Modeling ionograms with deep neural networks: applications to foF2 forecasting

Abstract

The ionosphere state parameters are of fundamental importance not only for radio communication but also for space weather. As most of the space phenomena, the dynamics are governed by nonlinear processes that make forecasts a challenging endeavor. We now have available enormous datasets and ubiquitous experimental sources that can help us finding the intricate regularities in these phenomena. In this work, we will focus on the forecasting of some parameters of the steady-state low latitude ionosphere. We used ionograms from Jicamarca Radio Observatory digisonde to train two neural networks. We produced forecasts of ionospheric parameters such as virtual heights and foF2 taking into consideration ionogram characteristics. These estimations were compared to the corresponding values obtained from the digisonde, the persistence model, and foF2 values obtained from the International reference ionosphere.