Prediction of vertical electron density of the ionosphere for calculating ionospheric corrections in radio navigation using sklearn machine learning libraries

The article discusses a method for predicting the vertical electron density of the ionosphere for calculating ionospheric corrections in radio navigation using machine learning libraries in Python. Various machine learning algorithms, such as Decision Tree Regressor and Random Forest, for modeling the dynamics of ionospheric conditions are analyzed. These methods are compared with neural networks. Measurements of ionospheric activity and geophysical parameters were used as input data. The resulting models demonstrated sufficient accuracy characteristics. The possibility of forecasting in real-world conditions was achieved, which allows for increasing the accuracy of satellite navigation systems and reducing the impact of ionospheric distortions on satellite signal measurements. Methods for decomposing images of predicted electron density distributions for transmitting the predicted distributions to remote navigation computing devices using the spherical harmonic decomposition method and spline interpolation are analyzed. The results confirm the effectiveness of machine learning methods for solving problems of spatial and temporal estimation of ionospheric parameters in radio navigation systems.