Abstract
Assessing the potential impacts of climate change on river flows is critically important for adaptation. Data from global or nested regional climate models (GCMs/RCMs) are frequently used to drive hydrological models, but now there are also very high-resolution convection-permitting models (CPMs). Here, data from the first CPM climate ensemble for the UK, along with the RCM ensemble within which the CPM is nested, are used to drive a grid-based hydrological model. The performance for simulating baseline (1981–2000) river flows is compared between the RCM and the CPM, and the projections of future changes in seasonal mean flows and peak flows are compared across Britain (1981–2000 to 2061–2080). The baseline performance assessment shows that (before bias correction) the CPM generally performs better than the RCM, and bias correction of precipitation makes both the RCM and CPM perform more similarly to use of observation-based driving data. The analysis of future changes in flows shows that the CPM almost always gives higher flow changes than the RCM. If reliable, these differences in flow projections suggest that adaptation planning for high flows based on use of regional data may be insufficient, although planning for low flows may be slightly over-cautious. However, the availability of CPM data only for one RCM/GCM is a limitation for use in adaptation as it under-samples the uncertainty range. There are significant challenges to the wider application of CPM ensembles, including the high computational and data storage demands.