Using sub-daily precipitation for grid-based hydrological modelling across Great Britain: Assessing model performance and comparing flood impacts under climate change

Abstract

Study regionGreat Britain. Study focusNational-scale grid-based hydrological models are usually run at fine spatial and temporal resolutions, but driving data are often not available at the required resolutions. Here, a recent observation-based hourly 1 km gridded precipitation dataset is applied with a 1 km hydrological model to simulate daily mean river flows. Performance is compared to use of equally-disaggregated and profile-disaggregated daily data, for a large number of catchments. Hourly and daily precipitation from a high-resolution convection-permitting climate model (CPM) are then used to drive the hydrological model for baseline (1980–2000) and future (2060–2080) periods, to investigate differences in potential peak flow changes. New hydrological insightsOn average, use of observation-based hourly data provides a clear improvement over equally-disaggregated daily data for high flows and peak flow bias, a small improvement for average flows and mean flow bias, but little difference for low flows. Performance in faster-responding catchments typically improves more; performance in some catchments degrades. Use of profile-disaggregated daily data provides the small mean flow bias improvement and some peak flow bias improvement, but other factors degrade. On average, future changes in peak flows from hourly CPM precipitation are only slightly larger than from equally-disaggregated daily data. Future work will look at simulation of hourly mean flows.