Archimedean copula-based bivariate flood-frequency analysis on Sukkur, Pakistan

Abstract

Floods are disastrous events in nature. In order to minimize the loss caused by these inevitable events, there is a need to estimate the risk of their occurrences. Several attempts have been made to study flood data in different parts of the world. In this paper, bivariate analysis of flood frequency using Archimedean copula is carried out on Sukkur Barrage, Indus River, Pakistan. The important characteristics of floods, namely peak flow and volume, are mutually dependent on each other. Copula-based technique to capture scale-free dependence structure is the most common technique meant for multivariate analysis of flood data. It is easy to construct and is a powerful technique for choice of marginal distributions. There are many types of Archimedean copulas selected for Archimedean copulas which have been tested on peak-flow and volume in the inflow data of Sukkur barrage, and in the light of graphical and statistical tests, Gumbel-Hougaard copula has been found to be the most appropriate one for the data under consideration. The Gumbel-Hougaard copula is used for obtaining primary and conditional return periods of flood characteristics. It also gives the Kendall return period for the multivariate analysis, which can be useful for risk based design of water resources project.