Present-day risk from winter storms in the United Kingdom

Winter storms cause significant impacts to a range of sectors in the United Kingdom (UK) (Hanlon et al., 2021; Kendon et al. 2023). The nature of winter storms is that they are associated with multiple hazards (for example strong winds, rain, and storm surges) which will most often occur as compounding hazards: simultaneously, over large areas or one after another (Bloomfield et al., 2023, 2024; Kew et al., 2024; Zscheischler et al., 2020). The complexity of the hazards from winter storms makes them challenging to plan for by resilience specialists and critical infrastructure operators (Bloomfield et al., 2023; Zscheischler et al., 2020). Strong winds and gusts, alongside compounding impacts from rainfall, for example, are known to lead to societal disruption, such as to energy distribution lines (e.g., Gonçalves et al., 2024).

Some recent work has focused on understanding which aspects of a storm’s development lead to compound impacts over the UK (e.g. Manning et al., 2024) but no existing UK storm classification has focused on informing hazards and impacts potential. In this project we explore the feasibility of a new framework to better assess risk from the compounding hazards in winter storms to facilitate better preparation by the resilience community. The approach is based on techniques used by operational meteorologists to anticipate the potential outcomes of storms. From literature review and expert interviews, we believe that this is a novel approach within the resilience planning setting.

When a storm approaches the UK, meteorologists must quickly determine the likely impact on a wide range of sectors, determine worst-case scenarios and build a picture of the level of predictability. One approach they use is to assess a range of aspects of the impending storm related to its dynamical development (we refer to these as ‘storm development metrics’) and use these to rapidly validate the predicted hazards by the forecast models, as well as to identify potential high-risk outcomes. They often refer to previous storms with similar characteristics to infer possible scenarios. We investigate whether these ‘storm development metrics’ can be used to create a ‘typology’ of storms that can separate storms by the plausible hazard scenarios that could occur, and hence simplify the task of assessing risk from storms now and in the future.

Different machine learning clustering techniques are applied to the development metrics from a large set of historical named storms that affected the UK to explore the discrimination power in the hazard space of the resulting cluster sets. This furthers the project aim to convert the technical understanding of operational meteorologists into more digestible information for resilience specialists, building capacity to manage the threat of multi-hazard storms.