Forecasting single-station ionospheric TEC over China using a combined DBO-LSTM model during geomagnetic storms

The ionospheric total electron content (TEC) in the ionosphere is a crucial parameter for studying ionospheric variations and space weather. Short-term prediction of the ionosphere is of significant importance for near-Earth space environment monitoring. This study proposes a hybrid model combining Dung Beetle Optimization (DBO) algorithm and Long Short-Term Memory (LSTM) neural network. By optimizing the number of neurons in the LSTM network, the dropout rate, the number of neurons in the fully connected layer, and the batch size, the prediction accuracy of the model can be improved to a certain extent. This paper utilizes TEC data from 24 GNSS observation stations of Crustal Movement Observation Network of China (CMONOC) and five parameters selected through Random Forest training, including f10.7, Lyman_alpha, SW Plasma Speed, Dst, and R Sun Spot indices to train the model. By forecasting ionospheric TEC during geomagnetic storms and comparing the model’s prediction results with those of LSTM and RNN models, it is found that the DBO-LSTM model outperforms the LSTM and RNN models on the test set. During severe geomagnetic storms, the RMSEs of the DBO-LSTM, LSTM, and RNN models at the low-latitude station GDZH are 5.34 TECU, 5.91 TECU, and 6.38 TECU, respectively. The proposed model demonstrates good predictive performance during geomagnetic storms.