Design flood estimation of cascade reservoirs based on vine-copula flood regional composition

The study region is the Yalong River basin, the largest tributary of the Jinsha River in China. The most likely flood regional composition (MLFRC) method has been widely used to estimate the design flood of cascade reservoirs in the operation period, in which the Gumbel copula or t-copula is often selected to fit the theoretical joint distribution. However, the Gumbel copula is only suitable for 2 or 3 cascade reservoirs, while the t-copula has great uncertainty in parameter estimation. In this study, we introduce vine-copula into MLFRC framework to estimate design flood of cascade reservoirs, and comprehensively compare the universality of different copulas through statistical experiments and a case study on the cascade reservoirs in Yalong River basin. The results indicated that: (1) Vine-copula significantly outperforms Gumbel copula and t-copula in terms of fitting the theoretical joint distribution. (2) The vine-copula based MLFRC method can successfully solve the different flood regional composition schemes most accurately in the simulated scenarios. (3) Compared with the design floods in construction period, the design floods in operation period in the Yalong River basin decrease significantly due to the regulation of upstream reservoirs. The 1000-year design flood at basin outlet section decrease by about 40 %. • Introduce vine-copula for design flood estimation of cascade reservoirs. • Compare different copula-based flood regional composition methods by statistical experiment. • Vine-copula based MLFRC method performs best with the smallest limitation in practice. • Increase hydropower generation with design floods in operation period.