Quantitative Prediction of High‐Energy Electron Integral Flux at Geostationary Orbit Based on Deep Learning

Abstract

The deep learning method of long short-term memory (LSTM) is applied to develop a model to predict the daily >2-MeV electron integral flux 1 day ahead at geostationary orbit. The inputs to the model include geomagnetic and solar wind parameters such as Kp, Ap, Dst, solar wind speed, magnetopause subsolar distance, and the value of >2-MeV electron integral flux itself over the previous five consecutive days. The model is trained on the data from the periods 1999-2007 and 2011-2016, and the efficiency of the model is tested on the 2008-2010 period. We experiment with different input combinations and find that when the model takes daily >2-MeV electron integral flux, daily averaged magnetopause subsolar distance, and daily summed Kp index as inputs, the prediction efficiencies for 2008, 2009, and 2010 are 0.833, 0.896, and 0.911, respectively. This value reaches 0.900 for 2008, when hourly >2-MeV electron integral flux, hourly magnetopause subsolar distance, and daily summed Kp index are taken as inputs, with training on the remaining data from 19 June 2003 to 13 April 2010. The prediction efficiencies of the persistence model and the 27-order autoregressive model for the same tested time period are 0.679 and 0.743, respectively. Therefore, the model developed based on the LSTM method can improve the prediction efficiency significantly for daily >2-MeV electron integral flux 1 day ahead at geostationary orbit.

Impact of Significant Time‐Integrated Geomagnetic Activity on 2‐MeV Electron Flux

“…It implies that this dynamics is at least partially predictable. Considering Int ( ap ) events and the related IntAE measure should therefore provide new opportunities to improve existing forecasting capabilities that rely either on complex physics‐based numerical models or on machine learning techniques (e.g., see Camporeale et al, ; Horne et al, , ; Li et al, ; Ma et al, ; Su et al, ; Wei et al, ). In particular, machine learning models that include lagged inputs will be able to integrate such geomagnetic indices automatically within their network (e.g., see Camporeale et al, ).…”

Data‐Driven Forecasting of Low‐Latitude Ionospheric Total Electron Content Using the Random Forest and LSTM Machine Learning Methods

“…For example, Tan et al (2018) applied the LSTM method to forecast the Geomagnetic Kp index using historical solar wind, interplanetary magnetic field and the Kp index itself as input. Wei et al (2018) were able to achieve one day lead time forecasting of high-energy electron integral flux at geostationary orbit using the LSTM deep neural network machine learning method and the Kp, Ap, Dst, solar wind speed, magnetopause subsolar distance and the 2-MeV electron integral flux itself as input.…”

“…Wei et al. (2018) were able to achieve one day lead time forecasting of high‐energy electron integral flux at geostationary orbit using the LSTM deep neural network machine learning method and the Kp, Ap, Dst, solar wind speed, magnetopause subsolar distance and the 2‐MeV electron integral flux itself as input. LSTMs are particularly popular in speech recognition, solar power forecasting, traffic prediction and others (Gensler et al., 2016; Graves et al., 2013; Zhao et al., 2017).…”

Modeling the Relationship of ≥2 MeV Electron Fluxes at Different Longitudes in Geostationary Orbit by the Machine Learning Method

“…Fukata et al [33], Xue and Ye [34], Ling et al [35], Guo et al [36], and Shin et al [37] developed models by neural networks to predict ≥2 MeV electron fluences within the next 1-3 days at one location in GEO orbit. Wang et al [38] used support vector machine and Wei et al [39] used a deep learning method to predict ≥2 MeV electron fluences in the next day. These models require external parameters as inputs, such as ≥2 MeV electron daily fluences one or a few days ahead, solar wind parameters (solar wind speed and dynamic pressure, interplanetary magnetic field, etc.…”