BIVARIATE FLOOD FREQUENCY ANALYSIS USING GUMBEL COPULA

Abstract

A copula based methodology is presented in this study for bivariate flood frequency analysis over a station over a Kelantan river basin located in Northeast Malaysia. The joint dependence structures of three flood characteristics, namely, peak flow, flood volume and flood duration were modelled using Gumble Copula. Various univariate distribution functions of flood variables were fitted with observed flood variables to find the best distributions (eg. generalized pareto, log-normal, exponential, gamma distribution, weibull, gumbel, cauchy). The results of study revealed that different variable fits with different distributions and the correlation analysis among variables showed a strong association. Cumulative joint distribution functions (CDF) of peakflow and volume, peakflow and duration and volume and duration revealed that return period of joint return periods are much higher.