Modeling ionograms with deep neural networks: Applications to forecasting

The state parameters of the ionosphere are of fundamental importance not only for space weather studies but also for technological applications such as satellite radio communications. As with many geophysical phenomena, the dynamics of the ionosphere are governed by nonlinear processes that make ionospheric forecasting a challenging endeavor. However, now, we have available enormous datasets and ubiquitous experimental sources that can help us find the intricate regularities in these phenomena. We aim to forecast ionograms for different solar activity times and database sizes with deep neural networks. Also, we will make hyperparameter tuning for each training set. Due to the neural network's extrapolation of virtual heights for all frequencies given to the model, we will estimate foF2 using recurrent neural networks to identify the last frequency of each ionogram. The predictions will be compared to measurements collected with the Digisonde system at the Jicamarca Radio Observatory, to a persistence model, and International Reference Ionosphere (IRI) model estimations. Finally, we will present preliminary results on an extension of the neural networks to produce electron density forecasts.