Simulation modeling for effective management of irrigation water for winter wheat

Abstract

Lack of water resources, low irrigation efficiency, and inappropriate irrigation decisions severely restrict agricultural production in arid and semi-arid regions. Therefore, rapid and accurate decision-making regarding crop irrigation in real time is necessary. This study optimized irrigation scheduling by using information on different meteorological years and obtained the average soil water content (0–60 cm) before each irrigation, the corresponding irrigation time, and the water available for irrigation. The relative development speed of winter wheat and the amount of water available for irrigation were considered, and a dynamic irrigation water limit model was constructed. Winter wheat field experiments over 3 years (2016–2019) were followed by an evaluation of the regional applicability of the decision support system for the agrotechnology transfer model. A long short-term memory network effectively predicted air temperature and solar radiation; the R2 and root mean square error values were 0.802–0.964% and 12.53–23.9%, respectively. Public weather forecasts can be used to accurately predict rainfall, with 87.3% and 57% accuracy rates for forecasts of no rain and rain, respectively. Compared with traditional irrigation, the use of this dynamic irrigation lower limit for irrigation forecasting can increase yield and attendant net benefits. When two and three irrigation treatments were applied during the winter wheat growth period, the 3-year average yield increased by 10.3% and 4.4%, respectively, and the net benefit increased by 19.1% and 7.4%, respectively. The proposed method avoids relying on only field experiments to determine the irrigation lower limit and allows for the effective implementation of optimized irrigation schedules and the dynamic correction of irrigation plans in arid and semi-arid areas.