An aided Abel inversion technique assisted by artificial neural network-based background ionospheric model for near real-time correction of FORMOSAT-7/COSMIC-2 data

Abstract

The assumption of spherical uniformity while the retrieval of electron density profiles from the Global Navigation Satellite Systems-Radio Occultation (GNSS-RO) observations is often violated and introduces significant errors in the retrieved electron density profile data. This paper presents an improved Abel-inversion technique by incorporating the horizontal gradients in the ionosphere, which are routinely derived from the Artificial Neural Network (ANN) based background NmF2 (peak electron density of F2-layer) model (ANNC2) assimilated with near real-time Constellation Observing System for Meteorology, Ionosphere, and Climate-2 (FORMOSAT-7/COSMIC-2) NmF2 data. The ANNC2-aided Abel inversion is then implemented for more accurate retrieval of electron density profiles from COSMIC-2 in real-time. It is found that the ANNC2-aided inversion has improved the electron density values around the F2-region and below, which yields a clear separation between two anomaly crests. Further, the ANNC2-aided Abel inversion had significantly reduced the artificial plasma caves beneath the equatorial ionization anomaly crests. Furthermore, COSMIC-2 NmF2 observations obtained from both classical and the ANNC2-aided Abel inversion are compared with the ground-based Digisonde data and found that the ANNC2-aided inversion gives the better results. This study provides some new insights on the aided Abel inversion technique assisted by ANN models for the real-time correction of Abel retrieved electron density profiles.