Radio Occultation (RO) is a well-established remote sensing technique that uses Global Navigation Satellite System (GNSS) signals to sound the Earth’s atmosphere. GNSS-RO measurements provide high-resolution, vertical profiles of physical parameters from the lower atmosphere (troposphere and stratosphere), e.g., refractivity, dry temperature, pressure, and water vapour, with primary application in weather forecasting and climatology models. The upper atmosphere (ionosphere) is also sounded during measurements, in which information about total electron content, electron density profiles, and scintillation indices compose the RO ionospheric data product.

The ionosphere is a dispersive medium composed of ionized particles. It is extensively conditioned by Solar activity and shows seasonal, geographical, and day- and night-time variation. Despite the benefit of the upper atmospheric data, the ionosphere influences the retrievals in the lower atmosphere by (i) adding an inherent systematic bias in bending angles, i.e., residual ionospheric error (RIE), and (ii) disturbing the signal amplitude and phase, i.e., scintillation, in the presence of irregularities regions on the electron density along the ray path, e.g., equatorial plasma bubbles. In this dissertation, both aspects are investigated by modelling the equatorial ionosphere, and its small-scale irregularities in simulations of occultation events to (i) reproduce the effects observed in measurements and (ii) assess methods that can extract information about the ionosphere and support its monitoring and modelling.

The multiple phase screen method was applied to model the GNSS signal propagation through quiet and disturbed ionospheric conditions. The small-scale irregularities in the F-region were modelled by a single slope power law to yield moderate to strong scintillation in the signals. Results were assessed according to the amplitude and phase scintillation indices, RIE, the standard deviation of the retrieved bending angles, and power spectral density (PSD). A subset of these parameters was taken as features to train a classifier based on the support vector machine algorithm. The purpose of this model was to detect RO measurements affected by ionospheric scintillation. Specifically, those in which PSD could provide further information about the irregularities according to the scintillation theory. Additionally, the back propagation (BP) method and its capability to estimate the mean distance between the receiver and irregularities were evaluated.

Applying spectral analysis techniques to RO measurements may contribute to the characterization of small-scale irregularities in equatorial plasma bubbles. The results from simulations applying the single-slope power law to model the irregularities showed a good agreement with the selected cases. The automatic detection of occultations affected by ionospheric irregularities has achieved similar performance to models trained with ground-based measurements. Furthermore, the BP method can add the estimation of the mean location to the spectral analysis information. Such tools can enlarge the amount of ionospheric data retrieved -- especially for occultations with extended vertical range and when combined with other sounding techniques.