基于Copula函数的多站洪水过程随机模拟研究

Stochastic Simulation Study of the Flood Process Based on Multi-dimensional Copula Function

洪水过程随机模拟能够生成大量与历史洪水统计参数相似且形状各异的洪水过程,而水库群联合防洪调度需要结合流域内各水文站洪水过程统筹洪水资源的调度,实现洪水资源的科学调度,亟需发展多站洪水过程随机模拟方法。鉴于此,提出一种基于Copula函数的多站洪水过程随机模拟方法,该方法先利用二维Copula函数完成主站洪水过程的模拟,然后利用条件混合法构建三维Copula函数,依次模拟其余站的洪水过程。以长江中上游8处水文站为研究对象的结果表明,各站模拟洪水与历史洪水统计参数的相对误差较小,除了北碚和武隆站的最大值外,各水文站统计参数的相对误差在5%左右,能够反映历史洪水过程的统计特征,能够为水利工程防洪规划设计、水库群联合调度及防洪调度风险分析等工作提供支撑作用。

The stochastic flood process model can generate a large number of flood processes with shapes varying but statistically similar to historical flood data.The combined flood control dispatching of reservoir groups requires coordinated flood resource scheduling based on flood processes from various hydrological stations in the basin,aiming to achieve scientific allocation of flood resources.Therefore,there is an urgent need to develop a multi-station stochastic flood process simulation method.In view of this,a multi-station stochastic flood process simulation method based on the Copula function was proposed.This method first used a two-dimensional Copula function to simulate the flood process of the main station,and then used a conditional mixing method to construct a three-dimensional Copula function to sequentially simulate the flood processes of the remaining stations.The results of studying eight hydrological stations in the middle and upper reaches of the Yangtze River indicate that the relative error between the simulated flood data and historical flood statistical parameters is small.Except for the maximum values at Beibei and Wulong stations,the relative errors of the statistical parameters at the hydrological stations are about 5%.This reflects the statistical characteristics of the historical flood process and can support work related to flood control planning and design for hydraulic projects,combined dispatching of reservoir groups,and risk analysis of flood control dispatching.