Recent Developments in Ionosphere Forecasting: a Machine Learning Perspective

Ionosphere plays an important role in radio communication, positioning and navigation as well as in various Earth observation techniques based on electromagnetic wave propagation. Thus, modeling, monitoring and forecasting of the ionosphere has been a rather active up-to-date research topic and various models have been proposed by different researchers. The availability of globally distributed dual frequency GNSS observations and ionosphere products delivered by the IGS and other institutions provided an unprecedented time series data source for developing ionosphere forecasting models. In addition, the availability of software tools for massively parallel numerical algorithms programmable into Graphical Processing Unit hardware have delivered a boosted computation power available to researchers.  In parallel, the application of machine learning and especially deep learning methods not only into the Ionosphere research but also to various research on Earth sciences have increased. In this work, an overview of recent developments in ionosphere forecasting research is presented with a spot on those which use especially machine learning and deep learning techniques. The opportunities and challenges are listed with a classification of different approaches in the literature. An outlook is provided for further research directions in the use of learning techniques for long and short term forecasting of Ionosphere. And finally, a potential interoperability in dissemination and the use of recently developed forecasting models are discussed.