Groundwater recharge can greatly contribute to the crop growth in farmland for the sustainable development of regional production in modern agriculture. It is crucial to precisely quantify the rational utilization of groundwater sources. Among them, groundwater recharge can be broadly classified into empirical and mechanistic models in recent years. However, some limitations are still remained in practical application, due to the complexity and highly nonlinear of groundwater recharge. In this study, a series of experiments were conducted on the winter wheat under different groundwater levels and irrigation treatments using 24 clustered large-scale lysimeters. The AquaCrop model was calibrated and validated using the soil moisture content, canopy coverage, biomass, and grain yield of winter wheat as evaluation indicators. The calibrated AquaCrop model was then utilized to simulate the growth and development of winter wheat, as well as the soil water dynamics under varying irrigation treatments and groundwater depths. An estimation model was proposed for the groundwater recharge in the period of winter wheat growth. Thereby, the critical groundwater levels were determined for the different growth stages, and the threshold depth of water table suitable for winter wheat growth. Results showed that: 1) The AquaCrop model was performed best on the growth and development of winter wheat under different irrigation treatments and groundwater levels. Simultaneously, the better performance was achieved in the statistical parameters between the simulated and measured values of soil water content, canopy coverage, biomass, and grain yield during the growth period, with the determination coefficient (R²)≥0.8, the relative error (RE) and the normalized root mean square error (NRMSE) are both＜16%. 2) The validated AquaCrop model was employed to simulate the growth of winter wheat under 52 scenarios of groundwater depth and irrigation treatments. Subsequently, the nonlinear regressions of soil water storage - groundwater recharge intensity and groundwater depth - soil water storage were fitted during the growth period of winter wheat. The groundwater recharge intensity was calculated to compare with the measured values for verification. The statistical parameters fell well within the acceptable error range of the model, such as R²≥0.873, RE≤9.22%, RMSE≤0.316 mm/d and NRMSE≤11.89%. 3) The critical tables of groundwater were calculated in different growth stages of winter wheat under various irrigation treatments. The critical tables of groundwater were 3.28, 3.36 and 3.54 m, respectively, for the winter wheat during the seedling, frozen and reviving-jointing stages under irrigation. Furthermore, the critical table of groundwater was 3.38 m under the sufficient irrigation treatment in the whole growth period. There was the various behavior in the different growth periods of winter wheat under the insufficient irrigation treatments. In general, the critical table of groundwater decreased with the increasing irrigation amount. The optimal threshold water table was ranged from 2.0 to 2.5 m for the winter wheat, according to the relationship between grain yields and groundwater levels under 52 simulated scenarios. Therefore, the yield and water use efficiency (WUE) of winter wheat were simultaneously reached their optimal levels under the insufficient irrigation treatment (at reviving and filling stages), which were 8.848 t/hm² and 2.17 to 2.43 kg/m³, respectively. The findings can also provide the technical support for the accurate assessments of groundwater contribution for the precision irrigation and decision-making management.making management.