Abstract:Pc5 geomagnetic pulsations can accelerate electrons in the radiation belts, which can pose adverse threats to both astronauts and satellites in space. The estimation of Pc5 waves in space is crucial to radiation belt dynamics studies and will help mitigate these challenges. Here, we explore the advantages of the Feed-forward Neural Network (FFNN) and Random Forest (RF) algorithm for effective estimation of Pc5 geomagnetic pulsations observed in space at geostationary orbit during solar cycle 23. The dataset used in this study is the vector magnetic field measurements retrieved from the Geostationary Operational Environmental Satellite-10 (GOES-10) and the solar wind parameters: Bz and Vx component of the solar wind in the Geocentric Solar Ecliptic (GSE) coordinate system, proton density, flow pressure, and plasma beta obtained from the OMNI Web database during part of solar cycle 23. Pc5 geomagnetic pulsations were extracted from the toroidal component of the magnetic field time series using a bandpass Butterworth filter. The continuous wavelet transform (CWT) was utilized to study the characteristics of the extracted wave in the time-frequency domain for its validation. The validated Pc5 events were used as the target in the model's development, with the solar wind parameters as the inputs. In addition to the solar wind parameters, we included an attribute of the magnetic field time series as an input variable in the model. The dataset is carefully divided to ensure effective training and testing of the models. Finally, we trained both models using the same inputs and targets and explored their estimation abilities. The model was tested during the maximum, descending, and minimum phases of solar cycle 23. Both the FFNN and RF models have a similar estimation, with average cross-correlation score (R) values of 0.74 and 0.73 and corresponding average root mean squared error (RMSEs) of 0.16 nT and 0.67 nT, respectively. The model was deployed to investigate the response of Pc5 waves during three storm days in each testing year. The machine learning (ML) model outputs showed good coherence with the observed Pc5 waves. To validate the models, we studied the correlation between the estimated Pc5 events with the Kp index, and a good correlation was seen to exist between both events. This validates the good performance of the developed models. This work will aid in the study of radiation belt dynamics and the construction of electron depletion regions in the radiation belt.

72‐hour Time Series Forecasting of Hourly Relativistic Electron Fluxes at Geostationary Orbit by Deep Learning

Predicting Geostationary (GOES) 4.1–30 keV Electron Flux Over All MLT Using LEEMYR Regression Models

An Hourly Prediction Model of Relativistic Electrons based on Empirical Mode Decomposition

Forecasting Solar Energetic Proton Integral Fluxes with Bi-Directional Long Short-Term Memory Neural Networks

Dynamic Prediction Model of Relativistic Electron Differential Fluxes at the Geosynchronous Orbit

Forecasting Ionospheric foF2 Based on Deep Learning Method

Research Progress in High-Energy Electron Flux Prediction Methods in Geosynchronous Orbit

Global Ionospheric TEC Forecasting for Geomagnetic Storm Time Using a Deep Learning‐Based Multi‐Model Ensemble Method

A forecast model of geomagnetic indices from the solar wind fluids observations based on long short-term memory neural network

Short-term Forecast of High-Energy Electron Flux Based on GPR

Deep Learning Based Solar Flare Forecasting Model. I. Results for Line-of-sight Magnetograms

Predicting Solar Energetic Particles Using SDO/HMI Vector Magnetic Data Products and a Bidirectional LSTM Network

The Short‐time Prediction of the Energetic Electron Flux in the Planetary Radiation Belt Based on Stacking Ensemble‐Learning Algorithm

Relativistic Electron Flux Prediction at Geosynchronous Orbit Based on the Neural Network and the Quantile Regression Method

Machine Learning-Based Short-Term GPS TEC Forecasting During High Solar Activity and Magnetic Storm Periods

Deep Learning the Forecast of Galactic Cosmic-Rays Spectra

Forecasting Solar Energetic Particle Events During Solar Cycles 23 and 24 Using Interpretable Machine Learning

Enhanced Forecasting of Global Ionospheric Vertical Total Electron Content Maps Using Deep Learning Methods

Ionospheric TEC Prediction Based on Ensemble Learning Models

Machine learning techniques for estimation of Pc5 geomagnetic pulsations observed at geostationary orbits during solar cycle 23

Influences of Solar Wind Parameters on Energetic Electron Fluxes at Geosynchronous Orbit Revealed by the Deep SHAP Method