The surface water resource in Weibei upland is scarce and the groundwater is the main water supply source for city and rural area. In order to analyze the groundwater variation characteristics, based on the shallow observation hydrometeorology data of Weibei Upland from 2000 to 2009, the variation factors were defined as four types (irrigation-exploitation, infiltration-evaporation, infiltration-exploitation & runoff and runoff-exploitation) according to the relationship between the recharge-discharge condition with natural and human activity influencing factors. The groundwater variation characteristics of each type were analyzed and the groundwater depth was simulated and predicted by autoregressive integrated moving average model (ARIMA). The results showed that the model meet the actual data very well.