Advancing the potential of the metastatistical extreme value framework for extreme flood estimation in German catchments

Abstract

River floods are among the most destructive natural disasters, with their risk projected to increase due to socioeconomic and climate changes, posing a growing global threat. Heavy-tailed behavior in flood distributions serves as a key indicator of extreme flood likelihood, emphasizing the need for accurate identification and prediction of such patterns. This dissertation advances the understanding of heavy-tailed flood distributions while introducing the Metastatistical Extreme Value (MEV) framework for predicting and managing the extreme flood events in practical applications. A statistical approach that accounts for different runoff-generation processes is applied and tested using daily streamflow time series from 182 streamflow gauges in Germany. The results provide new insights into heavy-tailed flood behavior and offer a robust method for predicting extreme floods. This method is less sensitive to limited data, and is applicable across diverse hydroclimatic conditions.