Optimizing regional irrigation water use by integrating a two-level optimization model and an agro-hydrological model

Water scarcity has been a crucial issue for sustainable irrigated agriculture in the arid regions. In these regions where conserving water is paramount, optimal allocation and utilization of irrigation water is particularly important. In this study, a process-based regional economic optimization (PBREOP) model was developed for maximizing irrigation water use efficiency and economic benefit of an irrigation system. The PBREOP model is a two-level optimization model with combined use of an agro-hydrological model (SWAP-EPIC). The first level (farm scale) dealt with the optimal distribution of irrigation water and cropping pattern considering various crops and soils in a subsystem, using a non-linear programing technique. The second level (district scale) sought out the optimal strategy for irrigation water allocation among different subsystems using a dynamic programing algorithm. The crop water production functions (CWPFs) were an important component of the first-level objective function. They were derived with the SWAP-EPIC model considering different irrigation alternatives. The model was solved using the decomposition-harmonization method for large systems. The Yingke Irrigation District (YID) in the middle Heihe River basin, Northwest China was used as a case to test the PBREOP model. Nine CWPFs for three major crops and three major soils were firstly derived based on the simulations of different irrigation levels and climate conditions (20 years). Next, the PBREOP model for YID was established with 11 subsystems, and applied to the irrigation water use optimization under five water supply scenarios. Results showed that the total economic benefit in YID could be increased by 15% on average through the optimization of water allocation and cropping pattern with the same water supply amount as that of the current situation. A variation range of the risk was also obtained with considering the impacts of climate uncertainties. Scenario analysis showed that the total irrigation water could be reduced by 23% on average without benefit reduction when compared to the benefit of the present situation. Model test indicated that the proposed PBREOP model can efficiently optimize irrigation water use and cropping pattern on a regional scale.