Abstract: Abstract: Healthy groundwater cycle can ensure that water resources are used more efficiently and securely in northern irrigation district. In recent years, groundwater cycle condition in the irrigation district affected by climate change and human activities has changed greatly. Environmental problems such as the attenuation of groundwater storage capacity, hanging pump wells and the groundwater deterioration occur with the unhealthy groundwater cycle in some northern areas, which directly affect the safety and efficiency of water resource utilization in the irrigation district. Therefore, studies on response of groundwater cycle to environmental changes in the irrigation district are urgent and important. This study took Jinghui Canal Irrigation District in Shaanxi province as a research area, analyzed variations of characteristics of external environment factors for groundwater system and groundwater cycle elements over the years by trend analysis and spearman rank correlation test. A forecasting model of groundwater depth affected by external environment was established based on multivariate time series CAR model (Controlled Auto-regressive). Groundwater depth under different environmental scenarios were predicted using validated models. The prediction problem of complex nonlinear time series can be effectively solved by using CAR model. In order to evaluate the prediction effects of CAR model， its results were compared with those from other models including support vector machine (SVM) prediction model and radial basis function (RBF) network model. The results showed that the prediction effect of CAR model was much better than SVM model and RBF network model. The specific research results of this paper showed that the main external environment factors affecting groundwater cycle were precipitation, evaporation, irrigation intake water from canal head, and irrigation water ratio of channel and well. Precipitation was in a significantly decreasing trend while evaporation was in a unnotable increasing trend from 1955 to 2010. The Hurst index of precipitation and evaporation were 0.69 and 0.56 respectively. The irrigation intake water from canal head, the surface irrigation water use and the groundwater exploitation showed a decreasing trend from 1977 to 2010, and was reduced by 62.5%, 44.7%, and 34.5% respectively. With the decrease in the irrigation water ratio of channel and well, the groundwater depth tended to increase gradually. The decreased precipitation, the increased evaporation, also the reduced amounts of groundwater recharge and the increased amounts of excretion all led to groundwater level dropping gradually, which dropped from 395.4 m in 1977 to 383.6 m in 2010 and the cumulative decline was 11.8 m in nearly 34 years. The simulation results of groundwater depth under different environmental scenarios showed that in scenarioⅠwith the average precipitation of 512.5 mm, the suitable irrigation water ratio of channel and well for keeping groundwater cycle healthy was 1.53. When the irrigation intake water from canal head was 2.15×108 m3, and the groundwater exploitation was 1.39×108 m3, so that the groundwater level could be stabilized with an average level and the balance of groundwater recharge and discharge could be maintained. In scenarioⅡ with precipitation reduced to 486.9 mm, the suitable irrigation water ratio of channel and well was 1.61. Groundwater recharge and discharge balance could be reached when the irrigation intake water from canal head was 2.19×108 m3 and the groundwater exploitation was 1.36×108 m3.