A meta-heuristic approach for multivariate design flood quantile estimation incorporating historical information

Abstract

Abstract Design flood quantiles are crucial for hydraulic structures design, water resources planning and management, whereas previous multivariate hydrological quantile estimation methods usually do not consider historical flood information. To overcome such limitations, a meta-heuristic inference function for margins (MHIFM) approach, coupling meta-heuristic algorithm with a modified inference function for margins (IFM) method, is developed for modeling the joint distributions of flood peak and volumes with incorporation of historical flood information. Then, the most likely realization (MLR) and equivalent frequency combination (EFC) methods are employed for selecting multivariate design floods on a quantile iso-surface. The Danjiangkou reservoir located in Hanjiang River basin, the first pilot basin of most regulated water resources management policy in China, is selected as a case study. Application results indicate that the MHIFM approach shows good performance for estimating the parameters of marginal and joint distributions; moreover, the MLR method yields safer design flood quantiles than the EFC method in terms of highest routed reservoir water levels. The proposed MHIFM approach associated with the MLR method is safer and more rational for reservoir design, which would provide rich information as the reference for flood risk assessment, reservoir operation and management.