Prediction of Geomagnetically Induced Currents in Low-latitude Regions using Deep Learning

The present study proposes a geomagnetically induced currents (GICs) prediction model for three low-latitude locations, Huancayo, Peru, Addis Ababa, Ethiopia and Guam, United States. It employs the long short-term memory (LSTM) and bidirectional long short-term memory (BiLSTM) neural networks. The model’s performance was accessed using the interleaving odd-even data split (IntOE) approach as a benchmark. The geomagnetic field variation (dB/dt) derived from geomagnetic disturbance event on 31 March 2001, is applied as the GICs’ proxy. Results showed that employing both models, LSTM and BiLSTM with block division data split markedly enhanced prediction accuracy by up to 66% compared to IntOE. However, IntOE proves to be more effective for event-based validation.