Development of a Bayesian-copula-based Frequency Analysis Method for Hydrological Risk Assessment – the Naryn River in Central Asia

Y. R. Liu,Y. P. Li,Y. Ma,Q. M. Jia,Y. Y. Su

DOI: https://doi.org/10.1016/j.jhydrol.2019.124349

IF: 6.4

2020-01-01

Journal of Hydrology

Abstract:Copula functions are widely used in multivariate hydrological risk assessment. Incorporation of uncertainty analysis within copula functions can improve the degree of reliability to risk inference. In this study, a Bayesiancopula-based frequency analysis (BCFA) method is developed for supporting multivariate hydrological risk assessment under uncertainty, through integrating copula functions and Markov Chain Monte Carlo (MCMC) into a general framework. BCFA is not only capable of identifying multivariate characteristics of hydrological extreme events, but also useful for assessing multiple uncertain parameters in their dependence structure. BCFA is applied to a real case of flood risk assessment in the Naryn River, a typical high-mountain watershed with high frequency snowmelt flood in Central Asia. Major findings can be summarized: (i) the uncertainty range of joint distribution is mainly influenced by the value of flood variable and the selection of copula function, where the uncertainty range would decreases along with flood variable value; (ii) in a same flood event, the probability of both flood duration and flood volume reaching extreme large values is highest, and the probability of either flood peak or flood duration exceeding a certain threshold is highest; (iii) high peak flows (large than 1828.4 m(3)/s) are usually accompanied with flood durations less than 24.4 day or flood volumes less than 17.6 x 10(3)m(3).d/s. The findings are helpful for providing decision support of flood management and disaster prevention.