关键词: 融雪, 洪水, 联合设计, 峰量拟合, MCMC, Copula函数, 重现期

Abstract: Driven by multiple characteristic attributes, flood is inherently affected by multiple concurrent or continuous driving factors, which aggravates the unpredictability of flood occurrence. In many risk assessment and design applications, the multi-hazards of extreme situations and compound events are often ignored. In view of the design flaws of univariate floods, the peak volume marginal distribution was constructed using GEV (Generalized Extreme Value) distribution and GPD (Generalized Pareto Distribution). The joint probability distribution under extreme conditions was established using the Gaussian Copula function with parameters optimized by MCMC (Markov Chain Monte Carlo). Furthermore, the design flood process line was derived with 500-year-event flood as an example. The control basin of the Manas River was taken as the research area. Results demonstrate that the joint distribution fitting effect of the Gaussian Copula function with MCMC-optimized parameters is better than that of the correlation index method and the local optimization algorithm. The design value under mutual influence of the variables is correspondingly greater than the flood design value under the control of a single variable. The design growth rate of 90 h in the design flood process line is 24.19%. Therefore, the new flood control standard established by the joint distribution under parameter optimization provides a more scientific and reasonable basis for the safety design of flood control and disaster mitigation for the Manas River reservoir in flood season.

Key words: snowmelt, flood, joint design, peak fitting, MCMC, Copula function, recurrence period