A novel irrigation canal scheduling model adaptable to the spatial-temporal variability of water conveyance loss

Abstract

The water conveyance loss of irrigation canals shows significant spatial-temporal variability both along the canal reach and over time. An accurate representation of these variances in irrigation canal scheduling is essential to reduce water conveyance loss. However, the existing optimal irrigation scheduling models (OISMs) still have deficiencies in dealing with spatial-temporal variability of water conveyance losses, which seriously limits the model performance in many irrigation areas with high spatial-temporal variability. Borrowing a concept from hydrology, we propose a Dynamic Calculation Method of canal water conveyance loss (DMWCL), in which antecedent water conveyance index (AWCI) is defined to represent the effects of antecedent soil water content upon infiltration losses, and the integral-flow balance method is adopted to describe the characteristics of canal discharge decreasing along the canal reach. Based on the DMWCL, we present a novel canal scheduling model with the potential to adapt to spatial-temporal variability (DMWCL-CS). The new model is applied to the Hetao Irrigation District, Inner Mongolia, a typical agricultural watershed in China. Our results across the case studies show that the canal water delivery schedule derived by the DMWCL-CS model reduces water conveyance loss and improves irrigation performance, ranging from 2.49 million m3 to 10.60 million m3 in 2014 and from 0.60 million m3 to 6.17 million m3 in 2016, respectively, compared to that of three additional models. Interestingly, we find that the water delivery schedule of the DMWCL-CS model accelerates the wetting process of the canal bed soil to reduce the water conveyance loss and increases the discharge to adapt to the characteristics that the discharge decreases along the canal reach. The DMWCL-CS model provides essential insights for irrigation managers to formulate a canal water delivery schedule that adapts to spatial-temporal variability. The proposed framework is generic and can be integrated into any OISM.