Abstract/Summary

Hydrological droughts threaten public water supply and significantly impact the natural environment and other sectors. Quantifying the plausible worst case drought in the present day climate and understanding the impacts of climate change on future droughts help decision-makers formulate plans to enhance resilience. However, existing studies are often dominated by a top-down approach characterised by a cascade of uncertainty which particularly suffers from an insufficient exploration of low-likelihood, high-impact outcomes. This thesis aims to navigate uncertainty in the hydrological impacts of climate change by introducing a physical climate storyline approach. Drought storylines created from process understanding, pooling of ensemble reforecasts and sampling from large ensemble climate model simulations enhances risk awareness and aids robust decision-making for water resources planning. Retrospective storylines of past droughts created by perturbations made to observed events embraces downward counterfactual thinking through identification of different ways in which events could have turned out worse. A more severe precipitation deficit in autumn 1975 for east England or an even drier winter 1975/76, both of which could have arisen from natural climate variability, could have led to a reduction in accumulated river flows by up to >50% at slow-responding catchments in East Anglia. Storylines of the 1976 drought could have surpassed the maximum intensity of the 1921-22 drought by up to >30% to become the most severe post-1891 drought for the region. Similarly, drier preconditions of the 2010-12 drought and a plausible third consecutive dry winter could have led to significantly more severe conditions across the UK. A third dry winter, in particular, could have led to conditions matching the benchmark multi-year drought for east England (1989-1993), highlighting the fact that the drought could be seen as a near miss. Placing the observed drought in a 2°C warmer climate is estimated to exceed mean deficit of the benchmark 1989-93 drought by >60% at some of the worst affected catchments. Initialized large ensemble model simulations is an emerging way to create high-impact, low-likelihood storylines. Storylines of the 2022 drought in East Anglia created by combining observations with pooled ensemble reforecasts enable decision-makers to explore the likelihood of worst-case river flow trajectories, identify high impact combination of physical climate drivers to increase risk awareness during an on-going event. Further, application of the UNprecedented Simulated Extremes using ENsembles (UNSEEN) technique estimates that the chance of unprecedented high (low) summer temperature (rainfall) increases from 5.7% (8.8%) in the present day to 58.3% (18.1%) in a 3°C warmer world. The larger sample simulations are ideal for searching for high impact drought sequences where the physical credibility of simulated events may be verified more easily compared to statistical methods. The utility of the current generation of large ensemble simulations to sample for multi-year drought storylines is further discussed with drawbacks due to spatial resolution, the need for bias adjustment and the under-estimation of weather system persistence (such as atmospheric blocking). The various methods presented in this thesis provide evidence of the magnitude of present and future extreme droughts. Outstanding research gaps to ensure the physical credibility of drought storylines include the need to improve quantification of observational data uncertainty and better characterisation of hydrological model realism during extreme droughts. The demonstrated value of the storyline approach contributes to the diversification of approaches used in water resources planning and could form a core part of planning to assist in stress-testing water resources systems and enhancing resilience to future droughts.