Flood generation processes – a tool for understanding hydrological changes and improving predictions

A wide variety of processes controls characteristics of river flood events. Classifying flood events by their causative processes may assist in understanding the emergence of extremes and support the detection and interpretation of their changes. We show observational evidences of considerable changes in the frequency of different flood generation processes in Europe in the past decades that are likely to manifest in the shifts in the dominant processes by the end of the century under high emission scenario. Furthermore, we show that we can use the information on different event generation processes for diagnosing limitations of conceptual hydrological models and deep learning-powered forecasting tools paving the way to their improvement. Our ongoing work on socio-economic impacts of floods generated by different processes indicates that their future shifts and the limitations of our state-of-the-art models might have dire consequences for the flood preparedness in Europe.