Non-Stationary Multivariate Extreme Value Analysis in the Context of Coastal Applications

Coastal communities along the Chinese mainland are increasingly exposed to compound extremes in which intense precipitation coincides with energetic sea states. Traditional stationary and univariate design methods are inadequate because they neglect non-stationary marginal behavior and evolving dependence between drivers. This thesis applies a transformed-stationary extreme-value analysis (tsEVA) coupled with a time-varying copula model (tsEVA 2.0) to quantify joint, non-stationary extremes of daily precipitation and daily maximum significant wave height (SWH) at nine coastal locations (L1-L9) over 1961-2022. Following the tsEVA methodology, each input series is normalized to stabilize mean and variance, then modelled in the tails using a peaks-over-threshold Generalized Pareto Distribution with percentile-based thresholds. Fitted models are back-transformed to obtain time-varying marginal extremes. Precipitation and SWH peaks are paired by time proximity, mapped into probability space, and their dependence is represented by a Gumbel copula with parameters evolving in 30-year moving windows. Comparisons between the beginning and end of the time series are evaluated using AND-type bivariate 10- and 50-year return-period isoline. Results showed temporal and spatial variabilities in coupling intensity. Dependency strengthens at L2, L5, L7, and L9, where joint isolines indicate higher compound exceedance frequencies at fixed nominal return periods. L1 shows stationary dependence, whereas L3, L4, L6, and L8 exhibit weakening coupling and reduced co-occurrence for given marginal thresholds. Among all sites, L5 provides the most compelling demonstration of the tsEVA 2.0 framework. The study demonstrated that time-varying dependence can substantially alter compound risk even when marginal trends are weak. The tsEVA2.0 framework provides reproducible, site-specific joint design information to support coastal defense and climate adaptation along the Chinese coast.