An analysis of change in flood indicators in the Elbe and Rhine River: implications for the selection of design discharges

---

Christoph Mudersbach, Jens Bender, Fabian Netzel

Wednesday 1 july 2015

11:30 - 11:45h at North America (level 0)

Themes: (T) Extreme events, natural variability and climate change, (ST) Hydrological extremes: floods and droughts

Parallel session: 9I. Extreme events - Flood Drought

---

Within this investigation, we focus on the time-dependent characteristics of common design discharge values of the gauges Neu Darchau (Elbe River) and Cologne (Rhine River). We follow the questions, if the discharge characteristics of the Elbe and Rhine River have changed over the last decades and how much the design discharges (i.e. HQ100) are affected by the latest extreme events (e.g. for the Elbe River in 2002, 2006, 2011, and 2013). Hence, we conduct (i) trend estimations for different flood indicators, (ii) trend and seasonality analyses of flood frequencies and (iii) an assessment of time-dependencies of flood frequencies by using probabilistic extreme value statistics with both block maxima and peak-over-threshold approaches. Time-dependent changes in extreme value statistics can be investigated using a non-stationary extreme value statistics approach. Here we use a quasi non-stationary extreme value approach in order to analyse the influence of single extreme events on the extreme value statistics. The quasi non-stationary approach is based on stationary extreme value distributions and a stepwise analysis of different time series lengths. This procedure reflects the real situation in statistical or engineering practice, where design values have to be steadily verified due to new data. A test method is introduced by which the significance of the temporal changes in design discharges can be assessed. This method is based on a stochastic time-series generator in conjunction with Monte-Carlo-simulations. Assuming that the natural discharge time-series follow an autoregressive process of 2nd order (AR2-model), a large ensemble of synthetic time-series can be built. By analysing the synthetic time-series confidence intervals and significance levels can be reliably estimated. Thus, the time-dependent changes in design discharges can be assessed either as a part of the natural variability of the underlying process or rather as inhomogeneities.