Disentangling the Mechanisms Linking Extreme Precipitation Types to Extraordinary Floods: An Assessment in Taiwan

The frequency of extraordinary floods (EF) has risen globally in recent years, often accompanied by substantial economic losses and fatalities. However, these changes are not uniform and exhibit pronounced spatial and temporal variability. Extreme precipitation (EP) is considered one of the key factors triggering EF and is associated with various weather systems. However, studies examining how different types of EP contribute to EF, particularly in terms of the spatiotemporal variability of these relationships in response to climate change, remain limited.

This study proposes a framework to assess the impacts of different types of EP on the occurrence of EF. We use data from Taiwan as a case study, where multiple flood-associated types of EP occur, including tropical cyclones (TCs), mesoscale convective systems (MCSs), and frontal systems (FSs). To examine differences among EP-EF groups, a mechanism-based framework is developed to classify EP types and flood events. Meanwhile, weather types are identified using an unsupervised k-means clustering approach based on three groups of variables: precipitation, storm-related characteristics, and topographic controls. EF events are defined using a peak-over-threshold (POT) approach and are linked to their corresponding weather types. Our findings reveal varying temporal trends across different EP-EF groups, providing insights into how the spatiotemporal variability of extreme rainfall affects the occurrence of extraordinary floods.