| Enhancing Ionospheric Forecasts over Egypt Using a Deep Learning Approach Based on GPS-VTEC Data |
| Paper ID : 1034-ISCH |
| Authors |
|
Hassan Mahdy Nooreldeen *1, Ayman Mahrous2, Ayman M Ahmed1, Mohamed Yossuf3 1Egyptian Space Agency (EgSA) 2Institute of Basic and Applied Sciences, Egypt-Japan University of Science and Technology, Alexandria, Egypt, 3Physics Dept, Faculty of Science, Helwan University, Cairo, Egypt |
| Abstract |
| Ensuring reliable communication and navigation systems requires accurate ionospheric modeling and forecasting, particularly in regions like Egypt, where space weather can significantly disrupt ionospheric conditions. The scarcity of observational data in this area has limited prediction accuracy, with models such as IRI2020 often relying on interpolation methods that may not provide the needed precision. To overcome these challenges, we developed an advanced deep learning model that integrates Recurrent Neural Networks (RNN), particularly Long Short-Term Memory (LSTM) networks. This study utilized three years of GPS-derived Vertical Total Electron Content (VTEC) data from GNSS stations distributed across Egypt, segmented into three regions based on latitude. By incorporating geomagnetic indices (KP, Dst, and F10.7) alongside GNSS-derived VTEC, the LSTM model was trained on a time series dataset, enabling precise spatial and temporal forecasts of VTEC. Our model achieved a forecasting accuracy of 95%, significantly outperforming the traditional IRI2020 model, which showed higher RMSE values. These findings highlight the effectiveness of machine learning techniques in enhancing ionospheric predictions in Egypt, offering valuable insights into ionospheric dynamics and behavior. |
| Keywords |
| Space Weather, Ionosphere, VTEC |
| Status: Abstract Accepted (Oral Presentation) |