On the validation of the Bayesian inference-based empirical model for GNSS scintillation indices at high latitudes

The Bayesian inference-based empirical model for GNSS scintillation indices (S4 and ) at high latitudes is constructed using data from Canadian Hight Arctic Ionospheric Network (CHAIN). Solar wind parameters, such as wind speed (VSW), ram pressure (SW), east-west (By) and north-south (Bz) components of the interplanetary magnetic field (IMF), the solar radio flux index (F10.7), the Sun's declination and the SuperMAG electrojet index (SME) are used as driving parameters for the model. Data from CHAIN stations located in different high-latitude regions, including the auroral zone, the cusp, and the polar cap, were included into the training data set, while data from other CHAIN stations were used to validate the performance of the model. A statistical analysis of different time periods, including geomagnetically quiet and disturbed periods, shows that the model captures most of the variations in scintillation indices. The errors of the model are presented in terms of the bias (mean value) and variability (standard deviation). Ways to the model improvement are suggested and discussed in detail.