Quantifying the Effect of Solar Storms on Total Electron Count (TEC) in the Ionosphere over U.S. Sector Using Neural Networks

A study of large solar storms in solar cycles 23 & 24 is presented to quantify their effects on the total electron content (TEC) in the ionosphere. We study the dependence of TEC over North America on various storm parameters, including the duration of the storm, its intensity, and the rate of change of the ring current response. These parameters are inferred autonomously and compared to TEC values obtained from the CORS network of GPS stations. To quantify the effects we examine the difference between the storm-time TEC value and an average from 5 quiet days during the same month. These values are added over a grid with 1 deg x 1 deg spatial resolution in latitude and longitude over the US sector. To study correlations between the various parameters and the quantified delta TEC value, non-linear autoregressive neural networks are used. The weightage of each input variable provide data on the importance of each input on the resultant TEC change. The results of this work are compared to recent TEC studies to investigate the effects of large storms on the distribution of ionospheric density over large spatial and temporal scales.