Comparison of Empirical and Numerical Methods for Estimating Seepage Losses in Unlined Irrigation Canals: in Laelay-Wukuro Irrigation Scheme, Eastern Tigray, Northern Ethiopia

Abstract

Efficient irrigation water management requires accurate estimation of seepage losses, particularly in unlined canals. The objective of this study was to quantify seepage losses and to evaluate the performance of empirical and numerical methods for estimating seepage in unlined irrigation canals. The study was conducted in the Laelay Wukro irrigation scheme, Eastern Tigray Northern Ethiopia, during the 2024 irrigation season. Seepage losses were assessed in six unlined secondary canal sections (SC1–SC6) with soil textures ranging from loamy sand to clay loam and hydraulic conductivities between 0.30 and 10.72 cm/hr. The methodology involved field measurement of seepage losses and their comparison with estimates obtained from six empirical equations (Davis–Wilson, Moritz, Molesworth and Yennidunia, Swamee, Ingham, and Muskat) and a numerical model (SEEP/W) implemented in GeoStudio software. Model performance was evaluated using statistical indicators including the coefficient of determination (R²), root mean square error (RMSE), mean absolute error (MAE), coefficient of residual mass (CRM), percentage average error (PAE), and bias. Results showed that the SEEP/W model provided the closest agreement with measured seepage losses (R² = 0.94, RMSE = 13.25, MAE = 24.12, CRM = 0.02, PAE = −15.4%, and Bias = −0.69), with an average deviation of 2.6% from observed values. In contrast, empirical methods significantly underestimated seepage losses due to their simplified assumptions. The study concludes that the SEEP/W numerical model is a more reliable and realistic tool for seepage analysis in unlined canals and can support improved canal design, seepage control, and irrigation water management in similar hydrogeological settings.