Modelling with copulas has become an increasingly popular field in recent years, as copula models are used in finance, medicine, geodesy and, last but not least, hydrology. In this diploma thesis we first present the basics of the theory needed to understand the model itself. We then focus on the use of copula models in hydrology, and introduce various measures of dependence, graphical tools, and estimation methods that help us find the most appropriate copula model. The analysis from the sample data set is then used to show the relationship between the maximum annual flow and the corresponding volume of the Harricana River. From the data presented and analyzed through work, we have seen that several copula families provide satisfactory models for Harricana River data. Not surprisingly, most of these models are extreme value copulas. With the help of additional analyzes, we come to the conclusion that the most optimal model is Tawn's type 1 copula, which is an asymmetric copula. Although we limited ourselves to the bivariate case in our analyzes, most of the tools presented could be generalized to the multidimensional case. As the number of variables increases, the complexity of the models also increases, so the construction of appropriate copula models remains an open question.
|
