The role of antecedent conditions in the generation of large floods using long continuous simulations

Continuous simulation has proven to be a promising base for flood frequency analysis since it avoids some of the shortcomings of other methods, such as assumptions about antecedent conditions or omission of relevant processes. In the EXAR project (hazard information for extreme flood events on the rivers Aare and Rhine), we have elaborated long continuous simulations of streamflow using a hydrometeorological modeling chain. This chain consists of a stochastic weather generator that provides precipitation and temperature series to a hydrological model, whose outputs are finally processed with a hydrological routing system, including and emulating the effect of regulated lakes, bank overflow and floodplain retention. As a result, distinctively long (several 100’000 years) continuous simulations are available at hourly resolution.

These simulations do not only cover streamflow but also other model internal fluxes and such as snow pack and soil moisture storage. With that, they allow to infer which hydro-meteorological story lines lead to extreme floods, i.e. floods with return periods of 100 years and higher. The story lines cover the important aspects of antecedent conditions, triggering precipitation, and their spatial and temporal interaction from the sub-catchment scale up to the large basin scale. The resulting story lines already cover a very broad range of possibilities due to the recombination of observations by the stochastic weather generator and the continuous simulation. They may help to further develop targeted story lines beyond what we already observed in changing climatic conditions.