Routing climate model runoff from CMIP6 to project future changes in global river discharge

Hydrological extreme events are expected to change in frequency and magnitude due to anthropogenic climate change. Climate impact studies investigating effects on floods and hydrological droughts require knowledge on discharge along river networks. However, global climate models (GCMs) focus on runoff at the grid cell level, have only a coarse resolution, and typically address runoff routing externally. To bridge this gap, atmospheric data (e.g., precipitation, temperature) from GCMs are fed into global hydrology models (GHMs). While this approach can benefit from the additional detail of GHMs, which are dedicated to resolve the terrestrial water balance, it can only consider a limited number of GCM projections. This implies that the substantial spread imposed by both GCM uncertainty and internal climate variability may be underestimated. To overcome this limitation, we route daily runoff from multiple models contributing to the 6^th phase of the Coupled Model Intercomparison Project (CMIP6) along the river network. For this we use the multiscale routing model mRM which can be flexibly adapted on a range of spatial scales by deriving an upscaled river network from high-resolution morphological data. The fidelity of the considered modelling chain is carefully evaluated in light of the underlying assumptions and the scale mismatch between the spatial resolution of the GCMs and the routing model. The new global discharge projections are used to explore the effects of anthropogenic climate change on mean and extreme river flow considering the uncertainty imposed by models contributing to the CMIP6 archive.