Using Artificial Neural Network Training for Space Weather Products

Artificial Neural Network (ANN) training or machine learning has been used to derive two of space weather products: solar EUV (26-34 nm) flux and ionospheric TEC. The solar EUV flux has been obtained through training with solar radio fluxes at 410, 610, 1415, 2695, 4995, and 8800 MHz. The ANN training shows that the 610 and 1415 MHz are the dominant contributors while the 2695 and 4995 MHz have a minor contribution to estimate EUV flux. This is consistent with the report that significantly better neutral density modeling can be obtained using F30 cm (1000 MHz) versus F10.7cm (2800 MHz) [Bruinsma, 2015]. On the other hand, TIMED/GUVI 135.6 nm radiances have been ANN trained with ground GPS TEC data. The GUVI data extend the ionospheric TEC coverage over the world, especially over the oceans where there are few ground GPS stations.