Kutiev, Ivan

A prediction method based on a simple auto-regressive model has been developed for short-term prediction of ionospheric characteristics. The method determines the auto-correlation function for the hourly values of the parameter of interest, using the time series from the previous 25 days. The resulting weighting coefficients can then be used to forecast future values of the parameter. The method has been applied to predict f0F2 up to 24 h ahead for stations Uppsala, Slough, Poitiers and Sofia. Error statistics are presented