Use of high temporal resolution data to identify the key drivers and locations where walls of water occur in the UK

Abstract

Walls of water (WoW) are a subset of flash floods characterised by an extremely fast increase in the discharge rate of rivers. In the UK, WoWs, events where an almost instantaneous increase in river flow happens, are responsible for several deaths, even when the maximum peak flow of the said event is not as noticeable. Using a national 15-minute continuous dataset, this study identified WoWs for catchments in the UK. Next, the antecedent atmospheric conditions for these WoWs were extracted from gridded datasets. Furthermore, catchment descriptors such as catchment area, elevation, slope, land use, and permeability of every catchment were downloaded from the National River Flow Archive. Finally, with the use of machine learning algorithms, that is, tree regressions and neural networks, this study identified vulnerable catchments and key conditions for WoWs to occur. Early results indicate that WoWs are not solely driven by rainfall intensity and that larger catchments (>500km) with low permeability are the most vulnerable to these hazards. Further studies using additional atmospheric conditions, i.e., temperature and windspeed will allow a better understanding of the drivers of these events.