This master thesis consists of two parts. In the fi rst part we deal with the estimation of the
distributions of annual univariate fl ood maxima, in order to be able to make statements
about a fl ood of the century. For this purpose, we use common estimation methods of
extreme value theory: the block maxima method, the Hill estimator as an application
of the maximum domain of attraction, and the peaks over threshold method. The focus
will be on the derivation of the diff erent statistical methods and their application to the
two measuring stations Munich and Landshut at the Isar; the underlying theory will be
summarized at the beginning without detailed proofs.
In the second part, we will consider accumulation risks and ask the question, with which
probability we will expect losses by a fl ood event at two diff erent locations at the same
time. To model the bivariate fl ood maxima, we will use extreme value copulas. Finally, we
would like to determine bivariate stress scenarios with the help of statistics for bivariate
block maxima, into which our results on the century fl ood events from the fi rst part will
be incorporated. In doing so, we will also use Hofkirchen, a measuring station at the
Danube.
«
This master thesis consists of two parts. In the fi rst part we deal with the estimation of the
distributions of annual univariate fl ood maxima, in order to be able to make statements
about a fl ood of the century. For this purpose, we use common estimation methods of
extreme value theory: the block maxima method, the Hill estimator as an application
of the maximum domain of attraction, and the peaks over threshold method. The focus
will be on the derivation of the diff erent statistical m...
»