River flow amplification under climate change: attribution and climate-driven storylines of the winter 2023/24 UK floods

A warming climate is expected to alter the magnitude, frequency and spatial pattern of floods. The widespread flooding observed in the UK and Western Europe over the winter half-year 2023/24 followed on from a number of other notable floods in 2013/14, 2015/16 and 2019/20. However, detecting a climate-driven trend in river flows is complicated by the influence of internal variability and relatively short observational records. End-to-end probabilistic attribution that includes river flows also remains challenging as hydrological responses do not scale linearly with changes in rainfall. Recent studies have encouraged the routine creation of event storylines in a forensic manner to explore a full range of plausible outcomes and enhance risk awareness of UNSEEN outcomes. Few studies to date have attempted to harmonise the different approaches when conducting retrospective analysis of hydrological extremes.

A consistent framework for post-event analyses of hydrological extremes is demonstrated here using the winter half-year 2023/24 UK floods as a case study. We aim to place the winter half-year 2023/24 in context of past climate change, consider the possibility of UNSEEN outcomes beyond historical observations in a present-day climate and appraise trend detectability over the 21^st century. The ‘ClimaMeter’ circulation analogue-based attribution approach suggests that a 6-month period with similar atmospheric circulation patterns to the observed winter half-year 2023/24 has become warmer and wetter (by an average 8.8%) in the recent past (1945-2021) compared to the more distant past (1850-1925). Monthly river flow reconstructions extended back to 1850 show river flows during analogue events have increased by 13.5%. Pooling seasonal hindcasts following the UNSEEN approach show the potential for river flows to be 46% higher than the observed given a worst-case storyline. A maximised rainfall storyline further explores consequences of slight changes to the tracks of two major winter storms which could have resulted in much larger rainfall accumulations. Finally, river flow simulations driven by a single-model-initial-condition large ensemble place observed trends in context of internal variability, suggesting early emergence of climate signals for winter half year river flows for some areas but a signal may not emerge for some regions until mid-21^st century. Our research provides a proof of concept in extending storyline attribution approaches to river flows and highlights the changing risk of winter flooding in the UK. The same framework for post-event analyses can be applied to future events and elsewhere globally to assist long-term planning for climate change adaptation.