Abstract:
In this paper, we present the preliminary results of a new global three‐dimensional (3‐D)
ionospheric model developed using artificial neural networks (ANNs) by assimilating long‐term
ionospheric observations from nearly two decades of ground‐based Digisonde, satellite‐based topside
sounders, and global positioning system‐radio occultation measurements. The present 3‐D model is named
ANN‐based global 3‐D ionospheric model (ANNIM‐3D), which is the extension of previous work on the
ANN‐based two‐dimensional ionospheric model by Sai Gowtam and Tulasi Ram (2017a, https://doi.org/
10.1002/2017JA024795) and Tulasi Ram et al. (2018, https://doi.org/10.1029/2018JA025559). The vertical
electron density profiles derived from ANNIM‐3D model are found to be consistent with the ground‐based
incoherent scatter radar observations at Jicamarca and Millstone Hill. The model results have been
thoroughly validated and found in good agreement with the ground‐based Digisonde and satellite in situ
observations at different altitudes. This model successfully reproduces the large‐scale ionospheric
phenomena like diurnal and seasonal variations of equatorial ionization anomaly and its hemispheric
asymmetries, ionospheric annual anomaly, and the main ionospheric trough. Also, the present model has
predicted the ionospheric response that is consistent with the neutral composition changes and meridional
wind circulations during disturbed geomagnetic activity periods. Finally, the merits and limitations of
this model and the scope for the potential improvements have been discussed.