Uncertainty assessment in river flow projections for Ethiopia's Upper Awash Basin using multiple GCMs and hydrological models

Uncertainty in climate change impacts on river discharge in the Upper Awash Basin, Ethiopia, is assessed using five MIKE SHE hydrological models, six CMIP5 general circulation models (GCMs) and two representative concentration pathways (RCP) scenarios for the period 2071-2100. Hydrological models vary in their spatial distribution and process representations of unsaturated and saturated zones. Very good performance is achieved for 1975-1999 (NSE: 0.65-0.8;r: 0.79-0.93). GCM-related uncertainty dominates variability in projections of high and mean discharges (mean: -34% to +55% for RCP4.5, - 2% to +195% for RCP8.5). Although GCMs dominate uncertainty in projected low flows, inter-hydrological model uncertainty is considerable (RCP4.5: -60% to +228%, RCP8.5: -86% to +337%). Analysis of variance uncertainty attribution reveals that GCM-related uncertainty occupies, on average, 68% of total uncertainty for median and high flows and hydrological models no more than 1%. For low flows, hydrological model uncertainty occupies, on average, 18% of total uncertainty; GCM-related uncertainty remains substantial (average: 28%).