary framework, offering scientific insights for coastal protection engineering design in these areas. [Methods] Tidal observation from the tide gauges in Wenzhou and Taizhou in Zhejiang Province was analyzed. Non-stationary assumption was validated via trend analysis. Non-stationary Generalized Extreme Value(GEV) and Generalized Pareto Distribution(GPD) models were constructed with time-dependent parameters, and Bayesian Markov Chain Monte Carlo(MCMC) sampling was applied for posterior parameter estimation. Model performance was evaluated using Akaike(AIC) and Bayesian(BIC) information criteria, while risk dynamics were quantified through effective return levels(ERL) and expected waiting time(EWT). [Results] Distinct trends were detected in all three stations, making the adoption of non-stationary extreme value theory models more appropriate. The GPD model outperformed GEV. Under non-stationary, the return period of the 100-year extreme water level is projected to reduce dramatically. [Conclusion] The GPD model is more suitable for non-stationary risk prediction due to its sensitivity to high-frequency extremes. The estimation result of extreme water levels are impacted by the selection of model thresholds. With the rise in sea level, the frequency and intensity of flood disaster risks in coastal areas will significantly increase, necessitating dynamic updates of coastal defense standards. The findings advances the method ological application of non-stationary extreme value theory in climate change adaptation and serves as a reference for coastal defense engineering design.