An accuracy-enhanced risk assessment framework for compound flood peak–volume effects using a mixed copula-probabilistic approach: a case study of the Yangtze River Basin, China

The severe impacts of watershed floods necessitate the elevation of standards in hydraulic structure engineering, where considering flood return periods are vital for safe and cost-effective designs. Traditional single-variable methods often overlook the simultaneous flood peak–volume effect, requiring a multi-variable approach for improved accuracy and effectiveness. This study introduces a mixed copula-based bivariate frequency method to examine the relationship between flood peaks and volumes using 72 years of data from the Hankou Hydrological Station. Our copula model, incorporating GEV, Gamma, Lognormal, and Normal distributions, captures the complex dependency between flood peaks and volumes, outperforming traditional univariate methods and standard copula models such as Clayton, Frank, and Gumbel. The mixed copula model reduces 2.5786% in d2 and 1.1072% in RMSE compared to the best-fitting single copula model, showing the capability of capturing complex tail dependence structures associated with extreme events. This method also improves the predictive accuracy of return period, increasing the 50-year and 100-year joint return periods by 8.45% and 10.95%, and the concurrent return periods by 61.79% and 91.88%, which can enhance safety and economic efficiency of hydraulic structures by making the tradeoff between failure risks and construction costs. These results demonstrates the robustness and reliability of our model to provide an accurate assessments of compound flood extremes and return periods to enable the design of resilient hydraulic structures.