Abstract: To ensure sustainable utilization of soil and water resources in canal-well irrigation district，it is very important to optimize the allocation of surface and groundwater. In order to reasonably determine spatiotemporal allocation scheme of water resources in canal-well irrigation district，a spatiotemporal optimal allocation model system for water resources coupled groundwater simulation model was developed in this study. The temporal optimal allocation model was subjected to 2 objectives of minimum water shortage and minimum difference of water shortage among months. The spatial optimal allocation model was subjected to objective of minimum difference of water shortage among canal units. The 2 models were solved by artificial fish swarm algorithm and particle swarm optimization，respectively. The Visual MODFLOW software was used to establish the three-dimensional groundwater simulation model of the irrigation area. Coupling variables between the simulation model and optimal allocation models were the monthly optimal supply result of surface water and groundwater, which were the inputs of groundwater simulation model. The spatiotemporal optimal allocation scheme of water resources were obtained by minimizing the sum variations of groundwater level of wet year, median year and dry year. The coupling system of spatiotemporal optimization of water resources and groundwater numerical simulation model might make full use of the surface water and groundwater, and also control the amplitude of groundwater level. Meanwhile, the spatiotemporal optimal allocation scheme of surface water and groundwater were obtained during wet year, median year and dry year. Jinghuiqu irrigation district is located in the middle part of Shaanxi province, between 34°25′20″N to 34°41′40″N, 108°34′34″E to109°21′35″E. The district is a typical canal-well irrigation district in the northwest of China，with an irrigation area of 90.3 khm². It is an important production area of agricultural and sideline products of Shaanxi. The simulation results in the irrigation district showed that the simulated water level fitted better the measured values for each observation well by the groundwater simulation model, with correlation coefficient of 0.96. Optimal results in Jinghuiqu irrigation district showed that water demand could be met in wet year and median year. Water shortage in dry year in 2020 and 2030 were 44.89 million m³ and 39.41 million m³, respectively. In 2020, groundwater exploitation quantity in wet year, median year and dry year were 114.39 million m³, 140.74 million m³ and 175.0 million m³, respectively. In dry year，groundwater irrigation was supplied in winner irrigation duration of December and January, and summer irrigation duration of June，July and August. Irrigation water shortage occurred in December, March，June, July and August, which were the duration of peak water demand and high sediment concentration. The average variations of groundwater level in wet year, median year and dry year were 0.49, 0.06, and -0.42 m, respectively. Balance between groundwater recharge and discharge basically realized. In 2030, groundwater exploitation quantity in wet year, median year and dry year were 124.57 million m³, 144.2 million m³ and 161.0 million m³, respectively. The average variations of groundwater level were 0.21, -0.08, and -0.26 m, respectively. Groundwater levels declined slightly. The model system of optimal model coupling groundwater simulation model can make full use of surface water and groundwater, and control the decline trend in the groundwater level. It is an effective way to achieve rational use of water resources and ecological health in canal-well irrigation district.