Estimation of Geomagnetic and Solar Indices by Global Ionospheric Maps With Use of Neural Networks

The paper presents a method for estimating geomagnetic and solar activity indices using global ionospheric maps of the total electron content. A convolutional artificial neural network is used as a regression model. The obtained neural network estimates show a high degree of correlation with real values of the indices. Thus, the correlation coefficient for estimating the F 10.7 index with its real value is 0.972 (cf. the correlation coefficient of the F 10.7 index with the global electron content index of the ionosphere is 0.882). The constructed neural network model allows us to analyze which areas of the global ionospheric maps of the total electron content show the most significant information about the level of solar and geomagnetic activity.