Uncertainty evaluation in univariate and bivariate flood frequency analyses using heterogeneous distribution
Investigating univariate and bivariate flood frequency is of vital value for hydrologic engineering design and flood risk management.However, the associated large uncertainties frequently misinterpret univariate and bivariate return periods, and further induce incoherent inferences in practical applications.Furthermore, the insufficient consideration of heterogeneity of flood series triggered by changing environment exacerbates the incoherence.Hence, designing corresponding frequency analysis frameworks coupling heterogeneous distribution in the context of sampling uncertainty constitutes the main objective of this study.The Weihe River Basin (WRB), China, is selected as the study area.The results indicate that (1) heterogeneous distribution exhibits better fitting performance and draws more credible design flood values at different return periods than homogeneous distribution, while more number of components (>2) of heterogeneous distribution induces the less robust and accurate fitting performance;(2) Further, although great uncertainty due to the limited sample size lies in design events selection from bivariate p-level quantile, Copula coupling heterogeneous distribution infers more theoretically coherent and reliable design events than coupling homogeneous distribution.
Flood frequency analysis Heterogeneous distribution Copula function Uncertainty estimation Weihe River Basin
Aijun Guo Jianxia Chang Yimin Wang
State Key Laboratory of Eco-hydraulics in Northwest Arid Region, Xian University of Technology, No.5 South Jinhua Street, Xian 710048, P.R.China
国际会议
法国巴黎
英文
47-54
2019-03-13（万方平台首次上网日期，不代表论文的发表时间）