Machine learning approach for prediction of total electron content and classification of ionospheric scintillations over Visakhapatnam region

Abstract

Ionospheric scintillations, which are due to ionospheric plasma density anomalies, negatively impact trans-ionospheric signals and the positioning accuracy of the global navigation satellite system (GNSS). One of the crucial variables for comprehending space weather conditions is the total electron content (TEC) of the ionosphere. It is vital to predict the ionospheric TEC before making efforts to enhance the GNSS system. In this article, the long short-term memory machine learning approach for TEC prediction is presented, based on which the ionospheric phase scintillations are identified and classified using popular classifiers: support vector machines and decision trees. In this article, the comparative analysis of these classifiers is presented using the standard performance metrics: accuracy, recall, precision, and F1 score.