Storylines of the 2022 UK drought using seasonal hindcasts at Anglian catchments and boreholes

In this study, we investigate the drivers of winter rainfall for the region of eastern England supplied by Anglian Water using a large sample of plausible winters in hindcasts from the ECMWF seasonal forecasting system SEAS5. The SEAS5 hindcast dataset (1982-2021) is used to provide a large sample of plausible winters (Dec, Jan, Feb - DJF). A series of meteorological indices (such as NAO, EA, Nino3.4, polar vortex strength and the SST tripole index) describing atmospheric circulation patterns are calculated from both observed winters (ERA5 reanalysis - 1960-2015) and for each winter in the hindcast dataset. K-means clustering of all the calculated indices are used to create clusters with similar characteristics. Storylines were created to represent plausible pathways of the 2022 drought assuming winter 2022/23 resembled each of the four winter clusters. Storylines was simulated by running GR6J and Aquimod using the top parameter set for the baseline period up until November 2022 after which hindcast rainfall and PET data for each winter (DJF) in the four winter clusters were appended in place of winter 2022/23. This dataset contains simulated river flows and groundwater level for each storyline and for each of the selected river catchments or borehole in the Anglian Water region. Each file is named with the NRFA station id / groundwater borehole name and the cluster number (i.e. C1, C2, C3 or C4). In the LTA100 experiment, spring (MAM), summer (JJA) and autumn (SON) 2023 were assumed to have 100% long term average (LTA) rainfall by selecting the closest years matching 100% LTA rainfall in the observations. In the LTA60 experiment, it is assumed that summer (JJA) 2023 follow 60% LTA seasonal rainfall