Improved prediction of I/sub F2/ and I/sub G/ indices using neural networks

摘要

The paper presents an investigation into the use of artificial neural networks to predict the values of the I/sub F2/ and I/sub G/ ionospheric indices. Since 1982, the World Data Centre C1 for Solar-Terrestrial Physics has produced predictions for these indices using an adaptation of the McNish-Lincoln (1949) technique for predicting sunspot numbers. It is demonstrated that significantly more accurate predictions are obtained using artificial neural networks, which form the basis of predictions which will in future be issued by the World Data Centre.