Assessing the multivariant effect on floods using the coupled SWAT -Copula model

There are several variables which are the triggering factor for floods, among them precipitation, soil moisture and snowmelt is a critical factor which plays a crucial role in snow covered mountain regions. This study uses Vine Copula to determine the dependency structure of the joint variable distribution between precipitation, soil moisture and snowmelt in Beas River basin. In this study, the SWAT model is coupled with the Vine Copula model to conduct multivariate analysis for flood using the sub-basins parameters. For this purpose, the 45-year data which is generated from SWAT model were used. The Vine copula technique approach requires the marginal distributions for each variable and different copula function that combines the marginal data in a tree structure to generate a joint distribution. Considering the range of the variables eight univariant marginal functions were chosen. Once marginal distribution is determined, 18 list of copulas were used to analyse the correlation of variables in pairwise. Later tree sequence of R-, D- and C-vine copulas were analysed in the study. Finally, according to the structure and nature of the data, R-vine copula was selected as the best copula and the relevant tree sequence was later used. Kendall’s tau test was used to check the correlation of the variables in pairwise and showed good correlation. This study proves to be an effective approach in improvising the flood prediction and control of flood risks.