Prediction of irrigation return flows through a hierarchical modeling approach

The quantity of water available for irrigation is getting scarce in many countries and it assumes great importance for assured crop production, especially in view of the erratic behavior of the monsoon. Thus, there is a pressing need to improve the water efficiency of irrigation systems. One-way of improving the efficiency of the irrigation system is reusing the return flow from the irrigation system. This task requires quantification of return flow, which still remains as a grey area in irrigation water management. The estimation of return flow from the irrigation system is usually obtained using thumb rules depending upon the site-specific conditions like command area conditions and soil properties. In this paper, a hierarchical modeling technique, namely, regression tree is developed for return flow estimation. Regression tree is built through binary recursive partitioning. The effective rainfall, inflow, consumptive water demand, and percolation loss are taken as predictor variables and return flow is treated as the target variable. The applicability of the hierarchical model is demonstrated through a case study of Periyar–Vaigai Irrigation System in Tamil Nadu, India. The model performance shows a good match between the simulated and the field measured return flow values. Results of statistical analysis indicated that the correlation coefficients are high for both single as well as double crop seasons.