Work

How do global soil databases influence streamflow simulation? A SWAT+ study in the Lake Tana sub-basin, Ethiopia

Year: 2026

Type: article

Abstract: Lake Tana sub-basin in Ethiopia This study evaluated three global soil datasets, DSOLMap, HWSD, and DSMW, for simulating streamflow in the Lake Tana sub-basin, Ethiopia, using the SWAT+ model. Model calibration (1995–2002) and validation (2003–2007) were performed using a multisite approach, with performance assessed before and after calibration using KGE, PBIAS, and R² metrics. Before calibration, HWSD consistently outperformed other datasets across all stations and periods, providing the most accurate monthly streamflow simulations and long-term mean flows. In contrast, DSOLMap exhibited the weakest performance at most stations in both periods, alongside DSMW at Gilgel Abay, indicating that these databases are less suitable for pre-calibration streamflow simulation in the study area. After calibration, database performance varied, highlighting the importance of parameter optimization. During 1995–2002, DSMW at Gilgel Abay and DSOLMap at Gumara showed the best performance, while HWSD remained robust but ranked slightly lower. During 2003–2007, no single database consistently dominated; top performance varied by watershed: DSOLMap and HWSD in Gilgel Abay, DSOLMap and DSMW in Gumara, and HWSD in Megech. Overall, soil database performance varies by station, simulation period, and calibration status. HWSD performs best for uncalibrated simulations, while post-calibration performance depends on site-specific conditions, underscoring the need to consider local watershed characteristics and calibration goals, especially in data-scarce regions. • Three global soil databases in SWAT+ were evaluated for 1995–2002 and 2003–2007 periods. • HWSD gave the most accurate pre-calibration streamflow across stations and periods. • DSOLMap performed weakest pre-calibration at most stations in both periods. • No database consistently outperformed the others across all stations and periods after calibration. • Best performers databases varied by station, period and calibration status.

Source: Journal of Hydrology Regional Studies

DOI: 10.1016/j.ejrh.2026.103455