Groundwater recharge estimation in the Ziway Lake Watershed, Ethiopian Rift: an approach using SWAT and CMB techniques

The community in the Ethiopian Rift is highly dependent on groundwater for human consumption, irrigation purposes, and is also essential for the availability of surface water resources (e.g., rivers, lakes, and wetlands). Therefore, it is essential to quantify the spatial and temporal rate of replenishment of the aquifers in the region for sustainable water resource development and management. The aim of this study is to estimate the amount of recharge and its spatial distribution in the ZLW (Ziway Lake Watershed), situated in the central Main Ethiopian Rift (MER). For this purpose, the two widely applied groundwater recharge estimation methods, such as the SWAT (Soil Water Assessment Tool) and CMB (Chloride Mass Balance) techniques, were employed. Hydro-meteorological data from 1979 to 2014 was utilized for SWAT model setup, calibration, and validation purposes. The performance evaluation showed acceptable ranges of model evaluation indicators (R² = 0.67–0.73, NSE = 0.63–0.7, and PBIAS = -0.83–-12.1). The SWAT– simulated mean annual recharge (RCHG) showed a general decreasing trend from the opposite–facing rift margins (160 mm/year) towards the rift floor (35 mm/year) with an average value of 98.1 mm/year. Though the magnitude is different, a similar RCHG pattern is obtained using the RCHG rate estimated using the CMB approach over the same watershed, with a mean annual RCHG of 163.82 mm/year (18.39% of the precipitation). The SWAT–simulated RCHG value is lower than the RCHG computed by the CMB approach, indicating the importance of preferential RCHG, which the SWAT model does not capture. Therefore, the SWAT–simulated RCHG should be taken as the lowest plausible groundwater replenishment, while the RCHG estimated by the CMB method represents the highest plausible RCHG estimate for the watershed characterized by highly fractured volcanic and sediment aquifers. Besides the spatial variability, the simulation result also indicates that RCHG is highly variable with time. The RCHG rate is strongly associated with the annual precipitation trend and tracks a bimodal annual pattern of the dry and wet seasons. The RCHG variability across the study area is largely controlled by climate and topography, though the small–scale pattern within the physiographic regions is connected to variability in LULC types and soil hydraulic characteristics. Quantification of the RCHG for the ZLW has immense importance for policymakers and water resource experts for further water resource study and developmental activities in the region.