基于Copula-Monte Carlo法的水库下游洪水概率分布研究

Flood Probability Distribution Estimation under the Influence of Upstream Reservoir Regulation Based on Monte Carlo Method

利用Copula函数构造水库调洪控制时段洪量和区间洪峰的联合分布,通过调洪函数将水库入库洪量转换为下泄洪峰,随机模拟得到两分区洪峰流量,提出了推求水库下游洪水概率分布的Copula-Monte Carlo(Copula-MC)方法。应用清江流域隔河岩水库实例进行验证,推求下游高坝洲断面的洪水概率分布,并与忽略洪水相关性的独立MC法进行比较,结果表明,独立MC法计算的洪峰流量系统偏小,会低估高坝洲断面的防洪风险;Copula-MC法得到的100年一遇高坝洲断面设计洪水与天然情况下相比减小32.6%。Copula-MC法充分考虑了上游水库断面和区间洪水的空间相关性及地区组成的随机性,结果合理可行。

The joint distribution of controlling flood volume at reservoir site and peak discharge at interval basin is firstly constructed based on Copula function, then the reservoir inflowing flood volume is transformed into outflow peak discharge by using flood regulation function and the flood peak discharge at the two sub-basins is got by using stochastic simulation method, and finally the Copula-Monte Carlo （Copula-MC） method for flood probability distribution estimation is developed. The Qingjiang basin is chosen as the case study and the peak discharge probability distribution of Gaobazhou site is estimated under the influence of Geheyan Reservoir by using Copula-MC method, and the results are compared with Independence-MC method that ignores the correlation. The results demonstrate that the flood peaks of Independence-MC method are systematically smaller than those of Copula-MC method, which may lead to an increased risk in Gaobazhou site, and the 100-year design peak discharge estimated by Copula-MC method at Gaobazhou site has been cut by 32. 6%. The calculation results of proposed Copula-MC method are reasonable because it can capture the spatial correlation of floods occurred at each sub-basin and reflects the stochastic characteristics of flood region composition.