Application of parameter estimation and regionalization methodologies to ungauged basins of the Upper Blue Nile River Basin, Ethiopia

Runoff variability of large ungauged international basins affects water availability to downstream countries making management of transboundary water difficult and sometimes causing geopolitical tension. The key to successfully predict runoff and its fluctuation in both gauged and ungauged basins is a reliable hydrologic model and an appropriate parameter regionalization approach. The Upper Blue Nile River Basin of Ethiopia is an example where data are limited, water resources are valuable, and competition for water among riparian countries is strong. To predict runoff variability from the basin as well as from ungauged subbasins, a water balance model with six parameters was tested. The model parameters were identified through a sensitivity analysis and Monte Carlo experiments and their plausible search ranges were constrained. Six parameter regionalization methods (global mean, subbasin mean, multiple regression, regional calibration, aggregated calibration, and volume fraction calibration) were evaluated by model performance, volume error, and runoff volume fractions of subbasins. The results showed that the high interdependency of model parameters affects the development of regional relationships. The model parameters regionalized in different basin scales were not transferable to one another due to different basin responses. The study found that regional calibration is suitable for simulating runoff of sites (100–10,000 km2), and volume fraction calibration is suitable for subbasins (14,000-50,000 km2). Lumped calibration managed to simulate runoff dynamics of lumped basins (65,000 - 180,000 km2) representing two or more aggregated subbasins