A numerical framework to advance agricultural water management under hydrological stress conditions in a data scarce environment

Abstract

Agriculture in Ethiopia has a historical dependency on rainfed (wet season) and surface water-based (dry season) irrigation projects, the performances of which rely heavily on JJAS precipitation. This is more sensitive in the Upper Blue Nile (UBN) basin, where precipitation uncertainty and hydrological droughts often overshadow the effectiveness of these irrigation schemes, challenging the water-food security efforts in general. To explore this issue, we developed a numerical framework, based on a groundwater model using MODFLOW-NWT, coupled with the outputs of the crop model DSSAT. The framework was implemented in an irrigated site namely, the Koga irrigation scheme, located within in the UBN Basin, Ethiopia. The coupling of groundwater modelling with crop water modelling in data scarce environments is a key contribution of this work. The groundwater model was calibrated with in situ data collected via a Citizen Science Initiative (CSI) for hydraulic head (H) and soil moistureθ; and against a distributed hydrological model (CREST) simulated evapotranspiration (ET). Normalized root mean squared error (NRMSE) values of 0.047, 0.05, and 0.06 were obtained through calibration for H, θ, and ET, respectively. The model was specifically used to simulate the vadose zone water availability due to irrigation, and water-food security aspects associated with extremely dry years. We investigated the hydrologic effect of three irrigation scenarios (non-regulated: NREG, regulated: REG, and regulated with groundwater pumping: REG+GW) aimed at reducing crop water stress derived by one-way coupling of MODFLOW with the crop model Decision Support System for Agrotechnology Transfer (DSSAT).