What is a drought-to-flood transition? Pitfalls and recommendations for defining consecutive hydrological extreme events

Research into rapid transitions between hydrological drought and flood is growing in popularity, in part due to media-reported catastrophic impacts from recent events. Droughts and floods are typically studied as events that are independent from one another. Thus, a clear definition and assessment of the methods used to define consecutive drought-to-flood transition events does not yet exist. Here, we use a series of eight catchments that have experienced real-world impacts from drought-to-flood transitions as case studies. We assess the suitability of and differences between event selection methods applied to observational data. We demonstrate that different threshold level methods can alter the characteristics of selected events. The number and timing of transitions differs substantially between threshold level approaches in highly seasonal regimes as opposed to those with a weaker seasonality. The time period used to define the maximum interval between drought and flood also influences whether transitions are detected. We show that the probability of a transition occurring within a set time window could vary substantially between different methodologies and catchments. We also show that previously applied methodologies would likely fail to detect transition events that have been broadly impactful in the historical record. For the eight case study events taken from media, governmental and scientific reports, only three of the transitions were successfully detected. We qualitatively assessed the streamflow time series of the case study catchments, and outline a number of potential pitfalls in the event detection process. Finally, we make recommendations regarding methodological choices in the context of potential impacts of interest, and outline some priorities for future methodological development and research.