Forecasting global ionospheric total electron content (TEC) using deep learning

Abstract:

Ionospheric total electron content (TEC) from global ionospheric maps (GIM) is widely applied to scientific research about space weather impacts, so there is great interest in the community in short-term ionosphere forecasting. In this study, the long short-term memory (LSTM) neural network (NN) is applied to forecast the 256 spherical harmonic (SH) coefficients used to construct GIM based on multiple input data, including historical time series of the SH coefficients, solar extreme ultraviolet (EUV) flux, disturbance storm time (Dst) index, and hour of the day. Comparing to those models with no external solar radiation or geomagnetic activity drivers concatenated into the LSTM layer, our developed LSTM model is able to improve the forecast of the SH coefficients by including the solar EUV flux and Dst index. After using the developed LSTM model, the global hourly TEC maps are reproduced from 256 predicted SH coefficients by using the SH function, and a comprehensive evaluation is carried out with respect to the CODE GIM TEC. Results show that the proposed approach performs well during both quiet and storm times. Moreover, typical ionospheric structures, such as equatorial ionization anomaly (EIA) and storm-enhanced density (SED), are well reproduced from the predicted TEC maps during storm time. It is also important to note that the proposed approach shows competitive performance in predicting global TEC when compared to traditional IRI-2016 and NeQuick-2 models.