SME geomagnetic index data forecast based on wavelet transform and LSTM neural networks

• DOI

Solar-Earth relations are the sequences of transformations and energy impact transfers. The inflow of large amounts of energy into the geospheric shells leads to the development of magnetic storms. The influence of the southward component of the interplanetary magnetic field Bz causes increased geomagnetic activity. This paper proposes to use the SME index obtained in the SuperMAG project to analyze geomagnetic activity. In this paper, we propose an approach to forecast the SME geomagnetic activity index based on interplanetary magnetic field data and joint application of wavelet transform and LSTM neural network architecture. The effectiveness of the approach for different sets of input data can be evaluated using the generated neural network models. The constructed neural network model makes it possible to forecast the SME for the coming hours. The analysis of the results during calm and disturbed periods of geomagnetic activity showed the dependence of data approximation quality on the input data set for the neural network model and the forecast depth.

Polozov, Y. SME geomagnetic index data forecast based on wavelet transform and LSTM Neural Networks. In: Solar-Terrestrial Relations and Physics of Earthquake Precursors. STRPEP 2023. Springer Proceedings in Earth and Environmental Sciences. Springer, Cham. 2023, pp. 186-196. Doi: https://doi.org/10.1007/978-3-031-50248-4_20