Neural networks in ionospheric prediction and short-term forecasting

This paper discusses appropriateness of the neural network approach for ionospheric prediction and short-term forecasting. The idea of using the neural network technique to predict the future behaviour of a time series of the vertical-incidence ionospheric data has been presented and endorsed in the framework of the COST 251 project. It is based on the well-known fact that in recent years attention has turned to the application of neural network techniques to analysing and predicting the behaviour of non-linear time series. In particular, the behaviour of time series of the solar-terrestrial data such as the solar sunspot number, the solar 10.7 cm flux, and the geomagnetic activity indices.