Hydrology of extraordinary floods - event analyses

• Applicant: Professor Dr. Andreas H. Schumann, since 9/2023
• Subject Area: Hydrogeology, Hydrology, Limnology, Urban Water Management, Water Chemistry, Integrated Water Resources Management
• Term: from 2017 to 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 278017089

In the last two decades several extreme river floods in Germany occurred that exceeded or were similar in magnitude to the largest floods since the beginning of systematic flood observations. The characteristics of the largest floods on record differ in a number of ways from their smaller counterparts leading to the question: “What are the main processes and process combinations of extraordinary floods?” To answer this, a methodology to analyse flood events was developed in the first phase based on analyses of hydrographs and their typing into flood categories. Several statistical methods were originated to consider the differences of flood generation processes. Based on this, a Machine Learning Approach will be developed to automatically differentiate flood types. It will be tested in several river basins to specify the main hydrological processes of flood generation, resulting in extreme floods. The results will be used in a uni- and multivariate statistical mixing model to compare the quantiles of the different flood types at one gauge and between gauges and to specify more precise critical values for combinations of flood characteristics.Based on the derived flood categories, a stochastic flood simulator was developed. This simulator is able to model type-specific flood peaks and volumes based solely on precipitation and runoff coefficients. Nevertheless, the probabilities of extreme floods are underestimated. Hence, a more detailed analysis of flood generation is planned, which has to be based on deterministic modelling to specify the relevance of initial and boundary conditions of extreme floods. For this, a distributed hydrological modelling approach will be developed to consider the spatial and temporal rainfall conditions as well as the non-linearity of hydrological processes and the overlay of flood waves. It will be applied for sensitivity analyses, based on flood events with different initial and boundary conditions (“if-then” analyses). The model will be validated by its ability to represent the frequencies of floods and their main characteristics. It will be used to generate extreme events, which result from rare coincidences of several factors.Extreme floods also define the upper tail of distribution functions. To consider them appropriately, the results of the deterministic model will be transferred into the stochastic model by a specific component. The result will be a stochastic event generator which considers the frequency of event types, their specific distributions of event characteristics and extreme events explicitly. More precisely, the deterministic part will correct the right tail of the stochastic part. The developed methodology will be applied to gauged and ungauged basins to test its ability for regionalization. It will be used to integrate deterministic aspects in stochastic modelling and bridge the gap between events which are very seldom but most relevant for flood statistics.

• DFG Programme: Research Units
• Subproject of: FOR 2416: Space-Time Dynamics of Extreme Floods (SPATE)
• Ehemalige Antragstellerin: Dr. Svenja Fischer, until 8/2023