Prediction of the F2 layer peak height of ionospheric dynamical parameters using a dual-element improved neural network

Abstract

ABSTRACT The ionosphere is an integral element of the Earth and reflects the variations of the Earth’s space weather and solar activity. Since extreme weather can cause ionospheric disturbances, changes in the ionosphere can indirectly enable early warning of extreme weather. The major intention of predicting the peak height of the ionospheric F2 layer (hmF2) in this paper is to acquire ionospheric variations over a period of time in a local area to facilitate future extreme weather warning research. In this paper, a dual element LSTM-CNN (long short term memory-convolutional neural network) prediction model is proposed to predict the hmF2. The performance of the proposed model is assessed by comparing it with other popular models such as SARIMA (seasonal differential autoregressive moving average), LSTM (long short term memory), BP (back propagation neural network) and IRI2016 (international reference ionospheric model) models. The outcome demonstrates that the prediction effect with the proposed model is remarkably excellent in comparison with the remaining four models. Furthermore, the proposed model has better sensitivity to rapid changes in parameters. The outcomes indicate that the forecasting model of this study has high prediction capabilities.