A New Model for Ionospheric Total Electron Content: the Impact of Solar Flux Proxies and Indices

As part of the development of a new high-resolution empirical model for ionospheric total electron content (TEC), we examine several proxies and indices that capture variability in solar flux. The new model is based on two decades (2000 - 2019) of high-resolution GNSS TEC data from the CEDAR Madrigal database and has an unprecedented 30 min temporal resolution. The model is developed at 45 degrees latitude, 0 degrees longitude and includes input parameters of solar flux, geomagnetic index, and periodic terms capturing annual, semiannual, four-monthly, and three-monthly variations. The model is tested for sensitivity to different solar flux proxies and indices. These include representations of solar flux by (1) TIMED SEE extreme ultraviolet (EUV), (2) SOHO SEM EUV, (3) the F10.7 index, (4) the Mg II Core-to-Wing ratio, (5) the Lyman Alpha Index, (6, 7) two versions of the S10.7 index, (8) a corrected version of the F10.7 index, (9) the P10.7 index, (10) the E10.7 index from the Solar Irradiance Platform, and (11) the EUV produced by version two of the Flare Irradiance Spectral Model (FISM2). Ultimately, the model constructed using FISM2 EUV performed best when compared to the training dataset, showing a root mean squared error of 1.9539 TECu. We compare the resulting model to TEC data from the first three months of 2020 and to the IRI-2016 model. The model notably captures variations in TEC on the scale of weeks which are not shown in the IRI model and closely predicts TEC variation in the early months of 2020.