Ionospheric forecast over Europe based on an autoregressive modelling technique 
driven by solar wind parameters 
Tsagouri I., Koutroumbas K. and Belehaki A. 
 
 
The reliable performance of several applications including HF communications, 
and satellite positioning and navigation systems during periods characterized by 
solar and geomagnetic disturbances requires accurate ionospheric predictions 
under all possible geophysical conditions. In practice, the ionospheric forecast 
is widely approached by empirical methods which are based on the correlation of 
the ionospheric disturbances with various geophysical indices, but since there 
is no efficient geophysical index to predict the ionospheric storm onset, its 
magnitude and duration, the ionospheric forecast remains an unsolved and a 
challenging problem. 
Based on recent advances in ionospheric storm dynamics which correlate the 
ionospheric storm effects with solar wind parameters such us the magnitude and 
the orientation of the interplanetary magnetic field (IMF) and on the 
availability of these parameters in real time by the ACE spacecraft from the 
vantage L1 point, an new ionospheric forecasting method was envisaged and 
proposed in this contribution. Because it takes about an hour for the solar wind 
to travel from where ACE is, in L1 point, to the Earth, telemetry from ACE 
allows alerts of geospace storms to be issued nominally an hour in advance of 
their occurrence effects to geospace environment. The proposed method is based 
on the fusion of two diverse techniques, an autoregression forecasting algorithm 
capable for real time ionospheric predictions, namely Time Series AutoRegressive 
(TSAR) model and an empirical method for predicting the onset and for scaling 
ionospheric disturbances during geomagnetic storms based on th 
 e solar wind parameters, namely Storm Time Ionospheric Model (STIM). Validation 
tests of TSAR’s performance show that the method provides very successful 
results for forecasts 15 min and 1 h ahead and statistically reliable results 
for predictions up to 24 hours ahead especially for quiet conditions. During 
storm intervals, the prediction error is rather small during the initial phase 
of the storm, with a general tendency to significantly increase as the storm 
evolves and recovers. On the other hand, STIM is designed to scale the 
ionospheric storm-time response, providing ionospheric predictions from 18 to 40 
hours ahead. The statistical evaluations of the methods predictions during a 
significant number of storm events gave evidence for statistically accurate 
predictions during all storm phases. In particular, STIM proved able to 
successfully capture the physical processes that govern the ionospheric storm 
onsets and their temporal evolution from the onset towards the recove 
 ry. All the above indicate that TSAR’s predictions take subs! 
 tantial 
advantage from the output of the STIM method at least during transitional 
periods such as the onset and the recovery of a storm event. The cooperation of 
the two methods’ significantly improves TSAR’s prediction accuracy and 
consistency especially for the longest prediction windows (e.g. 24 h ahead). 
Validation tests are carried out to verify the efficiency of the integrated 
algorithm in providing reliable ionospheric forecasts for the European region.