The classic Klobuchar model is widely used in navigation and positioning. However, the ionospheric delay correction accuracy is difficult to meet the high-precision positioning. In this paper, the observation data provided by the International Global Navigation Satellite Systems Service Center are used to calculate the Vertical Total Electron Content (VTEC) with the Klobuchar model and the Dual-frequency correction model. Then the Auto-regressive Integrated Moving Average (ARIMA) model is used to forecast the error of the 9th day and 10th day between the Klobuchar model and the dual-frequency correction model base on the error of the former eight days. The forecast results are used to improve the model. Finally, the accuracy of the improved model is to be evaluated in different environment and different latitudes. The results show that the average relative accuracy of the improved Klobuchar model is 71.66% and 69.69% in the ionospheric active period and ionospheric quiet period, respectively. The improved Klobuchar model is more consistent with the dual-frequency correction model, and can better to reflect the temporal evolution characteristics of the ionosphere.