A multivariate design framework for river confluences

Abstract

Throughout the last decade copula functions were widely used to assess a wide range of hydrological problems, often focusing on two distinct variables. In many of these studies it was ignored whether the two variables of interest actually occurred simultaneously (e.g. two annual maximum time series were analysed in a multivariate statistical framework). Here we introduce a novel approach to derive bivariate design events using copula functions allowing both simultaneous and non-simultaneous occurrence of the variables to be modelled. The methodology is exemplarily applied to assess the combined flood occurrence at the confluence of the rivers Rhine and Sieg (Germany). The results underline the validity of the methodology. Employing a hydrodynamic numerical model furthermore shows that commonly used statistical approaches to select a single design event out of a vast number of possible combinations can be critical for practical design purposes.Editor Z.W. Kundzewicz; Associate editor S. Grimaldi