Flood Typing and Mixed Population Study

Abstract

Flood frequency analysis is critical for managing flood risk, designing hydraulic structures, and developing floodplain management strategies. Accurately estimating the flood peak of varying associated risks can significantly reduce potential hazards to water infrastructure and human life. Despite long-standing recognition of the diverse nature of floods caused by different hydrological and meteorological processes, traditional flood frequency analysis methods often assume a homogeneous population of flood events. This study investigates five flood frequency approaches suitable for mixed flood distribution populations and evaluates their applicability to rivers experiencing floods from multiple hydrological processes. We aim to address three key research questions: (a) when application of a mixed frequency model is needed, (b) how an expert-based flood classification compares with a purely data-driven flood classification technique, and (c) which mixed flood frequency model can best estimate flood quantiles for varying return periods. First, we systematically conduct synthetic experiments to explore the application of the considered models, followed by their application to selected natural basins in the USA. The implications of this study extend to various stakeholders, including engineers, planners, and policymakers. By adopting a mixed population flood frequency analysis, decision-makers can gain a deeper understanding of flood risk, leading to more informed and effective flood mitigation strategies. Future research directions include refining the classification of flood-generating mechanisms and exploring the application of mixed population models in different climatic and geographical contexts.