Water shortage and soil salinization are the main issues threatening the sustainable development of agriculture and ecology in the Hetao Irrigation District (HID). The application of water-saving practices is required for sustainable agricultural development. However, further study is required to assess the effects of these practices on water and salt dynamics in the long term. In this study, the impacts of different water-saving practices on water and salt dynamics were investigated in the HID, Northwest China. The SAHYSMOD (integrated spatial agro-hydro-salinity model) was used to analyze the water and salt dynamics for different water-saving irrigation scenarios. The results indicate that the SAHYSMOD model shows a good performance after successful calibration (2007–2012) and validation (2013–2016). The soil salinity of cultivated land in the middle and upper reaches of the irrigation district decreased slightly, while that in the lower reaches increased significantly over the next 10 years under current irrigation and drainage conditions. It is predicted that if the amount of water diverted is reduced by up to 15%, the maximum water-saving volume could reach 650 million m³ yr^–1. For the fixed reduction rate of total water diversion, the prioritized measure should be given to reduce the amount of field irrigation quota, and then to improve the water efficiency of the canal system. Although a certain amount of water can be saved through various measures, the effect of water saving in the irrigation district should be analyzed comprehensively, and the optimal water management scheme should be determined by considering the ecological water requirement in the HID.

The water movement research in irrigation districts is important for food production. Many hydrological models have been proposed to simulate the water movement on the regional scale, yet few of them have comprehensively considered processes in the irrigation districts. A novel physically based distributed model, the Irrigation Districts Model (IDM), was constructed in this study to address this problem. The model combined the 1D canal and ditch flow, the 1D soil water movement, the 2D groundwater movement, and the water interactions among these processes. It was calibrated and verified with two-year experimental data from Shahaoqu Sub-Irrigation Area in Hetao Irrigation District. The overall water balance error is 2.9% and 1.6% for the two years, respectively. The Nash–Sutcliffe efficiency coefficient (NSE) of water table depth and soil water content is 0.72 and 0.64 in the calibration year and 0.68 and 0.64 in the verification year. The results show good correspondence between the simulation and observation. It is practicable to apply the model in water movement research of irrigation districts.