Basin-scale hydrological impacts from climatic changes: a model assessment in the Horn of Africa

Water resources modelling and management in arid catchments must face three main challenges: 1) the quantification of flood magnitudes, 2) the quantification of groundwater recharge for sustainable water extraction, 3) the evaluation of shifts in surface/subsurface hydrologic partitioning as a consequence of climate change. Tackling these challenges requires the development and use of a diverse set of tools, ranging from continuous hydrological modelling to the estimation of rainfall in ungauged areas, to the evaluation of quantitative climate change scenarios. The present thesis develops and applies these tools in the context of the Gheba River Basin, with outlet at Adi-Kumsi (Tigray, Ethiopia). This area exhibits strong seasonal and spatial rainfall fluctuations, whose quantitative characterization is made very challenging by observational data scarcity. This study integrates ground-based observations (1999–2002), global precipitation datasets (CHIRPS, TRMM, ERA5, TerraClimate), and stochastic weather generation to produce an enhanced evaluation of available precipitation input and to evaluate hydrologic partitioning in the Gheba Basin. CHIRPS and TRMM outperformed other datasets in a comparison with observations, with CHIRPS exhibiting lower bias and TRMM capturing peak flows more effectively. ERA5 and TerraClimate showed significant biases, limiting their suitability for hydrological applications. A 100-year hourly precipitation dataset was then generated using a Bartlett-Lewis stochastic weather generator, calibrated on the 1999–2002 observed precipitation period. The generated dataset closely preserved key statistical properties of the observed data, including mean, variance, and wet-dry cycle patterns, though it underestimated extreme events. Three climate scenarios were applied, with the first scenario directly informed by CMIP5 simulation results, while the remaining two represent hypothetical experiments designed to explore potential climate extremes. These include: (i) Moderate Increase Scenario (+10% annual precipitation, +17% wet-season); (ii) Intensified Rainfall Scenario (+30% daily mean, +50% variance); and (iii) Prolonged Dry Scenario (-30% mean precipitation, +10–25% dry fraction). A HEC-HMS model was calibrated on the available rainfall and discharge observations and was subsequently used to evaluate the water balance at the basin scale. Results reveal that Scenario 2 led to a 54% increase in total precipitation, amplifying total flow (187%) and direct runoff (292%), posing heightened flood and erosion risks. Conversely, Scenario 3 resulted in a 43% reduction in precipitation, leading to a 53% decline in aquifer recharge and a 48% reduction in total flow, exacerbating drought conditions. Scenario 1 yielded an increase in infiltration and groundwater recharge. These findings demonstrate that hydrological shifts in response to climate perturbations are nonlinear and strongly influenced by precipitation distribution rather than mean changes alone. While some deviations were observed between imposed and modelled changes, these are expected given that the stochastic model was calibrated using only four years of observed precipitation and a single 100-year simulation. The results underscore the inherent variability in stochastic hydrological modeling, where changes in precipitation characteristics can lead to amplified or dampened hydrological responses. A longer observational record and ensemble-based simulations would improve alignment with imposed changes and strengthen confidence in projections. Despite these uncertainties, this study provides a robust framework for assessing climate-induced hydrological shifts in data-scarce, semi-arid environments. The results highlight the Gheba Basin’s dual vulnerability to both extreme flood and drought conditions, emphasizing the need for adaptive water resource management strategies to enhance resilience in the face of increasing climate variability.