Conventional groundwater numerical models often face limitations in predicting future groundwater levels due to the difficulty of obtaining precipitation and evaporation data. To address this issue, this study puts forward an improved groundwater level prediction method. We employ the autoregressive integrated moving average (ARIMA) model to predict time-series data for precipitation and evaporation, then integrate the prediction results with the groundwater flow model of groundwater modeling system (GMS) to simulate changes in groundwater levels in the Yang River Basin. This new method is then applied to analyze the historical meteorological data of the Yang River Basin from 2000 to 2020. The ARIMA model is used to predict precipitation and evaporation in 2021, and the predicted data are input into the GMS to conduct groundwater level simulation experiments. The results indicate that the GMS performs well in simulating groundwater levels in the Yang River Basin, with most NSE values ranging from 0.71 to 0.96, and RMSE values between 0.05 and 0.45 m, demonstrating high overall accuracy. The ARIMA model exhibits strong predictive performance for meteorological data, with prediction accuracy for evaporation outperforming that for precipitation. The combined approach of the ARIMA and GMS models demonstrates high accuracy and applicability, which can provide technical support for regional groundwater resource management. The proposed method can overcome the limitations of dependence on future data availability faced by conventional models, offering a viable reference for groundwater level prediction in similar regions.