The irregular changes of VTEC in extreme weather will increase the difficulty of prediction.The VTEC values of the ionosphere in Guilin in 2019 and 2020 provided by the European Orbit Determination Center(CODE)are used as the sample values.The sliding quartile distance method is used to detect the ionospheric VTEC disturbance law during the rainstorm in Guilin,and a variety of prediction models are established to com-pare the applicability and prediction accuracy of the models during the rainstorm.In this paper,five time series prediction models(ARIMA model,Single exponential smoothing model,Holt exponential smoothing model,Brown exponential smoothing model and Damped exponential smoothing model)are selected to predict the iono-spheric VTEC during the rainstorm in Guilin in the next 24 h.The results show that the ARIMA model has the best prediction effect within 4-9 h,and the RMSE and MAE of the predicted values are less than 2 TECu.Sin-gle exponential smoothing model and Damped exponential smoothing model are second,and Holt exponential smoothing model and Brown exponential smoothing model have the worst prediction effect.After more than 9 h,the prediction ability of each model is weakened,RMSE and MAE are greater than 3 TECu,which can not meet the needs of accurate prediction.
关键词
暴雨/VTEC/时间序列预测模型/桂林
Key words
heavy rain/VTEC/time series forecasting model/Guilin
维普
万方数据